JUDGMENT OF THE GENERAL COURT (Ninth Chamber)
12 December 2018 (*)
(Competition — Agreements, decisions and concerted practices — Market for perindopril, a medicinal product intended for the treatment of cardiovascular diseases, in its originator and generic versions — Decision finding an infringement of Article 101 TFEU — Patent dispute settlement agreement — Licensing agreement — Technology acquisition agreement — Restriction of competition by object — Restriction of competition by effect — Balance between competition law and patent law)
In Case T‑684/14,
Krka Tovarna Zdravil d.d., established in Novo Mesto (Slovenia), represented by T. Ilešič and M. Kocmut, lawyers,
applicant,
v
European Commission, represented by F. Castilla Contreras, B. Mongin and C. Vollrath, acting as Agents, assisted by D. Bailey, Barrister,
defendant,
APPLICATION under Article 263 TFEU for partial annulment of Commission Decision C(2014) 4955 final of 9 July 2014 relating to a proceeding under Article 101 and Article 102 TFEU [Case AT.39612 — Perindopril (Servier)] in so far as it concerns the applicant,
THE GENERAL COURT (Ninth Chamber),
composed of S. Gervasoni (Rapporteur), President, L. Madise and R. da Silva Passos, Judges,
Registrar: C. Heeren, Administrator,
having regard to the written part of the procedure and further to the hearing on 20 June 2017,
gives the following
Judgment
I. Background to the dispute
A. Perindopril
1 The Servier group, composed of Servier SAS and several subsidiaries (individually or jointly, ‘Servier’), developed perindopril, a medicinal product used in cardiovascular medicine, primarily intended for the treatment of hypertension and heart failure, by inhibiting the angiotensin converting enzyme.
2 The active pharmaceutical ingredient (‘API’) of perindopril, that is to say, the biologically active chemical substance which produces the desired therapeutic effects, takes the form of a salt. The salt used initially was erbumine (or tert-butylamine), which is in its crystalline form on account of the synthesis process applied by Servier.
1. The compound patent
3 The perindopril compound patent (patent EP0049658) was filed with the European Patent Office (EPO) on 29 September 1981. That patent was due to expire on 29 September 2001, but protection was prolonged in a number of EU Member States, including the United Kingdom, until 22 June 2003, in accordance with Council Regulation (EEC) No 1768/92 of 18 June 1992 concerning the creation of a supplementary protection certificate for medicinal products (OJ 1992 L 182, p. 1).
2. Secondary patents
4 In 1988, Servier also filed a number of patents with the EPO relating to processes for the manufacture of the perindopril compound with an expiry date of 16 September 2008: patents EP0308339, EP0308340 (‘the 340 patent’), EP0308341 (‘the 341 patent’) and EP0309324.
5 Servier filed new patents relating to erbumine and its manufacturing processes with the EPO between 2001 and 2005, including patent EP1294689 (known as ‘the beta patent’), patent EP1296948 (known as ‘the gamma patent’), and patent EP1296947 (known as ‘the alpha patent’ — ‘the 947 patent’).
6 The 947 patent application relating to the alpha crystalline form of erbumine and the process for its preparation was filed on 6 July 2001 and granted by the EPO on 4 February 2004.
7 Servier also filed national patent applications in several EU Member States before they were parties to the Convention on the Grant of European Patents, which was signed in Munich on 5 October 1973 and entered into force on 7 October 1977 (‘the EPC’). Servier filed, for example, patent applications relating to the 947 patent in Bulgaria (BG 107 532), the Czech Republic (PV 2003-357), Estonia (P 200 300 001), Hungary (HU 225340), Poland (P 348492) and Slovakia (PP 0149-2003). All the patent applications in question were filed on the same date: 6 July 2001. The patents were granted on 16 May 2006 in Bulgaria, on 17 August 2006 in Hungary, on 23 January 2007 in the Czech Republic, on 23 April 2007 in Slovakia and on 24 March 2010 in Poland.
B. The applicant
8 The Krka group is composed of the parent company, Krka Tovarna Zdravil d.d., and several subsidiaries in Slovenia and other countries (individually or jointly ‘Krka’ or ‘the applicant’).
C. Disputes relating to perindopril
1. Disputes before the EPO
9 Ten generic companies, including Niche Generics Ltd, Krka, Lupin Ltd and Norton Healthcare Ltd, a subsidiary of Ivax Europe which subsequently merged with Teva Pharmaceuticals Ltd filed opposition proceedings against the 947 patent before the EPO in 2004, seeking the revocation in full of that patent on grounds of lack of novelty, lack of inventive step and insufficient disclosure of the invention. On 27 July 2006, the EPO’s Opposition Division confirmed the validity of the 947 patent after Servier made some minor amendments to its original claims (‘the EPO decision of 27 July 2006’). Seven companies brought an appeal against that decision. Niche Generics withdrew from the opposition procedure on 9 February 2005, Krka on 11 January 2007 and Lupin on 5 February 2007. By decision of 6 May 2009, the EPO’s Technical Board of Appeal annulled the EPO decision of 27 July 2006 and revoked the 947 patent. Servier’s request for a revision of that decision was rejected on 19 March 2010.
2. Disputes before the national courts
10 The validity of the 947 patent has, moreover, been challenged by generic companies before the courts of certain Member States, notably in the United Kingdom and the Netherlands.
(a) Dispute between Servier and Krka
11 On 30 May 2006, Servier applied for an interim injunction in Hungary preventing the marketing of a generic version of perindopril placed on the market by Krka, as a result of the infringement of the 947 patent. That application was rejected in September 2006.
12 In the United Kingdom, on 28 July 2006, Servier brought an action for infringement of the 340 patent against Krka before the High Court of Justice (England and Wales), Chancery Division (Patents Court). On 2 August 2006, it also brought an action for infringement of the 947 patent against Krka and applied for an interim injunction. On 1 September 2006, Krka brought a counterclaim for annulment of the 947 patent and, on 8 September 2006, a separate counterclaim for annulment of the 340 patent. On 3 October 2006, the High Court of Justice (England and Wales), Chancery Division (Patents Court), granted Servier’s application for an interim injunction and denied the motion for summary judgment brought by Krka on 1 September 2006 seeking the invalidation of the 947 patent. On 1 December 2006, the infringement proceedings were discontinued as a result of the settlement reached between the parties and the interim injunction was lifted.
(b) Dispute between Servier and Apotex
13 In the United Kingdom, Servier brought an action for infringement before the High Court of Justice (England & Wales), Chancery Division (Patents Court), against the company Apotex Inc. on 1 August 2006, claiming infringement of the 947 patent, since Apotex had launched a generic version of perindopril in the United Kingdom on 28 July 2006. Apotex brought a counterclaim for annulment of that patent. An interim injunction prohibiting Apotex from importing, offering to sell or selling perindopril was obtained on 8 August 2006. On 6 July 2007, the High Court of Justice (England & Wales), Chancery Division (Patents Court), ruled that the 947 patent was invalid because it lacked novelty and inventive step over the 341 patent. Consequently, the injunction was lifted immediately and Apotex was able to resume selling its generic version of perindopril on the United Kingdom market. On 9 May 2008, the Court of Appeal (England & Wales) (Civil Division) dismissed Servier’s appeal against the judgment of the High Court of Justice (England & Wales), Chancery Division (Patents Court).
14 On 9 October 2008, the High Court of Justice (England & Wales), Chancery Division (Patents Court), awarded damages to Apotex in the amount of 17.5 million pounds sterling (GBP) on account of the loss of revenue suffered during the period when the injunction was in force. On 29 March 2011, however, the High Court of Justice (England and Wales), Chancery Division (Patents Court), ordered Apotex to repay that sum to Servier on the basis of the ex turpi causa principle, since a valid Canadian patent protected the perindopril compound until 2018 and Apotex produced and sold its product in Canada. However, the Court of Appeal (England and Wales) (Civil Division) set aside that decision by judgment of 3 May 2012. On 29 October 2014, the Supreme Court of the United Kingdom dismissed Servier’s appeal against the judgment of the Court of Appeal (England and Wales) (Civil Division).
15 In the Netherlands, on 13 November 2007, Katwijk Farma BV, an Apotex subsidiary, brought an action before the Rechtbank Den Haag (District Court, The Hague) for annulment of the 947 patent, as validated in the Netherlands. Servier applied for an interim injunction against Katwijk Farma on 7 December 2007, which was rejected by the Rechtbank Den Haag (District Court, The Hague) on 30 January 2008. Following the annulment of the 947 patent for the Netherlands on 11 June 2008 by the Rechtbank Den Haag (District Court, The Hague) in the context of the action brought by Pharmachemie BV, Servier and Katwijk Farma withdrew from the ongoing proceedings.
D. Patent dispute settlements
16 Servier entered into a series of settlement agreements with a number of generic companies with which it was involved in patent disputes.
17 On 27 October 2006, Servier entered into a settlement agreement and a licence agreement with Krka, supplemented by an amendment made on 2 November 2006.
18 The settlement agreement with Krka provides that the 947 patent also covers equivalent national patents (Annex B).
19 In accordance with the settlement agreement with Krka, in force until the expiry or the revocation of the 947 or 340 patents, Krka undertook to withdraw any existing claim against the 947 patent worldwide and against the 340 patent in the United Kingdom, and not to challenge either of those patents (‘the patents at issue’) worldwide in the future (Clause I(ii) and (iv)). Moreover, Krka and its subsidiaries were not authorised to launch or to market any generic form of perindopril which would infringe the 947 patent for the duration of the validity of that patent and in the country in which it was still valid, unless otherwise expressly authorised by Servier (Clause V). Similarly, Krka could not supply to any third party a generic version of perindopril that would infringe the 947 patent unless otherwise expressly authorised by Servier (Clause V(2)). In return, Servier was required to withdraw the proceedings pending worldwide against Krka based on the infringement of the 947 and the 340 patents, including its applications for interim injunction (Clause I(i)).
20 Pursuant to the licence agreement concluded with Krka for a period corresponding to the validity of the 947 patent (Article 5), Servier granted Krka an exclusive, irrevocable licence on the 947 patent to use, manufacture, sell, offer for sale, promote and import its own products which contain the alpha crystalline form of erbumine (Article 2) in seven Member States (‘the seven Member States’ or ‘the seven Markets’) namely the Czech Republic, Latvia, Lithuania, Hungary, Poland, Slovenia and Slovakia (Article 1). In return, Krka was required to pay Servier 3% royalties on its net sales prices throughout those territories (Article 3). Servier was entitled, in those States, to use the 947 patent directly or indirectly (that is to say for one of its subsidiaries or for one third party per country) (Article 2).
21 On 5 January 2007, Servier also entered into an assignment agreement with Krka.
22 Pursuant to the assignment agreement, Krka assigned two patent applications to Servier, one concerning a process for the preparation of perindopril (WO 2005 113500) and the other the preparation of formulations of perindopril (WO 2005 094793) (Article 1). The technology protected in those patent applications was used for the production of Krka’s perindopril.
23 Krka undertook not to challenge the validity of any patents granted on the basis of the applications at issue (Article 3).
24 In return for that assignment, Servier paid Krka EUR 15 million for each of the applications at issue (Article 2).
25 Servier also granted Krka a non-exclusive, irrevocable, non-assignable, royalty-free licence, with no right to sub-license (other than to its subsidiaries) on the applications or resulting patents, that licence being unrestricted in time, territory or scope of use (Article 4).
E. The Sector Inquiry
26 On 15 January 2008, the Commission of the European Communities decided to open an inquiry into the pharmaceutical sector pursuant to Article 17 of Council Regulation (EC) No 1/2003 of 16 December 2002 on the implementation of the rules on competition laid down in Articles [101] and [102 TFEU] (OJ 2003 L 1, p. 1) in order to identify the factors contributing to the decline in innovation in that sector, measured by the number of new medicines reaching the market, and the reasons for the delayed entry into the market of certain generic medicines.
27 The Commission published a preliminary report on the results of its inquiry on 28 November 2008, followed by a public consultation. On 8 July 2009, it adopted a communication giving a summary of its pharmaceutical sector inquiry report. The Commission stated, inter alia, in that communication, that the monitoring of patent settlements concluded between originator companies and generic companies should continue in order better to understand the use of that type of agreement and to identify those agreements that delay generic market entry to the detriment of EU consumers and may constitute an infringement of competition rules. The Commission subsequently published six annual reports on the monitoring of patent settlement agreements.
F. The administrative procedure
28 On 24 November 2008, the Commission carried out unannounced inspections, inter alia, at the applicant’s premises. The Commission sent requests for information to several companies, including the applicant, in January 2009.
29 On 2 July 2009 the Commission decided to open proceedings against Servier and several other companies, including the applicant.
30 In August 2009 and then between December 2009 and May 2012, the Commission sent further requests for information to the applicant. Between 2009 and 2012, the applicant was invited to attend a number of state of play meetings.
31 On 27 July 2012, the Commission issued a Statement of Objections to several companies including the applicant, which submitted its reply on 5 December 2012.
32 Following the hearing of the companies concerned between 15 and 18 April 2013, further state of play meetings were arranged and additional requests for information sent.
33 On 18 December 2013, the Commission granted access to evidence gathered or more widely disclosed after the Statement of Objections and sent the applicant a Letter of Facts to which it replied on 17 January 2014.
34 The Hearing Officer issued his final report on 7 July 2014.
35 On 9 July 2014, the Commission adopted Decision C(2014) 4955 final relating to a proceeding under Article 101 and Article 102 TFEU [Case AT.39612 — Perindopril (Servier)] (‘the contested decision’).
36 Pursuant to Article 4 of the contested decision, the applicant infringed Article 101 TFEU by participating in three agreements covering all Member States, except the Czech Republic, Croatia, Latvia, Lithuania, Hungary, Poland, Slovenia and Slovakia, for the period starting 27 October 2006 – except as regards Bulgaria and Romania, where the infringement started on 1 January 2007, Malta, where the infringement started on 1 March 2007, and Italy, where the infringement started on 13 February 2009 – and ending on 6 May 2009, except as regards the United Kingdom, where the infringement ended on 6 July 2007, and the Netherlands, where the infringement ended on 12 December 2007.
37 The Commission imposed a fine of EUR 10 million on Krka (Article 7(4)(a) of the contested decision). Krka, which is an addressee of the contested decision (Article 9 of that decision) is also to refrain from repeating the infringement penalised and from any act or conduct having the same or similar object or effect (Article 8 of the contested decision).
II. Procedure and forms of order sought
38 By application lodged at the Court Registry on 19 September 2014, the applicant brought the present action.
39 In the context of the measures of organisation of procedure provided for in Article 89 of the Rules of Procedure of the General Court, the Commission was invited to respond in writing to a question relating to data concerning the licence agreement which were made confidential in the contested decision. It sent its reply within the prescribed period.
40 Acting on a proposal from the Judge-Rapporteur, the Court (Ninth Chamber) decided to open the oral part of the procedure and, by way of measures of organisation of procedure pursuant to Article 89(3)(a) of the Rules of Procedure, put a written question to the parties, requesting them to answer that question at the hearing.
41 At the hearing on 20 June 2017, the parties presented oral argument and their answers to the written and oral questions put by the Court.
42 The applicant claims that the Court should:
– annul the contested decision in so far as it concerns the applicant, in particular Articles 4, 7(4)(a), 8 and 9 thereof;
– order such measures as justice may require;
– order the Commission to pay the costs.
43 The Commission contends that the Court should:
– dismiss the application;
– order the applicant to pay the costs.
III. Law
44 At the outset, the applicant presents a number of general arguments that it intends to develop in greater detail in the context of its pleas in law.
45 In that regard, it disputes, in particular, that the grant of an ordinary royalty bearing sole licence in the context of a settlement agreement under which the patent challenger obtains the possibility to enter the market prior to the expiration of the patent may be classified as ‘a reversed payment settlement agreement’ and may fall within Article 101(1) TFEU.
46 It adds that the Commission’s analysis effectively obliterates some of the fundamental legal principles on which the EU legal system is based, such as respect for intellectual property rights and the lawfulness of settlements relating to genuine disputes.
47 Lastly, it takes the view that ‘only a strict judicial reproach of the Commission’s position as provided in the [contested] decision would serve ... to confirm the general lawfulness of (patent) settlement agreements, and ... to reaffirm the importance of intellectual property laws, which confer exclusive/exclusionary rights on holders of patents’.
48 The applicant goes on to raise six pleas in law on which it bases its claims. The first plea alleges that there was no full and impartial examination of the situation at issue; the second plea alleges that there was no potential competition; the third plea alleges that there was no restriction by object as regards the settlement and licence agreements; the fourth plea alleges that there was no restriction by object as regards the assignment agreement; the fifth plea alleges that there was no restriction by effect and the sixth plea relates to the exemption laid down in Article 101(3) TFEU.
49 It is appropriate to examine first of all the third plea in law, relating to the settlement and licence agreements, then the fourth plea relating to the assignment agreement. By those two pleas, the applicant alleges that the Commission has not established the existence of a restriction of competition by object. It will then be necessary to examine the fifth plea, by which the applicant argues that the Commission has not demonstrated the existence of a restriction of competition by effect.
A. Third plea in law, alleging that there was no restriction of competition by object as regards the settlement and licence agreements
1. Arguments of the parties
50 By this plea, the applicant disputes, inter alia, the Commission’s assessment of the facts at issue, which led it to find a restriction of competition by object.
51 The applicant relies, in essence, on two separate complaints, alleging, first, that there was no restriction by object since the generic company was not induced to abandon its efforts to enter several EU markets with a generic product and secondly, that there was no restriction of competition by object since the use of a settlement was legitimate.
(a) The absence of restriction by object since the applicant was not induced to abandon its efforts to enter several EU markets with a generic product
52 The applicant, in the first place, disputes the existence, in the settlement and licence agreements, of any inducement, in the form of a value transfer, not to pursue its efforts to enter several EU markets with a generic product.
53 It states that the settlement agreement entailed absolutely no monetary consideration.
54 The applicant submits, furthermore, that the conditions under which the licence was granted cannot be regarded as constituting an inducement not to compete with Servier.
55 It points out that under the licence agreement concluded with Servier, Krka was to pay royalties in the amount of 3% of the net sales of perindopril in alpha form made throughout all of the territory in which the licence applied.
56 In its view, that licence agreement was a genuine ordinary quid pro quo commercial agreement, under which both parties assumed rights and obligations of equal commercial value, with no value being transferred either way.
57 The applicant disputes the Commission’s argument that the licence must be perceived as a market sharing arrangement and an inducement on the ground that it was granted as a sole licence limited to seven markets that have traditionally been important for the applicant.
58 The applicant also submits that the settlement and licence agreements cannot be regarded as a market sharing arrangement because the applicant and Servier were in fierce competition on the seven markets covered by the licence agreement, as was stated, furthermore, by the Commission in recitals 1725 and 2350 of the contested decision.
59 The applicant disputes that the grant of the licence may be perceived as a market sharing arrangement on the ground that it was an exclusive licence limited to seven national markets. It states, thus, that Servier’s decision to grant it a licence under the 947 patent for seven countries only and not for the whole of the EU is merely a demonstration of the exclusive nature of the rights conferred on the patent holder by intellectual property legislation. The applicant adds that ‘[t]he illogical “all or nothing rule” (i.e. to grant the licence for all the territories in which the patent holder has been granted patents or for none of them), which the Commission is evidently introducing is contrary to the principles of both patent law and proprietary legal systems (in accordance with which it is up to the patent holder/rights holder alone to decide how to exploit its patented technology) and has no support in EU case-law’. According to the applicant, that approach would result in at least ‘partial expropriation’.
60 The applicant also relies on paragraph 156 of the Guidelines on the application of Article 101 [TFEU] to technology transfer agreements (OJ 2004, C 101, p. 2, ‘the 2004 Guidelines on technology transfer agreements’), according to which ‘[t]he parties to a licence agreement are normally free to determine the royalty payable by the licensee and its mode of payment without being caught by Article [101(1)]’.
61 It also relies on the fact that the Commission stated, in footnotes 2349 and 2354 of the contested decision, that the royalty rate applied was fair.
62 The applicant criticises recital 1738 of the contested decision in which the Commission relied on the existence of an ‘opportunity cost of not concluding the Krka Settlement Agreement’ which Krka had allegedly estimated would amount to over EUR 10 million of lost profits over a period of three years.
63 It criticises the Commission’s interpretation of a statement it made according to which the settlement agreement would be equivalent, in three years, to ‘well above [EUR] 10 [million]’ of lost profits.
64 The applicant adds that the commercial value of any licence, as assessed ex ante, is, essentially, unpredictable.
65 It also states that, even if the Commission were able to assign a significant positive value to the licence which had been granted to the applicant, that would not enable the licence agreement to be classified as a ‘significant inducement’. Again according to the applicant, if any potential benefit to a party arising from a settlement was automatically viewed as a significant inducement to settle, this would by definition bring an end to all settlements.
66 The applicant also relies on paragraphs 204 and 206 of the 2004 Guidelines on technology transfer agreements, according to which:
‘204. Licensing may serve as a means of settling disputes or avoiding that one party exercises his intellectual property rights to prevent the other party from exploiting his own technology. Licensing including cross licensing in the context of settlement agreements and non-assertion agreements is not as such restrictive of competition since it allows the parties to exploit their technologies post agreement. However, the individual terms and conditions of such agreements may be caught by Article 81(1). […]
206. In cases where it is likely that in the absence of the licence the licensee could be excluded from the market, the agreement is generally pro-competitive ...’
67 The applicant submits that the statement made by Lupin and used by the Commission in recital 1730 of the contested decision was never communicated to it and that, moreover, such a statement made by a third party to the agreement has no probative value.
68 The applicant also submits that the Commission misinterpreted several of its statements. According to one of its statements, Servier and Krka reached an agreement on the ‘main points’ of the settlement, namely, the geographic scope of the licence, some of its commercial terms, and Krka’s undertaking to withdraw the opposition proceedings concerning the 947 patent and to refrain from entering the market for the duration of the validity of that patent. On the basis of those statements, the Commission concluded that there was an interdependence between Servier’s licence and Krka’s undertaking to withdraw from competition on certain markets. The applicant counters that it had agreed only to bring an end to proceedings initiated in the UK, to withdraw its opposition against the 947 patent, not to challenge that patent and not to market a generic version of perindopril infringing that patent for the duration of its validity but that this was not a ‘withdrawal from competition’.
69 It adds that, on the contrary, the licence agreement had the effect of increasing competition in several national markets. In that respect, it submits that, according to Commission Regulation (EC) No 772/2004 of 27 April 2004 on the application of Article [101(3) TFEU] to categories of technology transfer agreements (OJ 2004 L 123, p. 11), technology transfer agreements promote competition.
70 In the second place, the applicant submits that the settlement and licence agreements were, in the context in which they were adopted, a mutually acceptable compromise. They were not the result, therefore, of one of the parties inducing the other, but of the context in which they were concluded.
71 The applicant points out in that regard that Servier held several valid patents, including the 340 patent and the 947 patent, and that the EPO decision of 27 July 2006 had confirmed the validity of the 947 patent before the agreement was signed. According to the applicant, that led it, furthermore, to cease all activities on its generic perindopril in alpha form.
72 The applicant also submits that, on 3 October 2006, the High Court of Justice (England and Wales), Chancery Division (Patents Court), granted Servier’s application for an interim injunction and denied the motion for summary judgment brought by Krka on 1 September 2006 seeking the invalidation of the 947 patent.
73 It notes, lastly, that, on the date on which the settlement and licence agreements were adopted, it had at its disposal only perindopril in alpha form, that is to say a product which could be marketed only by infringing Servier’s patents. Moreover, it had large amounts of the API and the end product in stock.
74 The applicant concludes that, in that context, it was essential for it to end its disputes with Servier.
75 In the third place, the applicant submits that, independently of the settlement and licence agreement, it pursued its efforts to enter the perindopril market. Those efforts were therefore not diminished as a consequence of those agreements.
76 The applicant states that it started work on research into new forms of non-alpha perindopril. It adds that, twelve months after the adoption of the settlement and licence agreements, it successfully developed a non-alpha form of perindopril, conducted the necessary statutory studies, created the dossier and applied for the marketing authorisations.
77 It states that the choice not to challenge Servier’s patents, preferring instead to develop a non-alpha form of perindopril, was just as fraught with uncertainty as a choice to enter into disputes with Servier. Therefore, any difference as regards the effects on competition of one or other choice does not appear to be as obvious as the Commission held in the contested decision.
78 The applicant adds, in that regard, that although the EPO decision which finally invalidated the 947 patent was adopted in May 2009, it succeeded in inventing a non-infringing form of perindopril and was granted marketing authorisations only a few months later, in the autumn of the same year.
79 To conclude on this complaint as a whole, the applicant maintains that the evidence it submitted confirms that there was no transfer of value from Servier to the applicant under the settlement agreement. The evidence therefore made it possible to establish, in its view, that no ‘reverse payment’ or significant inducement was agreed to the benefit of the applicant under the settlement agreement. Furthermore, again according to the applicant, proper analysis of that evidence confirms that the applicant’s decision to enter into the settlement agreement was based solely upon its assessment of the validity of the patents at issue, whether its generic perindopril infringed Servier’s patents and the respective strength of the arguments presented by each of the parties to the litigation. Lastly, the applicant concludes that its independent efforts to enter the perindopril market were not affected by the settlement agreement.
80 The Commission submits that it is clear from the settlement and licence agreements, and from the context in which they were adopted, that Krka was encouraged not to challenge Servier’s patents and not to enter EU national markets other than those for which it had been granted a licence.
81 In the first place, the Commission submits that the level of royalties paid under the licence agreement was not decisive in identifying an inducement for Krka to accept restrictions of competition.
82 In its view, in the particular circumstances of the present case, the licence was the unlawful means used to secure Krka’s agreement to the restrictions on its entry into all of the States that were members of the European Union at the material time, with the exception of the seven States for which it had a licence agreement (‘the 18 to 20 markets’ or the ‘18 to 20 Member States’).
83 It adds, as regards the amount of EUR 10 million which is disputed by the applicant (see paragraph 62 above), that the approximation of the parties’ profit expectations under the licence is relevant as it helps to explain why the parties agreed to the licence as part of the settlement.
84 The Commission also states that it rejects the applicant’s assertions that the licence agreement must be assessed in isolation as an ordinary commercial agreement.
85 Lastly, it states that, in its view, what matters is whether the value transferred — whether financial or a commercial arrangement — is capable of inducing a competing generic company to enter into an anticompetitive agreement.
86 Furthermore, according to the Commission, one of the main arguments in the application — namely that Krka would have been unable to sell its perindopril in alpha form — is unfounded since Servier’s patents did not automatically prevent Krka from trying to sell its perindopril and at no point did a court of a Member State make a ruling to that effect.
87 In the second place, the Commission submits that the settlement was not concluded on the basis of an assessment by the parties to that agreement of the validity of the patent and of the merits of the litigation concerning it.
88 In particular, according to the Commission, there was no justification for granting an exclusive licence if Servier’s patents were valid.
89 Furthermore, according to the Commission, Krka overstates the impact of the EPO decision of 27 July 2006.
90 In the third place, the Commission submits that the fact that the applicant was able to develop a non-alpha form of perindopril did not alter the anticompetitive nature of the market sharing.
91 In particular, it points out that, even if Krka had managed to research, develop, produce and market its non-alpha perindopril, it would still have taken it three years to put it on the market.
(b) The absence of restriction by object as a result of the lawful nature of the use of a settlement
92 In the context of the present complaint, the applicant relies on four separate sub-complaints, alleging, first, that there were genuine disputes between the parties and that the conclusion of the settlement agreement was based on the parties’ evaluation of those disputes, secondly, that the non-challenge clause cannot be classified as an infringement of competition law, thirdly, that the non-marketing clause cannot be classified as an infringement of competition law, and fourthly, that the principles relating to the restriction of competition by object were applied incorrectly.
(1) The allegation that there were genuine disputes between the parties and that the conclusion of the settlement agreement was based on the parties’ evaluation of those disputes
93 The applicant submits that the settlement concerned genuine disputes between the parties and was concluded on the basis of the parties’ evaluation of the probable outcome of those disputes.
94 In that regard, the applicant points out that there were several open ongoing litigation proceedings between itself and Servier:
– the procedure initiated before the EPO by the applicant in 2004 with a notice of opposition filed against the 947 patent;
– the procedures initiated in 2006 by Servier before the High Court of Justice (England and Wales), Chancery Division (Patents Court), against the applicant, that is to say actions for infringement of the 340 patent and of the 947 patent and an application for interim measures;
– the counterclaims brought in 2006 by the applicant for the annulment of the 947 patent and the 340 patent.
95 The applicant submits that its decision to enter into an agreement with Servier was adopted on the basis, in particular, of an assessment of the validity of, inter alia, the 947 patent and of the quality of the arguments put forward by each of the parties to the litigation.
96 The applicant points out that it was unable to enter the perindopril market, not on account of the clauses of the settlement agreement, but on account of the exclusive rights granted to Servier by the public authorities in accordance with intellectual property law.
97 According to the applicant, the settlement agreement did not go beyond the scope of the patent, since it did not impose any restrictions on the applicant relating to non-alpha perindopril.
98 The applicant points out, in that regard, that it focused its research efforts on new non-alpha forms of API and formulation and therefore that it did not consent to any restriction of its competitive efforts. The applicant states that it even succeeded in developing a new non-alpha form of perindopril.
99 The Commission submits that that complaint must be rejected. It states, in particular, that, even a mutually acceptable compromise concluded in order to settle genuine disputes may, in some circumstances, infringe Article 101(1) TFEU.
(2) The allegation that the non-challenge clause cannot be classified as an infringement of competition law
100 The applicant submits that the non-challenge clause set out in the settlement agreement which it entered into with Servier cannot be classified as an infringement of competition law (or as a contributory factor to such an infringement) because it was included in a settlement agreement whose purpose was to bring an end to genuine disputes.
101 That conclusion complies, according to the applicant, with the principle set out in recital 1136 of the contested decision.
102 According to the applicant, the non-challenge clause included in the settlement agreement that it entered into with Servier related exclusively to the two patents that were the subject matter of a dispute settled by that agreement and was limited to their scope.
103 According to the applicant, its perindopril unquestionably infringed Servier’s patents. In its view, the Commission relies on hypothetical reasoning to conclude that it was possible for the applicant to introduce a non-infringing technology.
104 The applicant submits, furthermore, that it challenged the validity of the patents prior to the settlement agreement. It also states that that agreement did not have the effect of preventing other competitors from challenging the validity of the patent and that those competitors could rely on the arguments it had developed against the patents at issue.
105 According to the applicant, the Commission stated in the contested decision that there was no obligation to pursue to the end actions for annulment or revocation of competitors’ patents.
106 The Commission relied, in recital 1712 of the contested decision, on the fact that the applicant considered that it had a better case. According to the applicant, the documents intended to establish that do not have probative value since they pre-date the EPO decision of 27 July 2006.
107 Moreover, the applicant relies on point 242 of the Guidelines, on the application of Article 101 [TFEU] to technology transfer agreements (OJ 2014 C 89, p. 3, ‘the 2014 Guidelines on technology transfer agreements’), which acknowledge the possibility of a non-challenge clause in a settlement agreement.
108 The Commission contends, in essence, that the non-challenge obligation prevented Krka from establishing that its technology did not infringe Servier’s patents, and from testing the validity of those patents.
109 It adds that the applicant was in a particular position which enabled it, better than any other generic company, to challenge Servier’s patents.
(3) The allegation that the non-marketing clause cannot be classified as an infringement of competition law
110 The applicant submits that, in accordance with the principle to which the Commission refers in recital 1136 of the contested decision, the non-marketing clause set out in the settlement agreement that it concluded with Servier cannot be classified as an infringement of competition law (or as a contributory factor to such an infringement).
111 According to the applicant, the non-marketing clause set out in the settlement agreement was limited to the scope of the 947 patent, for the duration of the validity of that patent and in the countries where it was considered to be valid.
112 Moreover, again according to the applicant, that clause was an essential part of a genuine settlement the purpose of which was to bring an end to genuine disputes.
113 The applicant points out that it was unable to enter the perindopril market, not on account of the clauses of the settlement agreement, but on account of the exclusive rights granted to Servier by the public authorities in accordance with intellectual property law.It submits that it was lawfully — and not contractually — prevented from entering the alpha perindopril market. Furthermore, that was accepted by the Commission in recital 1720 of the contested decision.
114 Moreover, at the time when the agreements with Servier were signed, a ‘launch at risk’ strategy could not be regarded as being economically viable. Lastly, the applicant was not required to pursue to the end proceedings before the EPO or before the national courts before being able to proceed with a settlement with Servier.
115 The applicant points out, furthermore, that immediately after the 947 patent was declared invalid, it supplied perindopril in alpha form and itself entered the markets. It also points out that it succeeded in developing a new form of non-alpha perindopril.
116 The Commission submits that the non-marketing obligation was intended to prevent the applicant from marketing its perindopril in the Member States where Servier had not authorised it to do so.
117 The Commission adds that the non-marketing obligation changed the situation because it relieved Servier of having to prove that Krka was actually infringing one of its valid patents, which it would otherwise have had to do.
(4) The allegation that the principles relating to the restriction of competition by object were applied incorrectly
118 According to the applicant, by failing to take into account the restrictions of competition resulting from Servier’s valid patents, the Commission failed adequately to examine the economic and legal context of the settlement agreement and therefore could not, on the basis of its analysis and in the light of the case-law of the Court of Justice on restrictions by object, conclude that such a restriction exists in the present case.
119 The applicant concludes by stating that, having regard to four relevant facts, namely, that the settlement was adopted after the EPO decision of 27 July 2006, that it related exclusively to existing genuine disputes, that it did not involve any monetary consideration for the applicant and that the non-challenge and non-marketing clauses were confined to the patents at issue, that settlement is different from the other settlements concluded by Servier and does not constitute a restriction of competition by object for the purposes of Article 101(1) TFEU.
120 The Commission submits that it respected the principles established in the case-law and it refers, in that regard, to recitals 1104 to 1117 of the contested decision, in which it is stated, inter alia, that restrictions ‘by object’ are those which, ‘by their very nature’, can be regarded as being injurious to the proper functioning of normal competition and that there is no need to take into account the actual effects of an agreement which has as its object the prevention, restriction or distortion of competition within the internal market.
2. Findings of the Court
121 In order to respond to the plea raised by the applicant, it is necessary to determine the extent to which a settlement of patent litigation may constitute a restriction of competition by object, and it is therefore necessary, first of all, to set out the applicable law on restrictions of competition by object, intellectual property rights and settlements. Next, it is necessary to clarify how patent settlements may be reconciled with competition law where a commercial agreement, which is in principle autonomous, is coupled, as in the present case, with a settlement agreement. It will then be for the Court, in the light of all the considerations set out above, to assess the facts of the case in the light of the arguments put forward by the applicant.
(a) Restrictions of competition by object
122 Article 101(1) TFEU provides that all agreements between undertakings, decisions by associations of undertakings and concerted practices which have ‘as their object or effect’ the prevention, restriction or distortion of competition within the internal market are to be prohibited as incompatible with the internal market. According to settled case-law since the judgment of 30 June 1966, LTM (56/65, EU:C:1966:38, p. 249), the alternative nature of those requirements, indicated by the use of the conjunction ‘or’, leads to the need to consider, in the first place, the precise purpose of the agreement, in the economic context in which it is to be applied. Where, however, an analysis of the terms of the agreement does not reveal a sufficient degree of harm to competition, the effects of the agreement should then be considered and, for it to be caught by the prohibition, it is necessary to find that factors are present which show that competition has in fact been prevented, restricted or distorted to an appreciable extent (see judgments of 19 March 2015, Dole Food and Dole Fresh Fruit Europe v Commission, C‑286/13 P, EU:C:2015:184, paragraph 116 and the case-law cited, and of 16 July 2015, ING Pensii, C‑172/14, EU:C:2015:484, paragraph 30 and the case-law cited). However, where the anticompetitive object of an agreement is established, it is not necessary to examine its effects on competition (see judgment of 20 January 2016, Toshiba Corporation v Commission, C‑373/14 P, EU:C:2016:26, paragraph 25 and the case-law cited).
123 The concept of restriction of competition by object can be applied only to certain types of coordination between undertakings that reveal, by their very nature, a sufficient degree of harm to the proper functioning of normal competition that it may be found that there is no need to examine their effects (see, to that effect, judgments of 30 June 1966, LTM, 56/65, EU:C:1966:38, p. 249; of 11 September 2014, CB v Commission, C‑67/13 P, EU:C:2014:2204, paragraphs 49, 50 and 58 and the case-law cited; of 16 July 2015, ING Pensii, C‑172/14, EU:C:2015:484, paragraph 31; and of 26 November 2015, Maxima Latvija, C‑345/14, EU:C:2015:784, paragraph 20).
124 According to the case-law of the Court of Justice, in order to determine whether an agreement between undertakings reveals a sufficient degree of harm that it may be considered a ‘restriction of competition by object’ within the meaning of Article 101(1) TFEU, regard must be had to the content of its provisions, its objectives and the economic and legal context of which it forms part (see judgment of 16 July 2015, ING Pensii, C‑172/14, EU:C:2015:484, paragraph 33 and the case-law cited). When determining the economic and legal context, it is also necessary to take into consideration the nature of the goods or services affected, as well as the real conditions of the functioning and structure of the market or markets in question (see judgment of 19 March 2015, Dole Food and Dole Fresh Fruit Europe v Commission, C‑286/13 P, EU:C:2015:184, paragraph 117 and the case-law cited). Nevertheless, it must be borne in mind that the examination of the real conditions of the functioning and structure of the market in question cannot lead the General Court to assess the effects of the coordination concerned (see, to that effect, judgment of 11 September 2014, CB v Commission, C‑67/13 P, EU:C:2014:2204, paragraphs 72 to 82), since otherwise the distinction established in Article 101(1) TFEU would lose its effectiveness.
(b) Intellectual property rights and, in particular, patents
125 According to settled case-law, the right to property, which includes intellectual property rights, constitutes a general principle of EU law (judgment of 29 January 2008, Promusicae, C‑275/06, EU:C:2008:54, paragraph 62; see also, to that effect, judgment of 12 July 2005, Alliance for Natural Health and Others, C‑154/04 and C‑155/04, EU:C:2005:449, paragraph 126 and the case-law cited).
126 In addition, it must be noted that intellectual property rights are protected by the Charter of Fundamental Rights of the European Union. Under Article 17(1) of the Charter of Fundamental Rights, to which the Treaty of Lisbon has conferred the same legal value as the Treaties (Article 6(1) TEU), ‘[e]veryone has the right to own, use, dispose of and bequeath his or her lawfully acquired possessions’, ‘[n]o one may be deprived of his or her possessions, except in the public interest and in the cases and under the conditions provided for by law, subject to fair compensation being paid in good time for their loss’ and ‘[t]he use of property may be regulated by law in so far as is necessary for the general interest’. Article 17(2) of the Charter of Fundamental Rights states, moreover, that ‘[i]ntellectual property shall be protected’. Consequently, the guarantees provided for in Article 17(1) of the Charter of Fundamental Rights apply also to intellectual property. The Court of Justice has held that the recognition of intellectual property rights in the Charter of Fundamental Rights entails a need for a high level of protection of those rights and that it is necessary to strike a balance between maintaining free competition — in respect of which primary law and, in particular, Articles 101 and 102 TFEU prohibit anticompetitive agreements, decisions and concerted practices and abuses of a dominant position — and the requirement to safeguard the patent holder’s intellectual-property rights, guaranteed by Article 17(2) of the Charter of Fundamental Rights (see, to that effect, judgment of 16 July 2015, Huawei Technologies, C‑170/13, EU:C:2015:477, paragraphs 42 and 58).
127 As regards the reconciliation of patent law with competition law, the Court of Justice has ruled that it is possible that the provisions of Article 101 TFEU may apply to intellectual property law if the use of one or more patents, in concert between undertakings, should lead to the creation of a situation which may come within the concepts of agreements between undertakings, decisions of associations of undertakings or concerted practices within the meaning of Article 101 TFEU (judgment of 29 February 1968, Parke, Davis and Co., 24/67, EU:C:1968:11, p. 71). It further considered, in 1974, that although the existence of rights recognised under the industrial property legislation of a Member State is not affected by Article 101 TFEU, the conditions under which those rights may be exercised may nevertheless fall within the prohibitions contained in that article and that this may be the case whenever the exercise of such a right appears to be the object, the means or the consequences of an agreement (judgment of 31 October 1974, Centrafarm and de Peijper, 15/74, EU:C:1974:114, paragraphs 39 and 40).
128 Lastly, it must be noted that the specific purpose of awarding a patent is, inter alia, to ensure that the patentee, in order to reward the creative effort of the inventor, has the exclusive right to use an invention with a view to manufacturing industrial products and putting them into circulation for the first time, either directly or by the grant of licences to third parties, as well as the right to oppose infringements (judgment of 31 October 1974, Centrafarm and de Peijper, 15/74, EU:C:1974:114, paragraph 9). When granted by a public authority, a patent is normally assumed to be valid and an undertaking’s ownership of that right is assumed to be lawful. The mere possession by an undertaking of such an exclusive right normally results in keeping competitors away, since public regulations require them to respect that exclusive right (judgment of 1 July 2010, AstraZeneca v Commission, T‑321/05, EU:T:2010:266, paragraph 362).
(c) Patent dispute settlement
129 It is a priori legitimate for the parties to a dispute relating to a patent to conclude a settlement agreement rather than pursuing litigation before a court. As the Commission rightly stated in recital 1102 of the contested decision, companies are generally entitled to settle litigation, including patent litigation, and those settlements often benefit both parties to the dispute and allow for a more efficient allocation of resources than if litigation were to be pursued to judgment. An applicant is not required to pursue litigation which it voluntarily initiated. It should be added that the settlement of disputes before the courts, in addition to the fact that it generates a cost for society, cannot be regarded as the preferred and ideal route for conflict resolution. An increase in litigation before the courts may reflect failures or shortcomings which could be remedied in other ways or be dealt with by appropriate prevention actions. If the national systems for granting patents or that of the EPO were experiencing such difficulties, for example by being too liberal in granting protection to processes which are devoid of inventive character, those problems could not justify an obligation or even an incentive for undertakings to pursue patent disputes until a judicial outcome is reached.
130 In addition, paragraphs 204 and 209 of the 2004 Guidelines on technology transfer agreements, which are applicable at the very least to agreements concerning the licensing of technology, acknowledge the possibility of concluding settlement and non-assertion agreements which include the granting of licences and indicate that, in the context of such a settlement and non-assertion agreement, non-challenge clauses are generally considered to fall outside the scope of Article 101(1) TFEU. Point 235 of the 2014 Guidelines on technology transfer agreements, which replaced the 2004 Guidelines, also states that ‘settlement agreements in the context of technology disputes are, as in many other areas of commercial disputes, in principle a legitimate way to find a mutually acceptable compromise to a bona fide legal disagreement’. That paragraph also states that ‘[t]he parties may prefer to discontinue the dispute or litigation because it proves to be too costly, time-consuming and/or uncertain as regards its outcome’, and that ‘[s]ettlements can also save courts and/or competent administrative bodies effort in deciding on the matter and can therefore give rise to welfare enhancing benefits’.
131 It follows from all of the foregoing that, for the purposes of reconciling patent law and competition law in the particular context of settlements between parties to a patent dispute, a balance must be struck between, on the one hand, the need to allow undertakings to make settlements, the increased use of which is beneficial for society and, on the other hand, the need to prevent the risk of misuse of settlement agreements, contrary to competition law, leading to entirely invalid patents being maintained and, especially in the medicinal products sector, an unjustified financial burden for public budgets (see, to that effect, judgment delivered today, Servier and Others v Commission, T‑691/14, paragraphs 219 to 252).
(d) The reconciliation of patent settlement agreements and competition law
132 It should be noted that the use of a settlement to resolve a patent dispute does not exempt the parties from the application of competition law (see, to that effect, judgments of 27 September 1988, Bayer and Maschinenfabrik Hennecke, 65/86, EU:C:1988:448, paragraph 15, and of 8 September 2016, Lundbeck v Commission, T‑472/13, under appeal, EU:T:2016:449, paragraph 118; see, by analogy, judgment of 30 January 1985, BAT Cigaretten-Fabriken v Commission, 35/83, EU:C:1985:32, paragraph 33; see, also, paragraph 204 of the 2004 Guidelines on technology transfer agreements and point 237 of the 2014 Guidelines on technology transfer agreements).
133 The Court of Justice has thus held, in particular, that a non-challenge clause in respect of a patent, including when it was inserted into an agreement intended to settle a dispute pending before a court, might, in the light of the legal and economic context, restrict competition within the meaning of Article 101(1) TFEU (judgment of 27 September 1988, Bayer and Maschinenfabrik Hennecke, 65/86, EU:C:1988:448, paragraphs 14 to 16).
134 It is therefore necessary to identify the relevant factors which justify a conclusion that a non-challenge clause in respect of a patent and, more broadly, a patent settlement agreement restricts competition by object, bearing in mind that determining whether there is a restriction by object entails an examination of the content of the terms of the agreement in question, its objectives, and its economic and legal context (judgment of 11 September 2014, CB v Commission, C‑67/13 P, EU:C:2014:2204, paragraph 53).
135 As a preliminary point, it should be noted that a patent dispute settlement agreement may have no negative impact on competition. That is the case, for example, if the parties agree that the patent at issue is not valid and therefore provide for the immediate market entry of the generic company.
136 The agreements at issue in the present case do not fall into that category because they contain non-challenge clauses in respect of patents and non-marketing clauses in respect of products, which are, by themselves, restrictive of competition. The non-challenge clause undermines the public interest in eliminating any obstacle to economic activity which may arise where a patent was granted in error (see, to that effect, judgment of 25 February 1986, Windsurfing International v Commission, 193/83, EU:C:1986:75, paragraph 92) and the non-marketing clause entails the exclusion from the market of one of the patent holder’s competitors.
137 Nevertheless, the insertion of such clauses may be legitimate, but only in so far as it is based on the parties’ recognition of the validity of the patent in question (and, consequently, of the infringing nature of the generic products concerned).
138 First, non-marketing and non-challenge clauses are necessary for the settlement of some disputes related to patents. If the parties to a dispute were unable to make use of such clauses, the settlement of the dispute would be of no interest in cases in which both parties agree on the validity of the patent. It must, moreover, be noted in this connection that the Commission stated, in paragraph 209 of the 2004 Guidelines on technology transfer agreements, that ‘[i]t is inherent in [settlement agreements] that the parties agree not to challenge ex post the intellectual property rights covered by the agreement [since] the very purpose of the agreement is to settle existing disputes and/or to avoid future disputes’. It is equally necessary, in order to achieve that purpose, that the parties agree that no infringing product may be marketed.
139 Secondly, the insertion of non-marketing clauses merely, in part, reinforces the pre-existing legal effects of a patent which the parties explicitly or implicitly recognise as valid. A patent normally enables its holder to prevent its competitors from marketing the product covered by the patent or a product obtained through the process covered by the patent (judgment of 1 July 2010, AstraZeneca v Commission, T‑321/05, EU:T:2010:266, paragraph 362). By agreeing to a non-marketing clause, the generic company undertakes not to sell products likely to infringe the patent in question. If that clause is limited to the scope of the patent at issue, it may be regarded as essentially duplicating the effects of that patent, in so far as it is based on the recognition of the validity of that patent. As regards non-challenge clauses, the patent cannot be interpreted as affording protection against actions brought in order to challenge the validity of a patent (judgment of 25 February 1986, Windsurfing International v Commission, 193/83, EU:C:1986:75, paragraph 92). The effects of those clauses therefore do not overlap with the effects of the patent. However, when a non-challenge clause is adopted as part of the settlement of a genuine dispute in which the competitor has already had the opportunity to challenge the validity of the patent concerned and ultimately acknowledges that validity, such a clause cannot be regarded, in that context, as undermining the public interest in eliminating any obstacle to economic activity which may arise where a patent was granted in error (see paragraph 136 above).
140 The Commission itself stated, in the contested decision, that non-challenge clauses and non-marketing clauses were generally inherent in any settlement. It thus considered that ‘when in a patent dispute or patent litigation, a settlement is reached on the basis of each party’s assessment of the patent case before them, such a patent settlement is unlikely to infringe competition law even though it may contain an obligation on the generic company not to use the invention covered by the patent during the period of patent protection (e.g. a non-compete clause) and/or an obligation not to challenge the patent concerned in court (e.g. a non-challenge clause)’ (recital 1136 of the contested decision).
141 Thus, the mere presence, in settlement agreements, of non-marketing clauses and non-challenge clauses whose scope is limited to that of the patent in question does not — despite the fact that those clauses are, by themselves, restrictive (see paragraph 136 above) — justify a finding of a restriction of competition sufficiently harmful to be described as a restriction by object, where those agreements are based on the recognition, by the parties, of the validity of the patent (and, consequently, the infringing nature of the generic products concerned).
142 The presence of non-marketing and non-challenge clauses whose scope is limited to that of the patent in question is, however, problematic when it is apparent that the generic company’s agreement to those clauses is not based on its recognition of the validity of the patent. As the Commission rightly points out, ‘even if the limitations in the [settlement] agreement on the generic undertaking’s commercial autonomy do not go beyond the material scope of the patent, they constitute a breach of Article 101 [TFEU] when those limitations cannot be justified and do not result from the parties’ assessment of the merits of the exclusive right itself’ (recital 1137 of the contested decision).
143 In that respect, it should be noted that the existence of a ‘reverse payment’, that is to say a payment from the originator company to the generic company, is doubly suspect in the context of a settlement agreement. In the first place, it must be borne in mind that a patent is intended to reward the creative effort of the inventor by allowing him to make a fair profit from his investment (judgment of 31 October 1974, Centrafarm and de Peijper, 15/74, EU:C:1974:114, paragraph 9) and that a valid patent must, in principle, allow a transfer of value to its holder — for example, through a licence agreement — and not vice versa. In the second place, the existence of a reverse payment gives rises to doubts as to whether the settlement is actually based on the recognition, by the parties to the agreement, of the validity of the patent in question.
144 However, the mere presence of a reverse payment does not mean that there is a restriction by object. It is possible that some reverse payments, where they are inherent in the settlement of the dispute in question, may be justified. However, where an unjustified reverse payment occurs in the conclusion of the settlement, the generic company must then be regarded as having been induced by that payment to agree to the non-marketing and non-challenge clauses and it must be concluded that there is a restriction by object. In that case, the restrictions of competition introduced by the non-marketing and non-challenge clauses no longer relate to the patent and to the settlement, but rather can be explained by the conferral of a benefit inducing the generic company to abandon its competitive efforts.
145 It must be pointed out that, although neither the Commission nor the Courts of the European Union are competent to rule on the validity of the patent, it is nevertheless the case that those institutions may, in the context of their respective powers and without ruling on the intrinsic validity of the patent, find that it has been used abnormally, in a manner which has no relation to its specific subject matter (see, to that effect, judgments of 29 February 1968, Parke, Davis and Co., 24/67, EU:C:1968:11, pp. 71 and 72, and of 31 October 1974, Centrafarm and de Peijper, 15/74, EU:C:1974:114, paragraphs 7 and 8; see also, by analogy, judgments of 6 April 1995, RTE and ITP v Commission, C‑241/91 P and C‑242/91 P, EU:C:1995:98, paragraph 50, and of 4 October 2011, Football Association Premier League and Others, C‑403/08 and C‑429/08, EU:C:2011:631, paragraphs 104 to 106).
146 Inducing a competitor to accept non-marketing and non-challenge clauses, in the sense described in paragraph 144 above, or its corollary, accepting such clauses because of an inducement, constitutes an abnormal use of the patent.
147 As the Commission rightly stated in recital 1137 of the contested decision, ‘patent law does not provide for a right to pay actual or potential competitors to stay out of the market or to refrain from challenging a patent prior to entering the market’. Likewise, according to the Commission, ‘patent holders are not entitled to pay generic companies to keep them off the market and reduce the risks of competition, whether in the context of a patent settlement agreement or otherwise’ (recital 1141 of the contested decision). Lastly, the Commission correctly added that ‘paying or otherwise inducing potential competitors to stay out of the market [was] not part of any patent right, nor [was] it one of the means provided for under patent law to enforce the patent’ (recital 1194 of the contested decision).
148 Where an inducement has been found, the parties may no longer rely on their recognition, in the context of the settlement, of the validity of the patent. The fact that the validity of the patent is confirmed by a judicial or administrative body is, in that regard, irrelevant.
149 It is then the inducement, and not the recognition of the validity of the patent by the parties to the settlement, which must be regarded as the real cause of the restrictions of competition introduced by the non-marketing and non-challenge clauses (see paragraph 136 above), which — since they are in that case entirely illegitimate — therefore reveal a sufficient degree of harm to the proper functioning of normal competition that a restriction by object may be found.
150 Where they involve an inducement, the agreements in question must therefore be regarded as market exclusion agreements, in which the ‘stayers’ are to compensate the ‘goers’. Such agreements actually constitute a buying-off of competition and must therefore be classified as restrictions of competition by object, as follows from the judgment of 20 November 2008, Beef Industry Development Society and Barry Brothers (C‑209/07, EU:C:2008:643, paragraphs 8 and 31 to 34), and the Opinion of Advocate General Trstenjak in Beef Industry Development Society and Barry Brothers, (C‑209/07, EU:C:2008:467, point 75), referred to in inter alia recitals 1139 and 1140 of the contested decision. Moreover, the exclusion of competitors from the market constitutes an extreme form of market sharing and of limitation of production (judgment of 8 September 2016, Lundbeck v Commission, T‑472/13, under appeal, EU:T:2016:449, paragraph 435), which, in a context such as that of the agreements in question, reveals a degree of harm which is all the greater since the companies excluded are generic companies, the market entry of which is, as a rule, favourable to competition and which also contributes to the public interest in lowering the cost of healthcare. Lastly, that market exclusion is augmented, in the agreements at issue, by the fact that it is not possible for the generic company to challenge the patent at issue.
151 It follows from all of the foregoing that, in the context of patent dispute settlement agreements, a finding of a restriction of competition by object presupposes that the settlement agreement contains both an inducement in the form of a benefit for the generic company and a corresponding limitation of the generic company’s efforts to compete with the originator company. Where those two conditions are met, a finding of restriction of competition by object must be made in view of the harmfulness of that agreement to the proper functioning of normal competition.
152 Thus, where a patent settlement agreement contains non-marketing and non-challenge clauses, whose inherently restrictive nature (see paragraph 136 above) has not been validly called into question, the existence of an inducement for the generic company to agree to those clauses permits the conclusion that there is a restriction by object, and does so even if there is a genuine dispute, the settlement agreement includes non-marketing and non-challenge clauses whose scope does not exceed that of the patent at issue and that patent could — having regard, in particular, to the decisions adopted by the competent administrative authorities or courts — legitimately be regarded as valid by the parties to the agreement at issue at the time it was adopted.
153 In the contested decision, the Commission rightly examined whether the settlement agreement at issue in the present case involved a value transfer from the originator company to the generic company representing a ‘significant’ inducement, that is to say liable to lead the latter to accept non-marketing and non-challenge clauses, and concluded, having found such an inducement, that there was a restriction of competition by object.
154 The Commission thus adopted the inducement criterion, referred to below as the ‘inducement’ or ‘inducive benefit’ criterion, for the purpose of distinguishing settlement agreements which constitute restrictions by object from those which do not constitute such restrictions.
155 In some cases, the settlement agreement may provide for a reverse payment, that is to say a transfer of value without any quid pro quo, intended, inter alia, to offer compensation to the generic company. That payment then constitutes an inducement since it does not cover costs inherent in the settlement of the dispute. In other cases, as in the present case, the settlement agreement may be accompanied by a commercial agreement described as a ‘side deal’ since it is linked to the settlement of the dispute. It must be clarified how, in the latter case, the existence of an inducement may be established by the Commission.
(e) Side deals
156 It follows from Article 2 of Regulation No 1/2003 and from settled case-law that, in the field of competition law, where there is a dispute as to the existence of an infringement, it is incumbent on the Commission to prove the infringements found by it and to adduce evidence capable of demonstrating to the requisite legal standard the existence of the circumstances constituting an infringement (judgments of 17 December 1998, Baustahlgewebe v Commission, C‑185/95 P, EU:C:1998:608, paragraph 58, and of 8 July 1999, Commission v Anic Partecipazioni, C‑49/92 P, EU:C:1999:356, paragraph 86; see, also, judgment of 12 April 2013, CISAC v Commission, T‑442/08, EU:T:2013:188, paragraph 91 and the case-law cited).
157 In that context, any doubt on the part of the Court must operate to the advantage of the undertaking to which the decision finding an infringement was addressed. The Court cannot therefore conclude that the Commission has established the infringement in question to the requisite legal standard if it still entertains any doubts on that point, in particular in proceedings for annulment of a decision imposing a fine (see judgment of 12 April 2013, CISAC v Commission, T‑442/08, EU:T:2013:188, paragraph 92 and the case-law cited).
158 It is necessary to take into account the principle of the presumption of innocence resulting in particular from Article 48 of the Charter of Fundamental Rights. Given the nature of the infringements in question and the nature and degree of severity of the penalties which may ensue, the presumption of innocence applies, inter alia, to the procedures relating to infringements of the competition rules applicable to undertakings that may result in the imposition of fines or periodic penalty payments (see judgment of 12 April 2013, CISAC v Commission, T‑442/08, EU:T:2013:188, paragraph 93 and the case-law cited).
159 In addition, account must be taken of the non-negligible stigma attached to a finding of involvement in an infringement of the competition rules for a natural or legal person (see judgment of 12 April 2013, CISAC v Commission, T‑442/08, EU:T:2013:188, paragraph 95 and the case-law cited).
160 Thus, the Commission must show precise and consistent evidence in order to establish the existence of the infringement and to support the firm conviction that the alleged infringement constitutes a restriction of competition within the meaning of Article 101(1) TFEU (see judgment of 12 April 2013, CISAC v Commission, T‑442/08, EU:T:2013:188, paragraph 96 and the case-law cited).
161 It is not necessary for every item of evidence produced by the Commission to satisfy those criteria in relation to every aspect of the infringement. It is sufficient if the set of indicia relied on by the Commission, viewed as a whole, meets that requirement (see judgment of 12 April 2013, CISAC v Commission, T‑442/08, EU:T:2013:188, paragraph 97 and the case-law cited).
162 The existence of an anticompetitive practice or agreement must sometimes even be inferred from a number of coincidences and indicia which, taken together, may, in the absence of another plausible explanation, constitute evidence of an infringement of the competition rules (judgment of 7 January 2004, Aalborg Portland and Others v Commission, C‑204/00 P, C‑205/00 P, C‑211/00 P, C‑213/00 P, C‑217/00 P and C‑219/00 P, EU:C:2004:6, paragraph 57).
163 For example, although parallel behaviour may not by itself be identified with a concerted practice, it may, however, amount to a strong indication of such a practice if it leads to conditions of competition which do not correspond to the normal conditions of the market (judgment of 14 July 1972, Farbenfabriken Bayer v Commission, 51/69, EU:C:1972:72, paragraph 25).
164 Likewise, the presence of a ‘side deal’ — the expression used by the Commission in recital 1190 of the contested decision — may constitute, as regards the settlement of a patent dispute, a strong indication of the existence of an inducement and, consequently, of a restriction of competition by object (see paragraphs 144 to 152 above).
165 It should be explained in that respect that a side deal is a normal commercial agreement linked to a settlement agreement which contains clauses which are by themselves restrictive (see paragraph 136 above). Such a link exists, in particular, where the two agreements are concluded on the same day, where they are legally linked, the binding nature of one of the agreements being conditional upon the conclusion of the other agreement, or where, in the light of the context in which they are concluded, the Commission is able to establish that they are indissociable.It may be added that, the shorter the time between the conclusion of each agreement, the easier it will be for the Commission to establish that indissociable nature.
166 It should also be noted that the fact that the settlement agreement and the side deal are concluded on the same day or that there is a contractual link between them is an indication that those agreements form part of a single contractual framework. If those agreements were not concluded on the same day (and if there were no contractual link between them), one of the parties to the negotiation would grant the other party everything it wants without any certainty of ultimately obtaining the expected quid pro quo. That temporal or legal link between the two agreements is also an indication that they were negotiated together.
167 The side deal is a normal commercial agreement that could exist independently without the settlement of a dispute being at issue. Likewise, the conclusion of a settlement agreement does not require the concurrent conclusion of a commercial agreement. Thus, the two agreements do not need to be linked. Moreover, that linkage cannot be justified by the settlement of a dispute because the purpose of a side deal is not to reach such a settlement but rather to carry out a commercial transaction.
168 In addition, a side deal involves value transfers, of a financial or non-financial nature, between the parties. It may involve, in particular, the transfer of value from the patent holder to the generic company.
169 There is therefore a risk that the linking of a commercial agreement with a settlement agreement containing non-marketing and non-challenge clauses, which are, by themselves, restrictive of competition (see paragraph 136 above), is actually intended — under the guise of a commercial transaction, taking the form, as the case may be, of a complex contractual arrangement — to induce the generic company to accept those clauses, through a value transfer provided for in the side deal.
170 Consequently, the fact that a commercial agreement, which does not normally have the settlement of a dispute as its subject matter (see paragraph 167 above), and which serves as a vehicle for a transfer of value from the originator company to the generic company, is, in the circumstances set out in paragraph 165 above, linked with a settlement agreement containing competition-restricting clauses is a strong indication of the existence of a reverse payment (see paragraph 143 above).
171 However, the strong indication referred to in paragraph 170 above is not sufficient and the Commission must therefore support it with other consistent evidence justifying the conclusion that there is a reverse payment. Such a payment, in the specific context of side deals, corresponds to the part of the payment made by the originator company which exceeds the ‘normal’ value of the asset traded (or, as the case may be, to the part of the ‘normal’ value of the asset traded which exceeds the payment made by the generic company).
172 It must be emphasised, in that regard, that the Commission stated on several occasions in the contested decision that certain side deals concluded by Servier with generic companies had not been negotiated ‘at arm’s length’ (recitals 1351, 1950 and 1952).
173 It should be noted that the concept of ‘normal competitive conditions’, which is similar to that of ‘arm’s length’, even though it is not used in relation to agreements, decisions and concerted practices, is not alien to competition law, since it is used in the particular field of State aid in order to determine whether a State has acted like a private investor (judgment of 2 September 2010, Commission v Scott, C‑290/07 P, EU:C:2010:480, paragraph 68), that is to say, whether the advantage granted to the undertakings in question constitutes the normal remuneration for a quid pro quo obtained by the State.That concept may therefore constitute, by analogy, a relevant reference parameter when determining whether two companies that concluded a commercial transaction did so on the basis of economic considerations limited to the economic value of the asset traded, that is to say, for example, to its prospects of profitability, and, thus, at arm’s length.
174 Where there are indicia or evidence put forward by the Commission in order to support a finding that the side deal was not concluded at arm’s length, the parties to the agreements may present their version of the facts, supporting their claims with the evidence that they are able to put forward and which permit the conclusion that the commercial agreement, although linked to the settlement agreement, is justified by reasons other than the exclusion of a competitor by means of a reverse payment. The parties to the agreements may thus argue that the side deal was concluded at arm’s length by adducing relevant evidence concerning, for example, the industrial and commercial practices in the sector or the particular circumstances of the case.
175 In the light of all the evidence available to it and, as the case may be, the lack of an explanation or the lack of a plausible explanation from the parties to the agreements in question, the Commission may be justified in finding, following an overall assessment, that the side deal was not concluded at arm’s length, that is to say that the payment made by the originator company exceeds the value of the asset traded (or that the value of the asset transferred to the generic company exceeds the payment made by the latter). The Commission may thus conclude that there is a reverse payment (see paragraph 171 above).
176 A reverse payment, if it is not intended to compensate for costs inherent in the settlement, therefore constitutes an inducive benefit (see paragraph 144 above). That is the case where the purpose of a side deal is not to settle a dispute but rather to carry out a commercial transaction (see paragraph 167 above).
177 However, the parties to the agreement may still argue that the benefit in question is insignificant, if the amount of that benefit is insufficient to be regarded as a significant inducement to accept the competition-restricting clauses set out in the settlement agreement (see paragraph 153 above).
178 It should also be noted that, among side deals, the licence agreement has particular features which require a specific analysis of the conditions under which such an agreement may constitute an inducement leading to a finding of a restriction of competition by object within the meaning of Article 101 TFEU.
(f) Licence agreements
179 By way of exception to the considerations relating to side deals set out in paragraphs 164 to 170 above, the linking of a normal commercial agreement to a settlement agreement containing non-challenge and non-marketing clauses no longer constitutes a strong indication of a reverse payment where the commercial agreement in question is a licence agreement concerning the patent in dispute.
180 That exception can be explained by the fact that, while it is true that a licence agreement in relation to a patent does not have as its subject matter the settlement of a dispute, but rather the grant of permission to use that patent, it may nevertheless be justifiable — in contrast to the situation as regards other commercial agreements (see paragraph 167 above) — to link that licence agreement to a settlement agreement concerning a patent which is the subject matter of the licence.
181 In principle, a patent dispute arises when the generic company’s wish to enter the market comes into conflict with the patent owner’s wish to safeguard the rights that he derives from that patent. Authorising such entry by concluding a licence agreement thus appears to be a particularly appropriate means of resolving the dispute, since it satisfies the wishes of both parties to the dispute.
182 It is also acknowledged that the use of a licence agreement is an appropriate means of resolving a dispute. That is apparent from paragraph 204 of the 2004 Guidelines on technology transfer, according to which ‘licensing may serve as a means of settling disputes’. That paragraph is incorporated in paragraph 205 of the 2014 Guidelines on technology transfer agreements.
183 Linking a licence agreement to a settlement agreement is all the more justified since the presence, in a settlement agreement, of non-marketing and non-challenge clauses is legitimate only where that agreement is based on the parties’ recognition of the validity of the patent (see paragraphs 137 to 140 above). The conclusion of a licence agreement, which, for any licensee, makes sense only if the licence is actually used, is also based on the parties’ recognition of the validity of the patent. To that extent, the licence agreement thus supports the legitimacy of the settlement agreement, which fully justifies the linking of the two.
184 Since it appears justified to link a patent dispute settlement agreement to a licence agreement concerning the same patent, that linking, unlike the situation as regards the other side deals, does not constitute a strong indication of the existence of a reverse payment (within the meaning of that expression in relation to side deals, see paragraph 171 above).
185 It is therefore for the Commission to rely on indicia other than the mere linking of the licence agreement and the settlement agreement for the purpose of establishing that the licence agreement was not concluded at arm’s length and that it actually masks a reverse payment inducing the generic company to accept the non-marketing and non-challenge clauses (see paragraphs 170 to 175 above).
186 It should be noted that a finding of the existence of a reverse payment is less evident in the case of a licence agreement because such an agreement does not entail a financial transfer from the originator company to the generic company, but rather from the generic company to the originator company. Thus, in a licence agreement, the licensee pays a royalty to the patent holder.
187 There is, however, a transfer of value from the originator company to the generic company, since the royalty paid to the patent holder constitutes a quid pro quo for the benefit that the generic company receives from the licence agreement, namely the authorisation to use the patent in order to enter the market without risk.
188 It is therefore for the Commission to demonstrate that that quid pro quo is abnormally low, that is to say to such an extent that it cannot be explained by considerations limited to the economic value of the asset to which the contract relates (see paragraph 173 above), and that the licence agreement thus involves a reverse payment to the generic company.
189 It must be particularly clear that the transaction in question was not concluded at arm’s length in order to establish a sufficient degree of harmfulness for the purpose of classifying the settlement agreement as a restriction of competition by object, since the restriction of competition by the non-challenge and non-marketing clauses in the settlement agreement is mitigated by the licence agreement.
190 The non-marketing clause is thus rendered ineffective, at least in part. The licence agreement goes even further than a mere partial neutralisation of the effects of that clause, since it encourages the entry of generic products on the market by eliminating the litigation risk associated with the patent.
191 As regards the non-challenge clause, although its restrictive effects persist, they are limited by the fact that the licence allows market entry without a litigation risk. Although it is essential for the generic company to be able to challenge the validity of the patent when it enters the market at risk, that is less the case when it is authorised by the originator company to enter this market through a licence agreement.
192 At this stage of the analysis, it should be noted that, in the context of patent dispute settlement agreements, a finding of a restriction of competition by object presupposes that the settlement agreement contains both an inducement to the generic company and a corresponding limitation of the generic company’s efforts to compete with the originator company (see paragraph 151 above). It follows from the foregoing that, where there is a licence agreement, those two elements are mitigated, or even absent, with the result that a sufficient degree of harm to the proper functioning of normal competition (see, to that effect, judgment of 11 September 2014, CB v Commission, C‑67/13 P, EU:C:2014:2204, paragraphs 49 and 50 and the case-law cited) cannot easily be identified.
193 It should be added that the exception mentioned in paragraph 179 above is not contradicted by the fact that the linking of a licence agreement and a non-challenge clause are among the restrictions excluded from the exemption laid down in Article 2 of Regulation No 772/2004, or by the case-law of the Court of Justice, first set out in the judgment of 25 February 1986, Windsurfing International v Commission (193/83, EU:C:1986:75, paragraphs 89 and 92), and clarified in the judgment of 27 September 1988, Bayer and Maschinenfabrik Hennecke (65/86, EU:C:1988:448).
194 First, according to Article 5 of Regulation No 772/2004, the linking of a licence agreement and a non-challenge clause is one of the restrictions excluded from the exemption provided for in Article 2 of that regulation. However, that exemption, as well as that exclusion, apply, pursuant to Articles 2 and 5 of that regulation, only in so far as the agreements in question contain restrictions of competition falling within the scope of Article 101(1) TFEU. Consequently, the fact that the linking of a licence agreement and a non-challenge clause is one of the restrictions excluded from the exemption provided for in Article 2 of Regulation No 772/2004 does not support the conclusion that such linking is, in all circumstances, a restriction of competition within the meaning of Article 101(1) TFEU and, in particular, a restriction by object.
195 In that respect, the Court of Justice held that, whilst it is true that to grant the benefit of Article 101(3) to a given agreement presupposes that this agreement falls within the prohibition imposed by Article 101(1), the authorisation in Article 101(3) to grant that same benefit to categories of agreements does not imply that because a particular agreement comes within those categories it necessarily fits the descriptions set out in Article 101(1). Therefore, to grant exemptions by categories cannot amount, even by implication, to passing any pre-conceived judgment on any agreement considered individually (judgment of 13 July 1966, Italy v Council and Commission, 32/65, EU:C:1966:42, pp. 405 and 406).
196 Secondly, the Court of Justice indeed held that a clause in a licence agreement obliging the licensee not to challenge the validity of the patent was incompatible with Article 101(1) TFEU. It added that such a clause clearly does not fall within the specific subject matter of the patent, which cannot be interpreted as also affording protection against actions brought in order to challenge the patent’s validity, in view of the fact that it is in the public interest to eliminate any obstacle to economic activity which may arise where a patent was granted in error (judgment of 25 February 1986, Windsurfing International v Commission, 193/83, EU:C:1986:75, paragraphs 89 and 92).
197 However, in a judgment delivered two years later, in a case concerning a settlement agreement, the Court qualified the position it had adopted in the judgment of 25 February 1986, Windsurfing International v Commission (193/83, EU:C:1986:75), this time holding only that a non-challenge clause included in a patent licensing agreement may, in the light of the legal and economic context, restrict competition within the meaning of Article 101(1) TFEU (judgment of 27 September 1988, Bayer and Maschinenfabrik Hennecke, 65/86, EU:C:1988:448, paragraph 16). Although it also rejected, in that same judgment, the Commission’s proposal that the inclusion of a non-challenge clause fell outside the prohibition laid down in Article 101(1) TFEU where the agreement in question was intended to settle litigation pending before a court, it did not, however, conclude that all patent settlement agreements containing such a clause fell within the prohibition laid down in Article 101(1) TFEU.
198 It is true that the licensees under a licence agreement, are, as is clear from paragraph 112 of the 2004 Guidelines on technology transfer agreements, ‘normally in the best position to determine whether or not an intellectual property right is invalid’ and therefore to challenge it. That is why the linking of a licence agreement and a non-challenge clause is, in principle, prohibited (Opinion of Advocate General Darmon in Bayer and Maschinenfabrik Hennecke, 65/86, EU:C:1987:336, paragraph 8). However, where a licence agreement is concluded in the context of the settlement of a genuine dispute involving litigation between the parties concerned, the licensee has already had the opportunity to challenge the validity of the patent in question and if, ultimately, he agrees, without being induced, to a non-challenge clause (and a non-marketing clause), it is because he believes that the patent is valid. In that particular context of a settlement in which the parties ultimately agree that the patent is valid, the basis for prohibiting the linking of a licence agreement and a non-challenge clause no longer appears relevant, provided that the settlement agreement is based on the recognition by the parties to the agreement of the validity of the patent in question, and not on an inducement to the licensee to accept the non-challenge clause (and the non-marketing clause).
199 It follows from the foregoing that, where there is a genuine dispute involving litigation between the parties concerned and a licence agreement that is directly connected with the settlement of that dispute, the linking of that agreement to the settlement agreement does not constitute a strong indication of the existence of a reverse payment. In such circumstances, it is therefore for the Commission to demonstrate, on the basis of other evidence, that the licence agreement does not constitute a transaction concluded at arm’s length and thus masks a reverse payment (within the meaning of that expression in relation to side deals, see paragraph 171 above).
200 It must be determined, in the light of the foregoing considerations, whether the Commission was entitled to conclude, in the present case, that the settlement and licence agreements concluded between Servier and Krka could be classified as a restriction by object.
(g) The facts of the case
201 It is necessary, in the first place, to examine whether there were genuine disputes and whether the licence agreement appeared to have a sufficiently direct connection with the settlement of those disputes to justify its linking to the settlement agreement.
202 In that regard, first, it should be noted that there were genuine ongoing disputes between Servier and Krka at the time the agreement was signed and that those disputes came to an end following the settlement agreement, which provided, in Article I(i) and (ii), that both parties were to withdraw from the ongoing proceedings between them.
203 In 2004, ten generic companies, including Krka, had filed opposition proceedings against the 947 patent before the EPO, seeking the revocation of that patent in its entirety on grounds of lack of novelty, lack of inventive step and insufficient disclosure of the invention. On 27 July 2006, the EPO’s Opposition Division confirmed the validity of that patent following minor amendments to Servier’s original claims. Seven companies then brought an appeal against the EPO decision of 27 July 2006. Krka withdrew from the opposition proceedings on 11 January 2007, pursuant to the settlement agreement concluded with Servier.
204 Likewise, Servier had brought an action for infringement of the 340 patent against Krka before the High Court of Justice (England and Wales), Chancery Division (Patents Court). On 2 August 2006, it had also brought an action for infringement of the 947 patent against Krka and applied for an interim injunction. On 1 September 2006, Krka had brought a counterclaim for annulment of the 947 patent and, on 8 September 2006, a separate counterclaim for annulment of the 340 patent. On 3 October 2006, the High Court of Justice (England and Wales), Chancery Division (Patents Court), granted Servier’s application for an interim injunction and denied the motion brought by Krka on 1 September 2006. On 1 December 2006, the ongoing proceedings were discontinued as a result of the settlement reached between the parties and the interim injunction was lifted.
205 Secondly, both the settlement agreement and the licence agreement related to the disputes in question. The settlement agreement and, in particular, the non-marketing and non-challenge clauses which it contained, were limited to the scope of the patents which were the subject matter of the disputes between Servier and Krka. The licence agreement concerned the 947 patent and thus also had a direct link with those disputes.
206 Thirdly, there was, at the time the settlement and licence agreements were concluded, consistent indications capable of leading the parties to believe that the 947 patent was valid (see paragraphs 203 and 204 above).
207 Fourthly, although there were already meetings between Servier and Krka before the EPO decision of 27 July 2006 (see, inter alia, recital 837 of the contested decision), they had not resulted in an agreement (recitals 856 to 859 to the contested decision) and it was only after that decision that new negotiations began (recital 898 of the contested decision). The EPO decision of 27 July 2006 confirmed the validity of the 947 patent and was therefore, at the very least, one of the driving factors leading to the settlement and licence agreements.
208 Thus, having regard to the scope of the terms of the settlement agreement and the licence agreement and the context in which those agreements were signed, it must be held that the linking of those two agreements was justified and therefore does not constitute a strong indication of the existence of a reverse payment from Servier to Krka giving rise to the licence agreement (see paragraph 184 above).
209 In those circumstances, it is necessary to examine, in the second place, whether, in the present case, the Commission established, on the basis of indicia or evidence other than the mere linking of the licence agreement and the settlement agreement, that the licence agreement had not been concluded at arm’s length (see paragraph 185 above).
210 In this respect, it should be noted that it is common ground that, unlike the other agreements that were the subject of the contested decision, neither the settlement agreement nor the licence agreement gave rise to a financial transfer from Servier to Krka.
211 The licence agreement even provided that Krka was to pay Servier a royalty of 3% of its net sales.
212 It is true that the royalty constitutes the quid pro quo for the benefit received by the generic company under the licence agreement, namely the authorisation to use the patent in order to enter the market without risk. However, it was for the Commission to demonstrate that that quid pro quo was abnormally low and that the licence agreement thus gave rise to a reverse payment to Krka.
213 Although the Commission presented, in the contested decision, a number of factors suggesting that the licence agreement was beneficial for Krka’s commercial interests (see recitals 1738 to 1744 and, in particular, recital 1739), it did not demonstrate that the royalty rate of 3% was abnormally low, that is to say to such a degree that it could not be explained by considerations limited to the economic value of the patent to which the licence relates (see paragraph 188 above).
214 As regards the Commission’s assertion that the royalty rate was much lower than Servier’s operating profit for 2007 in the Czech Republic, Hungary and Poland, it is not necessarily abnormal that the rate of an operating surplus, which represents the gross profit derived from an activity, greatly exceeds the royalty rate of a licence agreement, which represents only the cost of the right of use of a patent.
215 The same reasoning can also be used to reject the Commission’s argument that the royalty represented a small proportion of Krka’s profit margins. A fortiori, the generic company would have no interest in concluding a licence agreement if the amount of the royalty did not enable it to generate a sufficient profit margin.
216 Finally, it is not abnormal that the royalty rate of a patent used by Krka was calculated on the basis of the sales price of Krka’s product and not on the basis of the sales price of Servier’s product.
217 All those elements, even taken together, can, at most, demonstrate that the price of the licence granted to Krka was favourable to its commercial interests, but do not suffice to establish that the transaction in question was not concluded at arm’s length, especially since the licence agreement provided that Servier could continue to market its product in the seven Member States to which the licence applied, either directly or through one of its affiliated companies or even via a single third party per Member State. The licence granted was therefore not exclusive, which limited its advantageousness to Krka since there was a risk that Krka’s product would be in competition with another generic product, whether marketed or produced by Servier or by a third party.
218 Moreover, it should be added that, in the judgment delivered today, Servier and Others v Commission (T‑691/14, paragraph 1072), it is noted that, during the hearing in that case, the Commission itself indicated that it did not dispute that the royalty was consistent with market practices.By stating in the contested decision — admittedly as a subsidiary point — that ‘it is not the low level of royalties but the fact that the sole licence was granted against a commitment not to enter or challenge Servier’s patents in a number of other, restricted markets, that is central to this analysis’ (footnote 2354), the Commission already showed that it incorrectly attached only secondary importance to the fact that the transaction might have been concluded at arm’s length.
219 It follows from the considerations set out in paragraphs 213 to 217 above that the Commission has not established that the royalty rate of 3% laid down in the licence agreement was abnormally low, that is to say to such a degree that it could not be explained by considerations limited to the economic value of the patent to which the licence relates. The Commission has therefore not established that the licence agreement does not constitute a transaction concluded at arm’s length.
220 Consequently, the Commission has not established the existence of a reverse payment resulting from the granting of a licence at an abnormally low price (see paragraph 173 above) and which, since it is not intended to compensate for the costs inherent in the settlement of a dispute (see paragraph 176 above), constitutes an inducement.
221 It follows that the Commission was not entitled to find in the present case that there was a restriction of competition revealing a sufficient degree of harm to be classified as a restriction by object.
222 That conclusion is not called into question by the other factors relied on by the Commission in the contested decision.
223 First, even if the licence agreement were inducive because it allowed, in the seven Member States — that is to say in a part of the market in respect of which the Commission did not find an infringement — the implementation of an advantageous duopoly between Servier and Krka, as the Commission indicated in the contested decision (see, inter alia, recital 1728, 1734 and 1742), that duopoly did not result from that agreement itself, but from the choices made by Servier and Krka after that agreement, namely, Servier’s choice not to grant a licence to another generic company or to sell its own generic version of perindopril at a low price (recital 1727 of the contested decision) and Krka’s choice not to adopt an aggressive pricing policy (recital 1744 of the contested decision).
224 The restriction by object found by the Commission, in particular the inducement which is one of the conditions of that restriction (see paragraph 151 above), concerns the settlement and licence agreements concluded between Servier and Krka, and not practices subsequent to those agreements and not determined by them.
225 Even if the duopoly in question could be regarded as an implementation of the settlement and licence agreements, it should be borne in mind that the Commission and the Courts of the European Union cannot, when examining whether an agreement restricts competition by object and, in particular, in assessing the economic and legal context of that agreement, completely ignore its potential effects (Opinion of Advocate General Wahl in ING Pensii, C‑172/14, EU:C:2015:272, paragraph 84). However, it is also apparent from the case-law that establishing the existence of a restriction of competition by object cannot, under the guise, inter alia, of the examination of the economic and legal context of the agreement at issue, lead to the assessment of the effects of that agreement (see, to that effect, judgment of 11 September 2014, CB v Commission, C‑67/13 P, EU:C:2014:2204, paragraphs 72 to 82), since otherwise the distinction between a restriction of competition by object and by effect laid down in Article 101(1) TFEU would lose its effectiveness. For the purposes of verifying the specific capability of an agreement to produce competition-restricting effects characteristic of agreements with an anticompetitive object, the analysis of the potential effects of an agreement must therefore be limited to those resulting from information objectively foreseeable at the time of the conclusion of that agreement (see, to that effect, judgment of 11 September 2014, CB v Commission, C‑67/13 P, EU:C:2014:2204, paragraphs 80 to 82, and the Opinion of Advocate General Wahl in ING Pensii, C‑172/14, EU:C:2015:272, paragraph 84).
226 In the present case, the alleged potential effects in question, that is to say the duopoly alleged by the Commission, are based on hypothetical circumstances which were therefore not objectively foreseeable at the time of the conclusion of the settlement and licence agreements.
227 In any event, the Commission, referring to the practice of pharmacy stock saturation and to a complaint to the Polish authorities alleging the existence of unfair competition, indicated in recital 1725 of the contested decision that ‘Servier’s attitude towards Krka in the seven licensed markets could hardly be described as one of cooperation’. Moreover, as is apparent from recital 1728 of the contested decision, the duopoly described by the Commission between Servier and Krka did not exclude a certain degree of competition between those companies.
228 Secondly, according to the Commission, the licence agreement was inducive, in this case, because it enabled Krka to enter certain markets without risk in return for its exclusion from other markets. From that perspective, where the scope of the non-marketing or non-challenge clauses is wider than that of the licence agreement and there is therefore a gap or an ‘asymmetry’ between those two agreements, according to the Commission’s wording in recitals 1706 and 1736 of the contested decision, it is then possible to conclude that there is an inducement, since the licence agreement, by allowing the generic company to enter certain parts of the market without risk, is actually intended to induce that company to agree to withdraw from other parts of the market, to the originator company’s advantage.
229 Those arguments cannot be accepted.
230 First of all, the approach put forward by the Commission, whereby the mere conclusion, even on normal market conditions, of a licence agreement linked to a settlement agreement containing restrictive clauses could constitute an inducement, would lead to a paradoxical outcome because, in that case, the wider the scope of a licence agreement, the greater the inducement and thus the easier it would be to find a restriction by object, unless the scope of the licence agreement were exactly identical to that of the settlement agreement.
231 The wider the scope of a licence agreement, especially in relation to the scope of the settlement agreement to which it is linked, the more that agreement is procompetitive, in view of the procompetitive effects of the licence, which encourages the market entry of a generic company and limits the competition-restricting nature of the non-marketing and non-challenge clauses in the settlement agreement (see paragraphs 190 and 191 above).
232 In that respect, it may be noted that in his Opinion in CB v Commission (C‑67/13 P, EU:C:2014:1958, point 55), Advocate General Wahl stated that the formalist approach to identifying a restriction by object was conceivable only in the case of conduct in respect of which it could be concluded that the unfavourable effects on competition outweighed the procompetitive effects.
233 In addition, the Commission’s argument, which leads to the patent holder being obliged to conclude a licence agreement covering the entire territory to which the restrictive clauses in the settlement agreement apply, does not respect the intellectual property rights of the patent holder and, in particular, his margin of discretion as regards the grant of licences (see, for a case where the patent holder is in a dominant position, judgment of 17 September 2007, Microsoft v Commission, T‑201/04, EU:T:2007:289, paragraph 331). That argument also disregards the margin of discretion that the parties to a dispute must have in order to reach a settlement in good faith.
234 Moreover, the conclusion of an ‘asymmetric’ licence agreement does not necessarily constitute — for a generic company that does not recognise the validity of the patent in question — a sufficient benefit that it would agree to the non-marketing and non-challenge clauses. For the benefit arising from such an agreement to be regarded as an inducement, it would have to offer that company compensation for the certain loss of expected profits resulting from the acceptance of a settlement with clauses prohibiting entry on certain geographic parts of the market. For a company that does not seriously believe that the patent is valid and which is able to enter the entire market covered by the non-marketing and non-challenge clauses, a licence with a geographic scope more limited than the scope of those clauses does not constitute an economically satisfactory outcome that could lead them to accept those clauses. It is true that the licence partially opens the way for that company to the market covered by the patent by offering it the possibility of obtaining the envisaged profits on that part of the market, but, if it is not established that the royalty rate of that licence is abnormally low in respect of that part of the market, that licence does not give that company any compensation for the other parts of the market, on which it could make a profit if the patent were annulled, and which it is now prevented from accessing.
235 In the present case, Krka’s expected earnings in the 18 to 20 markets to which the licence agreement did not apply were far from negligible. The Commission indicates in the contested decision that the expected earnings from Western European markets roughly matched those from the three largest of the seven markets (footnote 2348). Although it must be taken into account that the licence eliminates any risk of further infringement proceedings and that the profits that Krka could obtain through the licence agreement were therefore more certain, the importance that it could attach to such a risk depended to a large extent on its degree of conviction as to the validity of the patent. The fact that Krka recognised the validity of the 947 patent was therefore a decisive factor in its decision to choose a limited — but licence-protected — entry to the seven markets rather than a wider entry to all of the Member States’ markets subject to a significant risk of infringement because of the validity of that patent from Krka’s perspective.
236 Thirdly, with regard to the other elements that are supposed to establish the inducive nature of the licence agreement for Krka, it should be noted first of all that the fact that the latter estimated the opportunity cost of not entering into the agreement at more than EUR 10 million of ‘lost profits’ in three years (recital 1738 of the contested decision) is rather an additional indication of the fact that it considered that the 947 patent was valid. The profits in question corresponded to those expected if it entered or stayed on the seven markets. Thus, Krka seems to have considered that in the absence of an agreement with Servier, it was unlikely, if not impossible, that it would enter those markets at risk or stay on those markets, which confirms the fact that it acknowledged the validity of the 947 patent.
237 Next, although it is apparent from recital 1740 of the contested decision, which refers to recital 913 thereof, that the 18 to 20 markets were ‘traditionally less important for Krka’, the expected profits on those markets were far from negligible (see paragraph 235 above).
238 Thus, the elements set out in paragraphs 236 and 237 above do not establish that the licence agreement was an inducement for Krka.
239 Fourthly, the Commission’s finding that the settlement and licence agreements constituted market sharing between Servier and Krka (see the title of section 5.5.3 of the contested decision and, inter alia, recital 1745 thereof) is unfounded.
240 As regards the seven markets covered by the licence agreement, although the Commission did not find an infringement in respect of that part of the internal market, it nevertheless took account of the conduct of Servier and Krka on those seven markets, including the conclusion of the licence agreement, which the Commission classified as an inducement, in order to establish the existence of market sharing based on a distinction between the 18 to 20 Member States, on the one hand, and the seven Member States, on the other.
241 However, Servier was not excluded from the markets of the seven Member States where Krka and it were in competition (see paragraph 227 above).
242 Thus, there was no part of the market which, under the settlement and licence agreements, was reserved for Krka. It therefore cannot be concluded that there was market sharing — in the sense of a hermetic division between the parties to the agreement — of that part of the internal market.
243 Furthermore, it should be noted that, in those seven Member States, the licence agreement contributed to the entry or continued presence on the market of a generic competitor of the originator company. It therefore had a positive effect on competition by comparison with the previous situation in which the generic companies could only enter or remain on the market at risk, since the validity of the main patent in question — the 947 patent — had just been confirmed by the competent authorities (see paragraph 206 above) and there was a risk, which Krka perceived as a serious risk, that its product was infringing.
244 It should be added that the fact that, at the time the settlement and licence agreements were concluded, the national equivalents of the 947 patent had not yet been granted to Server in some of the seven markets, whereas Krka was already selling its product (recital 1755 of the contested decision), does not support the conclusion that the licence agreement had no positive effect on competition. Although it is true that Krka could already have entered the markets before the licence agreement without facing an immediate risk of infringement proceedings and although, consequently, the licence did not play a decisive role in relation to Krka’s entry of the markets in question, it nevertheless allowed Krka to remain on those markets without the risk of facing such a challenge.
245 The licence agreement’s positive effect on competition noted in paragraphs 243 and 244 above supports the conclusion that there was no market sharing as regards the seven Member States.
246 The licence agreement’s positive effect on competition is further confirmed by an extract from Krka’s reply to a request for information which appears in recital 913 of the contested decision. That extract states, inter alia, as follows:
‘Getting a license and withdrawing oppositions was considered as the best option for Krka at that time — to be able to sell perindopril on Krka’s key markets in [Central and Eastern Europe] immediately, it means in 2006.
According to all other scenarios, a launch was not possible earlier than in at least 2 years after July 2006, and even after such period a launch was not warranted (risk that 947 is maintained, development risks for non-alpha).’
247 The extract cited in paragraph 246 above supports the conclusion that Krka considered that, without a licence agreement, it would be impossible to enter or remain on the market in the seven Member States because of the 947 patent (see paragraphs 235 and 236 above).
248 As regards the 18 to 20 markets, that is to say the only part of the market in respect of which the Commission found an infringement, it should be noted that, since it has not been shown that there was an inducement (see paragraph 220 above), the non-marketing and non-challenge clauses must be regarded as arising from a legitimate patent dispute settlement agreement which is linked to a licence agreement (see paragraph 199 above). Such a contractual framework, based on the recognition of the validity of the patent, cannot, therefore, be classified as a market exclusion agreement.
249 Accordingly, no part of the market was unlawfully reserved for Servier.
250 The market sharing on which the Commission also based its finding of a restriction by object is therefore not established.
251 Fifthly, the Commission failed to demonstrate that Servier or Krka had intended to conclude a market sharing or market exclusion agreement, that Servier had intended to induce Krka not to compete or that Krka had intended to agree, in exchange for an inducive benefit, not to exert competitive pressure on Servier.
252 As a preliminary point, it should be borne in mind that it is normal for the activities which anticompetitive practices and agreements entail to take place in a clandestine fashion, for meetings to be held in secret, and for the associated documentation to be reduced to a minimum. It follows that, even if the Commission discovers evidence explicitly showing unlawful contact between traders, it will normally be only fragmentary and sparse, so that it is often necessary to reconstitute certain details by deduction (judgment of 25 January 2007, Sumitomo Metal Industries and Nippon Steel v Commission, C‑403/04 P and C‑405/04 P, EU:C:2007:52, paragraph 51). It must be noted, however, that the agreements at issue in the present case are genuine contracts which, moreover, were well publicised (recital 915 of the contested decision). Since the Commission could easily obtain the full content of the agreements at issue, the applicability of the case-law which has just been cited is less evident. Thus, inferences drawn from partial extracts of e-mails or other documents purporting to establish the intentions of the parties cannot easily call into question a finding based on the actual content of the agreements, that is to say on the legally binding relationship which the parties have decided to establish between themselves.
253 It should also be noted that, in the present case, documents which postdate the EPO decision of 27 July 2006, or the interim injunction of 3 October 2006 delivered in the United Kingdom against Krka, are best able to shed light on the intentions of the parties when they concluded the settlement and licence agreements. Those two events substantially altered the context in which the agreements were concluded, in particular as regards the perception that Krka, as well as Server, could have of the validity of the 947 patent.
254 As regards Krka, the documents on which the Commission relies in order to determine that company’s intentions (see, inter alia, recitals 849 to 854 and 1758 to 1760 of the contested decision, as well as the recitals to which they refer) concern periods before those events.
255 The extracts cited are, in any event, too fragmentary or ambiguous to establish — contrary to what the Court has repeatedly noted (see, inter alia, paragraphs 235, 236 and 247 above) — that Krka did not recognise the validity of the 947 patent and, a fortiori, that, at the time the settlement and licence agreements were signed, it intended to conclude market sharing or market exclusion agreements.
256 As regards Servier, the only extract from a document — which postdates the two events mentioned above — purportedly showing its anticompetitive intentions and which is referred to in the section of the contested decision devoted to those intentions (recitals 1761 and 1762), is the following: ‘4 years gained = great success’.
257 That extract appears in the record of a meeting of the top management of Servier, which refers to the judgment of 6 July 2007 of the High Court of Justice (England and Wales), Chancery Division (Patents Court), according to which the 947 patent was invalid for lack of novelty and inventive step of that patent in relation to the 341 patent.
258 Even assuming that it could be inferred from that extract that Servier’s management had considered, following that judgment, that the interest of the 947 patent lay in allowing it to gain an additional four years of protection, it cannot be concluded from this that, on 27 October 2006, when the settlement and licence agreements were concluded, Servier intended to conclude market sharing or exclusion agreements nor, a fortiori, can it be concluded that the settlement and licence agreements were restrictive of competition by object.
259 Furthermore, the observation made by another generic company, according to which ‘it would seem the rationale for this settlement from Servier’s view is that it protects the key markets where high level substitution and/or [international nonproprietary name] prescribing is prevalent’ (recital 1730 of the contested decision), cannot, even taken into account with all the other evidence relied on by the Commission, establish the existence of an intention on Servier’s part to adopt market sharing or market exclusion agreements with Krka.
260 Lastly, the Commission’s repeated references in the contested decision to a document entitled ‘Coversyl: defense against generics’ are not convincing. That document predates the EPO decision of 27 July 2006 and the interim injunction of 3 October 2006 delivered in the United Kingdom against Krka, which considerably limits its relevance (see paragraph 253 above). Moreover, it is apparent from the contested decision itself that that document does not elaborate a strategy concerning Krka, but, at most, that it follows from ‘the nature and structure of the document’ and from ‘the context in which reference is made to Krka’ that a defence was ‘considered’ against it (footnote 2386). Lastly, it is not apparent from the extracts from that document cited in the contested decision that Servier expressed doubts as to the validity of the 947 patent.
261 In any event, for the purposes of casting doubt on the conclusion reached by the Court in paragraph 221 above and establishing that the aim of the agreements at issue was — contrary to the conclusion implied by an analysis of their content and of the context in which they were concluded — the buying off of a competitor in order to exclude it from the market, it would fall to the Commission, in light of the considerations set out in paragraph 252 above, to produce a body of relevant and consistent evidence. The Commission has not been able to produce such evidence.
262 Sixthly, the fact that Krka continued to challenge Servier’s patents and to market its product even though the validity of the 947 patent had been upheld by the EPO decision of 27 July 2006 is not a decisive factor for the purpose of establishing the existence of a restriction of competition by object, since the fact that Krka continued to exert competitive pressure on Servier can be explained by Krka’s desire, despite the expected litigation risks, to strengthen its position in the negotiations that it was likely to have with Servier with a view to reaching a settlement.
263 In addition, continuing to challenge Servier’s patent did not cause Krka to run any further risks in terms of infringement. It merely increased its litigation costs. As for the fact that it continued to market its product, it limited itself to five Central and Eastern European markets, and the Commission indicated, in the contested decision, that Krka had ‘eventually ceased to consider entering at risk in the UK, France and other Western European markets in the aftermath of the Opposition [Division’s] Decision’ (recital 1693). In addition, in five of the seven markets covered by the licence, the equivalents of the 947 patent had not yet been granted (recital 1755 of the contested decision). Thus, the risks incurred by Krka, in at least some of the markets in which it remained, were limited.
264 Having regard to the considerations set out in the two preceding paragraphs, the fact that Krka continued to challenge Servier’s patents and to market its product even though the validity of the 947 patent had been upheld by the EPO Opposition Division does not — contrary to the Commission’s submissions — support the conclusion that the EPO decision of 27 July 2006 did not have a decisive impact on Krka’s perception of the 947 patent and, consequently, on its subsequent choice to agree to settle with Servier.
265 Seventhly, although the Commission has adduced some evidence showing that the settlement and licence agreements were the subject of commercial negotiations between Servier and Krka, with Krka seeing to maximise the advantages that it could gain from those agreements and even making the licence agreement a condition of its acceptance of the non-marketing and non-challenge clauses (see, inter alia, recitals 913 and 1746 to 1748 of the contested decision), that evidence, even taken together with all of the other evidence relied on by the Commission, does not establish that the licence agreement was not a transaction concluded at arm’s length, that is to say that the royalty rate of 3% stipulated in the licence agreement was not chosen on the basis of commercial considerations, but rather in order to induce Krka to accept the non-marketing and non-challenge clauses in the settlement agreement.
266 In addition, it should be borne in mind that the conclusion of a licence agreement, which, for any licensee, makes sense only if the licence is actually used, is based on the parties’ recognition of the validity of the patent (see paragraph 183 above). Thus, the fact that the generic seeks to obtain the licence agreement which is most favourable to his commercial interests is not sufficient to show that that company did not conclude the agreement in question on the basis of its recognition of the validity of the patent.
267 It should be added that an agreement benefiting Krka enabled it to enter the parts of the market where its position was strongest and where it could rapidly market or continue to market its product, which is beneficial to competition. Thus, the interests of a generic company such as Krka, which seeks to obtain from the originator company the licence that is most beneficial to its commercial interests, converge with those of the consumer since, due to the licence agreement, a generic company will rapidly enter the market or remain there.
268 It follows from all the foregoing that the conclusion set out in paragraph 221 above must be affirmed, since the settlement and licence agreements at issue do not reveal a sufficient degree of harm to competition that the Commission could validly conclude that they constituted a restriction by object. The plea is therefore well founded.
B. Fourth plea in law, alleging that there is no restriction of competition by object as regards the assignment agreement
1. Arguments of the parties
269 The applicant submits, inter alia, that the assignment of its patent applications did not restrict competition given that they were applications relating to the alpha form of perindopril, which could not, in any event, be marketed without infringing the 947 patent. Consequently, generic companies could not be expected to show any interest in acquiring that technology. Thus, the assignment of those applications to Servier was, according to the applicant, the only economically-viable solution.
270 The applicant also submits that the Commission’s claim that the technology it had assigned would have been essential to any generic company was misleading.
271 Furthermore, the applicant disputes the factors on which the Commission relied in order to conclude that the settlement, licence and assignment agreements were part of a single and continuous infringement seeking to restrict competition through the sharing of perindopril markets in the European Union: the short period of time between the signing of those agreements, the identity of the signatories, the common objective and the similar method of restriction of competition.
272 The applicant adds that no link can be established between the payment of EUR 30 million provided for in the assignment agreement and the settlement agreement. It states that there was no contractual guarantee concerning such a payment in the settlement agreement.
273 The applicant disputes that any restriction of competition arose from the assignment agreement. It asks how long it should have waited before concluding that assignment agreement. It points out that it allocated adequate funds to develop a non-alpha form of perindopril and that it continued to be a competitor to Servier also as regards the alpha form of perindopril because of the licence agreement included in the settlement agreement. It also states that, during the negotiations prior to the signing of the settlement agreement, it sought to obtain a licence with the widest possible territorial scope. The applicant concludes that the assignment agreement did not restrict competition.
274 The applicant submits that, by assigning patent applications to Servier, it merely responded to an offer made by Servier and negotiated the sale of those applications to the latter. It adds that, by the contested decision, the Commission intervened in the sensitive field of private property and the right to dispose of such property.
275 It submits, lastly, that the Commission failed to distinguish its claims against Servier under Article 102 TFEU from the infringement relating to the assignment agreement. The applicant points out, in that regard, that the Commission does not dispute the applicant’s right to dispose of its intellectual property.
276 The Commission submits that the plea must be rejected.
277 It notes, inter alia, that the fact that other generic manufacturers did not approach Krka to acquire its technology between October 2006 and January 2007 cannot be decisive.
278 The Commission also submits that the settlement agreement and the assignment agreement constituted a single and continuous infringement, namely a restriction of competition by object, concerning the sharing of the perindopril market in the European Union.
2. Findings of the Court
279 It is necessary, at the outset, to note the decisive grounds on which the Commission relied, in the contested decision, in order to reach the conclusion that the assignment agreement could be characterised as a restriction of competition by object.
280 The Commission first of all found that, under the assignment agreement, Krka had assigned two patent applications to Servier, one concerning a process for the synthesis of perindopril (WO 2005 113500) and the other concerning the preparation of perindopril formulations (WO 2005 094793), and that the technology covered by those patent applications was used for the production of Krka’s perindopril (recital 1770 of the contested decision).
281 On the basis of that finding, the Commission sought to demonstrate that the assignment agreement reinforced the competitive position of Servier and Krka which arose from the market sharing that had been established, according to the Commission, by the settlement and licence agreements (recitals 1766 and 1804 of the contested decision).
282 As regards, in the first place, Servier, the Commission observed that the transfer of Krka’s technology had taken place in specific market conditions in which there were very few alternative sources of potentially viable API technology independent of Servier (recitals 1766 and 1772 to the contested decision). According to the Commission, Krka’s technology, with which the European Pharmacopoeia requirements could be satisfied (recitals 1766, 1770 and 1793 of the contested decision), constituted ‘a “key” to enter the market’.
283 The Commission indicated the following in recital 1772 of the contested decision:
‘By removing Krka’s ability to freely license out or assign its technology to third parties, i.e. other generics, Servier effectively foreclosed the potential avenue of competition based on the use of Krka’s technology by third parties. Such technology could, for example, serve as a platform for new patent challenges. In combination with the Krka Settlement Agreement, ALA thus provided Servier with absolute protection from any remaining potential competition stemming from Krka’s technology.’
284 Thus, according to the Commission, by acquiring Krka’s technology, Servier was certain that Krka could no longer assign a technology which could have been useful to other generic companies. The Commission therefore concluded that the assignment agreement allowed Servier to reinforce the protection that it already enjoyed because of the non-marketing and non-challenge clauses in the settlement agreement (recitals 1805 and 1806 of the contested decision).
285 As regards, in the second place, Krka, the Commission considered not only that the latter ‘was aware that acquisitions of perindopril technology by Servier could lead to foreclosure of generic competitors’ (recital 1800 of the contested decision), but above all that it benefited from the licence that was granted back to it under the assignment agreement.
286 As regards that latter aspect, the Commission indicated that Krka could continue to use its technology on the markets of the seven Member States in which it was able to market its product under the licence agreement (recital 1806 of the contested decision). According to the Commission, Krka’s technology was useful, including for Krka, for the purpose of producing perindopril API with a purity level meeting the requirements of the European Pharmacopoeia. The favourable position that Krka already enjoyed on the seven markets due to the licence agreement was therefore maintained by the assignment agreement.
287 The Commission concluded that the purpose of the assignment agreement was to reinforce the market sharing put in place by the settlement and licence agreements (recitals 1803 and 1810 of the contested decision).
288 The Commission added that conclusion of the settlement and licence agreements and of the assignment agreement formed part of a single and continuous infringement restricting competition by sharing markets for perindopril in the European Union. The Commission relied, in that respect, inter alia, on the fact that those agreements pursued the same objective of market sharing between Servier and Krka (recital 1811 of the contested decision).
289 The Commission concluded the section of the contested decision devoted to the analysis of the restriction by object in relation to the various agreements between Servier and Krka by indicating that those agreements ‘followed [the] objective to share markets by preventing or limiting generic competition between, or to, Krka and Servier’ (recitals 1812 of the contested decision).
290 Lastly, it should be noted that the Commission considered that the assignment agreement introduced only an ‘additional’ distortion, as indicated by the heading of section 5.5.3.4 of the contested decision.
291 It follows from the foregoing considerations that the Commission’s finding of a restriction by object with regard to the assignment agreement is based on the previous finding of market sharing as a result of the settlement and licence agreements.
292 However, as indicated above (see paragraph 250 above), that finding is incorrect.
293 Consequently, the Commission’s finding of a restriction by object with regard to the assignment agreement must also be held to be invalid.
294 It should be added that the assignment agreement is not a ‘side deal’ to the settlement agreement, within the meaning of the considerations set out in paragraphs 164 et seq. above.
295 That assignment agreement was not concluded on the same day as the settlement agreement, there is no contractual link between the two agreements and the Commission has not established that they were indissociable (see paragraph 165 above).
296 The Commission even stated that there was no link between, on the one hand, the EUR 30 million payment by Servier to Krka under the assignment agreement and, on the other hand, the settlement agreement, in the sense that that payment did not constitute an inducement for Krka to accept the non-marketing and non-challenge clauses in the settlement agreement. That is apparent, inter alia, from the following extracts from the contested decision:
‘(1678) Two months later, Servier purchased from Krka patent applications for competing technologies to produce perindopril for EUR 30 million. Krka considered that Servier feared that this technology could otherwise be assigned or licensed to other competitors. While some elements point in the direction of the existence of a link between the Settlement Agreement and the payment of [EUR] 30 million by Servier, this decision does not draw any conclusion on this point, and the analysis of those agreements is not based on the existence of such a link.
…
(footnote 2419) Servier contests that there was a link between the payment for patent applications and the settlement agreement. (Servier’s reply to the Statement of Objections, paragraph 1084, ID 10114, p. 363). As it evidently flows from section 5.5.3.3.3, the assessment of the Krka Settlement Agreement does not consider the payment of EUR 30 million as an inducement for Krka to accept the restrictive settlement terms, and leaves open as undecided the question whether there was a link between the settlement agreement and the [assignment and licence agreement].’
297 Thus, the assignment agreement cannot compensate for the fact that the inducement which, according to the Commission, arises from the licence agreement and allowed it to conclude that the settlement agreement was actually intended to exclude one of Servier’s competitors is not established (see paragraph 220 above).
298 It follows from all of the foregoing that the Commission erred in finding, as regards the assignment agreement, a restriction of competition by object. The present plea is therefore also well founded.
C. Fifth plea in law, alleging that there is no restriction of competition by effect
1. Arguments of the parties
299 By this plea, the applicant disputes the assessment made by the Commission, in recitals 1813 et seq. of the contested decision, by which it considered that the agreements concluded with Servier constituted a restriction of competition by effect.
300 The applicant submits, first of all, that, in the light of the applicable case-law, in order to conclude that such a restriction exists, it is not sufficient to find that the agreements at issue were likely to have restrictive effects; it must also be established that competition has in fact been restricted. The applicant adds that the Commission has not proved this and thus has not satisfied the condition laid down by the case-law. Lastly, it states that the counterfactual analysis adopted by the Commission is ‘totally unrealistic’.
301 The applicant notes that, after the EPO decision of 27 July 2006, there were three options available to it, namely, to market its product despite that decision, not to market its product, or, lastly, to try to settle its disputes with Servier.
302 In that regard, it disputes the Commission’s assessment in recital 1826 of the contested decision, according to which, after the EPO decision, Krka continued to endeavour to enter western European markets.
303 The applicant explains that, since the EPO decision of 27 July 2006, it worked towards settling its disputes with Servier. In that context, the fact that it continued to have a presence on the five markets, which it had already entered previously, namely the Czech Republic, Hungary, Lithuania, Poland and Slovenia (recital 1681 of the contested decision), pending the outcome of the negotiations, cannot establish that it continued to endeavour to enter other markets.
304 The applicant adds that continuing with its litigation was merely defensive, since it was simply defending itself in proceedings brought by Servier and strengthening its powers of negotiation with the latter.
305 Furthermore, the applicant disputes the Commission’s conclusion in recitals 1827 to 1830 of the contested decision, according to which there was a significant likelihood that, in the absence of an agreement, Krka would have continued to challenge the validity of Servier’s patents and market its product ‘at risk’, that it would have sold its technology to third parties, and that its efforts to enter the market would have been successful.
306 In that regard, the applicant states that, after the EPO decision of 27 July 2006, it no longer believed that the 947 patent was invalid and that, therefore, it was not the non-challenge clause set out in the settlement agreement that prevented it from challenging that patent, since it would not have challenged it in any event, even in the absence of an agreement. At the hearing, the applicant also added that other parties, including Apotex, had pursued the litigation against the 947 patent and that, as a result, Krka’s withdrawal from the proceedings in which it participated had had no effect.
307 Similarly, the applicant disputes that it could have launched at risk in the markets other than the five in which it was already marketing its product. It took the view that launching at risk following the EPO decision of 27 July 2006 would have entailed unacceptable commercial risks that it was not prepared to bear. In that regard, it adds that it had no interest in entering all the EU markets. It also indicates that the Commission disregarded the dissuasive effect of the EPO decision of 27 July 2006.
308 It also disputes that it could have assigned its technology to third parties, since they were not interested.
309 Furthermore, the applicant states that, after an ex post analysis, it should be concluded that it could have entered the market only as from May 2009, that is to say after the decision of the EPO’s Board of Appeal declaring the 947 patent invalid (see paragraph 9 above).
310 The Commission submits that the EPO decision of 27 July 2006 did not prevent Krka from pursuing its litigation relating to Servier’s patents.
311 It relies on the fact that Krka successfully opposed an application for an interim injunction brought against it in Hungary prior to the settlement agreement.
312 It also states that the applicant’s product only potentially infringed a valid patent.
313 The Commission submits that Krka would have continued to be a competitive force in the seven markets concerned by the licence agreement. It states, in that regard, that Krka intended to launch its perindopril in Slovakia. It also states that Latvia and Slovakia belonged to Krka’s key markets.
314 Lastly, the Commission accepts that it is for it to establish, with a reasonable degree of probability, that the agreements between Servier and Krka had an appreciable negative effect on actual or potential competition within the European Union.
2. Findings of the Court
315 The Court has repeatedly held that, in order to determine whether an agreement is to be considered to be prohibited by reason of the distortion of competition which is its effect, the competition in question should be assessed within the actual context in which it would occur in the absence of the agreement in dispute (see judgments of 30 June 1966, LTM, 56/65, EU:C:1966:38, p. 250, and of 6 April 2006, General Motors v Commission, C‑551/03 P, EU:C:2006:229, paragraph 72; see, also, judgment of 11 September 2014, MasterCard and Others v Commission, C‑382/12 P, EU:C:2014:2201, paragraph 161 and the case-law cited). It is therefore necessary to show — by a comparison between the competition that existed when the agreement was in force and the competition that would have occurred if that agreement had not been concluded — that the competitive situation was worse when that agreement was in force.
316 As a preliminary point, it is necessary to clarify the approach adopted by the Commission, in the contested decision, in examining the restriction of competition by effect with regard, in particular, to the comparative step of that examination mentioned in paragraph 315 above.
(a) The approach adopted by the Commission
317 It is necessary, first of all, to note some of the general considerations, applicable to all of the agreements between Servier and the generic companies at issue in the contested decision, set out by the Commission in Section 5.1.7 of the contested decision, entitled ‘Assessment of patent settlement agreements with reverse payments as restrictions by effect pursuant to Article 101(1) [TFEU]’.
318 The Commission indicated, inter alia, that the examination of conditions of competition on a given market ‘must be based not only on existing competition between the undertakings already present on the relevant market but also on potential competition’ (recital 1215 of the contested decision).
319 The Commission noted, in recital 1219 of the contested decision, that, according to the Guidelines on the application of Article [101(3) TFEU] (OJ 2004, C 101, p. 97), both the ‘actual and potential effects’ of an agreement were to be taken into account, since the agreement only had to have ‘likely anti-competitive effects’. It referred, in that regard, to paragraph 24 of those guidelines, which is based on the judgment of 28 May 1998, Deere v Commission (C‑7/95 P, EU:C:1998:256, paragraph 77).
320 The Commission then set out its method. It indicated that it would show the restrictive effects of the agreements between Servier and the generic companies at issue in the contested decision by establishing, as a first step, that each of them had entailed the removal of a potential competitor, and then, as a second step, that the elimination of a single potential competitor was ‘likely to have effects on the competitive structure’ (recital 1219 of the contested decision).
321 The Commission therefore considered that the finding of the elimination of a potential competitor allowed it to establish only anticompetitive effects that were ‘likely’ to occur, that is to say ‘potential’ effects on competition (see paragraph 319 above).
322 The Commission stated the following in recital 1220 of the contested decision:
‘The assessment of restrictive effects should be carried out based on the facts at the time of the settlement, while also taking into account how the agreement was actually implemented. Some of the parties disagree and claim that the assessment should take into account all posterior factual developments, and not be based primarily on the situation at the time the agreements were concluded. ... when elimination of potential competition is at issue, looking at what actually happened may have little to do with what would likely have happened absent the agreement, a core question for the competitive assessment. This is all the more so where the agreement significantly changes the incentives of one party, or both, to continue to compete.’
323 In the rather ambiguously worded first two sentences of recital 1220 of the contested decision, the Commission acknowledged that it would not rely, for each agreement, on all the factual developments subsequent to its conclusion; rather, it would rely, at least primarily, on the facts at the time it was concluded. In order to justify that approach, the Commission then referred to the concept of ‘potential competition’, stating that, where the elimination of potential competition was at issue, taking into account certain actual events, in particular events subsequent to the conclusion of the agreement, is less relevant in order to show one side of the comparison referred to in paragraph 315 above, namely competition as it would have occurred absent an agreement.
324 That approach is confirmed by an extract from recital 1264 of the contested decision, in which the Commission considers that, where the elimination of a potential competitor is at issue, it is necessary to analyse the ‘potential future effects’ of the agreements.
325 Recital 1264 of the contested decision is contained in the section of the contested decision entitled ‘Prevailing market structure at the time of the settlement agreements’, which is primarily concerned with describing the gradual elimination of potential competitors to Servier by the conclusion of the various agreements examined by the Commission (recitals 1244 to 1269 of the contested decision).
326 It is true that, in the section of the contested decision entitled ‘Prevailing market structure at the time of the settlement agreements’, the Commission refers to certain events which actually occurred during the implementation of the agreements and which support the conclusion that the two companies that did not enter into an agreement with Servier continued to exert competitive pressure. The Commission thus notes that the 947 patent was invalidated in the United Kingdom as a result of the litigation pursued by one of those two companies, Apotex, in that country.
327 However, the Commission states that there was still a strong ‘possibility’, after the conclusion of the agreements that Servier entered into with various generic companies, that it would again try to reach an agreement with Apotex, and with the other company representing a potential threat to it (recital 1268 of the contested decision), even though the Commission could have observed, when it adopted the contested decision, that no such agreements had been concluded.
328 The Commission’s assertion referred to in paragraph 327 above confirms that, in order to show the competition that would have occurred had an agreement not been concluded (one side of the comparison mentioned in paragraph 315 above), it relied on a hypothetical approach, which was partly indifferent to the actual events that took place, in particular after the conclusion of the agreements.
329 The Commission thus relied on the premiss that it could, in the event of an agreement eliminating a potential competitor, show merely the potential effects of that agreement, that is to say those that the agreement is ‘likely’ to have, which allows it to base its description of competition as it would have occurred absent an agreement on hypotheses or ‘possibilities’ rather than on the events that actually took place, which it could have observed when it adopted its decision.
330 It follows from the foregoing that the Commission considered that, where it had established that an agreement excluded a potential competitor, it was not necessary, in order to determine the competition that would have occurred had that agreement not been concluded, to rely on the actual events that occurred, in particular after the conclusion of the agreement. Rather, the Commission considered — relying on its usual practice in taking into account the potential effects of an agreement, according to which it suffices to demonstrate that that agreement is ‘likely’ to have anticompetitive effects (see paragraphs 319 and 324 above) — that it could base its description of the competition that would have occurred had an agreement not been concluded on hypotheses or ‘possibilities’.
331 The Commission’s general approach having been set out, it must be determined whether, in the particular context of its analysis of the effects of the agreements between Servier and Krka on competition, the Commission took an approach consistent with that general approach.
332 In recitals 1813 and 1814 of the contested decision, that is to say the first recitals of the section devoted to the restriction by effect with regard to the agreements concluded between Servier and Krka, the Commission indicated that the purpose of that section was to determine whether those agreements ‘were likely to entail restrictive effects on competition’. Similarly, in the title of the conclusion of the section devoted to the restriction by effect with regard to the agreements concluded between Servier and Krka, the Commission indicated that those agreements ‘were likely to entail restrictive effects for competition’.
333 It follows from the wording used by the Commission in paragraph 332 above that its approach is based on the finding of potential effects of the agreements at issue (see paragraph 319 above).
334 Moreover, in order to carry out the comparison mentioned in paragraph 315 above, the Commission relied on the fact that, had an agreement not been concluded, Krka would have remained a ‘competitive threat’ to Servier (recitals 1828 and 1830 of the contested decision).
335 On the face of it, that ‘competitive threat’ allegedly eliminated by the agreements at issue relates, given its hypothetical nature, more to potential effects on competition than actual effects.
336 The elimination of the ‘competitive threat’ mentioned in paragraphs 334 and 335 above constitutes a key element in the Commission’s demonstration intended to establish that the competitive situation on the market became worse because of the settlement (see paragraph 315 above).
337 It is true that the Commission subsequently devotes — in relation to Servier’s market power which it previously examined (recitals 1817 to 1819 of the contested decision) — a section of the contested decision to the structure of the relevant market, characterised by a lack or shortage of sources of competition (recitals 1835 to 1846 of the contested decision).
338 However, it is the prior finding of the existence, absent an agreement, of a ‘competitive threat’, made in the foregoing section of the contested decision (recitals 1825 to 1834), which constitutes the starting point for the analysis of the market structure.
339 The Commission concludes its analysis of the structure of the market in question by indicating that there was a strong possibility that the remaining sources of competition at the time the agreements between Servier and Krka were concluded would be removed from competition by a future agreement or otherwise, but it does not specify whether that actually occurred during the period when those agreements were in force (recital 1846 of the contested decision).
340 The element mentioned in paragraph 339 above confirms the findings set out in paragraph 330 above. Thus, the Commission considered that, since it had established that the settlement agreement excluded Krka and that Krka was at least a potential competitor to Servier, it was not required, in order to demonstrate the competition that would have occurred had an agreement not been concluded (one side of the comparison, mentioned in paragraph 315 above), to take into account the events that actually occurred, which it could have observed at the time it adopted its decision. Rather, the Commission considered — relying on its usual practice in taking into account the potential effects of an agreement, according to which it suffices to demonstrate that that agreement is ‘likely’ to have anticompetitive effects — that it could base its description of the competition that would have occurred had an agreement not been concluded on hypotheses or ‘possibilities’.
341 The Commission’s analysis of the agreements concluded between Servier and Krka was therefore consistent with the general approach it had set out for examining the various settlement agreements found to be infringing in the contested decision.
342 Having noted the Commission’s approach to the comparative step of the examination of the restriction by effect mentioned in paragraph 315 above, it is necessary to determine whether the Commission was entitled to find that the agreements concluded between Servier and Krka restricted competition by effect.
343 That examination requires, as a preliminary point, a review of the relevant case-law.
344 In particular, in view of the approach taken by the Commission and the key role played in its reasoning by the multiple references to the ‘potential effects’ of the agreements and to the fact that they were ‘likely to entail restrictive effects on competition’, it is necessary to review the case-law, already referred to in paragraph 319 above, according to which account should be taken of the potential effects of an agreement, a concerted practice or a decision by an association of undertakings for the purpose of determining whether such measures fall within the scope of Article 101(1) TFEU.
(b) The relevant case-law in the present case
345 Whilst the Court of Justice, in the context of references for a preliminary ruling, has often reiterated the principle that Article 101(1) TFEU does not restrict the assessment of an agreement or practice to actual effects alone, since that assessment must also take account of the agreement’s potential effects on competition within the internal market (judgments of 21 January 1999, Bagnasco and Others, C‑215/96 and C‑216/96, EU:C:1999:12, paragraph 34; of 23 November 2006, Asnef-Equifax and Administración del Estado, C‑238/05, EU:C:2006:734, paragraph 50; of 28 February 2013, Ordem dos Técnicos Oficiais de Contas, C‑1/12, EU:C:2013:127, paragraph 71, and of 26 November 2015, Maxima Latvija, C‑345/14, EU:C:2015:784, paragraph 30), it has rarely had the opportunity to examine itself whether a practice or agreement has potential effects such that a restriction of competition may be found.
346 The Court of Justice first examined the taking into account of the potential effects of an agreement in the judgment of 17 November 1987, British American Tobacco and Reynolds Industries v Commission (142/84 and 156/84, EU:C:1987:490). In the case that gave rise to that judgment, the Commission had rejected a complaint and found that the agreements to which that complaint related did not constitute an infringement of the Treaty rules on competition (judgment of 17 November 1987, British American Tobacco and Reynolds Industries v Commission, 142/84 and 156/84, EU:C:1987:490, paragraph 1). The Court held in that case that, where the Commission finds that an agreement does not breach competition law, it is required not only to take account of the effects that the clauses of that agreement had at the time of their examination by the Commission but also the effects that they could have in the future in the light of the as yet unrealised possibilities they open to the parties. For example, in that case, an agreement relating to acquisitions of shareholdings in a competing undertaking gave the investing undertaking the possibility of reinforcing its position at a later stage by taking effective control of the other undertaking, which could have consequences on the competitive situation examined (judgment of 17 November 1987, British American Tobacco and Reynolds Industries v Commission, 142/84 and 156/84, EU:C:1987:490, paragraphs 37, 39, 54, 57 and 58).
347 Thus, according to the judgment referred to in paragraph 346 above, the Commission must take into account, in the examination of the effects of an agreement, not only the actual effects of clauses which are already being implemented when it adopts its decision, but also the potential effects of clauses which have not yet been implemented.
348 The Court subsequently acknowledged the taking into account of potential effects of an agreement in the judgment of 28 May 1998, Deere v Commission (C‑7/95 P, EU:C:1998:256). The case that gave rise to that judgment concerned a Commission decision following the notification of an agreement aimed at obtaining, under Article 2 of Council Regulation No 17 of 6 February 1962, First Regulation implementing Articles [101 and 102 TFEU] (OJ, English Special Edition 1959-1962, p. 87), a negative clearance, by which the Commission could certify, upon application by the undertakings concerned, that there were no grounds for action on its part in respect of an agreement. In its decision, the Commission had found that the agreement notified to it constituted a restriction of competition by effect.
349 In the case that gave rise to the judgment of 28 May 1998, Deere v Commission (C‑7/95 P, EU:C:1998:256), the General Court, and subsequently the Court of Justice, upheld that finding, which was based on the existence of potential effects.
350 The applicant in that case relied on the fact that the information exchange system established by the agreement in question had been applied for several years before the notification of the request for negative clearance to support its argument that the Commission’s assessment should be limited to taking into account the actual effects of the exchange of information. However, the General Court considered that that argument was not relevant, since the Treaty prohibited both actual and potential effects of agreements (judgment of 27 October 1994, Deere v Commission, T‑35/92, EU:T:1994:259, paragraphs 59 and 61).
351 It is necessary, however, to qualify the conclusion that the argument based on the fact that the agreements or practices in question had been implemented was ineffective.
352 First, the circumstances of the case were unusual, because the agreement for which negative clearance was requested had replaced a previous agreement which had not been notified to the Commission. The Commission therefore had to decide on the compliance of that new agreement with the competition rules, and not on that of the previous agreement. It is therefore not certain that the Commission would have been able to draw definitive conclusions as regards that new agreement from the application of the previous agreement, despite their similarity. As regards the new agreement, it had been applied for only a few months before the participants decided to suspend it. The Commission thus did not have the necessary perspective to examine its actual effects on competition (judgment of 27 October 1994, Deere v Commission, T‑35/92, EU:T:1994:259, paragraphs 2 and 4).
353 Secondly, the General Court, when examining the potential effects on competition of an agreement in the judgment of 27 September 2006, GlaxoSmithKline Services v Commission (T‑168/01, EU:T:2006:265, paragraph 163), indicated that the fact that the agreement in question had been suspended only a few months after its entry into force, until the adoption of the Commission decision contested in that case, led it to interpret the Commission’s examination of the agreement in question as being mainly devoted to its potential effects.
354 The General Court, in the judgment of 27 September 2006, GlaxoSmithKline Services v Commission (T‑168/01, EU:T:2006:265), therefore established an explicit link between the fact that an agreement has not been implemented and the examination of its potential effects.
355 Thirdly, in the judgment of 30 June 2016, CB v Commission (T‑491/07 RENV, not published, EU:T:2016:379, paragraphs 243, 247, 248 and 250), the General Court examined the potential effects on competition of a decision of an association of undertakings by taking into account the effects that the measures in question would produce if they were applied, which again establishes a link between the examination of the potential effects of the association’s decision and the fact that it has not yet been implemented. It should be highlighted that the Commission had distinguished, in the decision in question (Commission Decision C(2007) 5060 final of 17 October 2007 relating to a proceeding under Article [101 TFEU] (COMP/D 1/38606 — Groupement des cartes bancaires ‘CB’)), between the analysis of potential effects, that is to say those that the measures would produce if they were no longer suspended (recitals 261 et seq. of that decision), and the analysis of the effects that had occurred in the course of the period during which the measures at issue had been applied (recitals 310 et seq. of that decision).
356 It should be noted that, in the cases that gave rise to the judgments of 27 September 2006, GlaxoSmithKline Services v Commission (T‑168/01, EU:T:2006:265), and of 30 June 2016, CB v Commission (T‑491/07 RENV, not published, EU:T:2016:379), the Commission did not penalise the undertakings concerned, but ordered them to bring an immediate end to the infringement in question.
357 It should also be added that, in the cases mentioned in paragraph 356 above, it was the undertakings concerned that had referred the measures in question to the Commission (see, to that effect, judgments of 27 September 2006, GlaxoSmithKline Services v Commission, T‑168/01, EU:T:2006:265, paragraph 10, and of 30 June 2016, CB v Commission, T‑491/07 RENV, not published, EU:T:2016:379, paragraph 8).
358 Thus, in most of the cases in which the EU Courts have applied to an agreement, a concerted practice or a decision of an association of undertakings the case-law according to which a finding of restriction by effect may arise from the potential effects of those measures, the Commission decision at issue did not penalise past conduct constituting a restriction by effect, but rather prevented the occurrence of such conduct by envisaging the effects that the measures in question could have if they were applied. That was the situation, inter alia, in the case which gave rise to the judgment of 17 November 1987, British American Tobacco and Reynolds Industries v Commission (142/84 and 156/84, EU:C:1987:490), in which the Commission rejected a complaint by examining the effects that a clause of the agreement in question might have if the possibility that it provided were implemented.
359 There is therefore no previous case-law, concerning agreements, decisions and concerted practices, in which the Court of Justice or the General Court has accepted that the Commission may rely only on the potential effects of the measure at issue in order to find that an infringement has been committed and impose a fine on the infringers on the basis of that finding.
360 It appears paradoxical — where the clauses of an agreement have been implemented and their impact on competition can be measured by taking into account the relevant factual developments, including those subsequent to the conclusion of the agreement, which took place before the Commission issued its decision — to allow the Commission to demonstrate merely the anticompetitive effects that such clauses are likely to have and, to that end, to make the comparison mentioned at paragraph 315 above without taking those developments into account.
361 It also appears paradoxical to allow the Commission, in order to find that an infringement in the form of a restriction of competition by effect was committed (and can therefore be penalised by a fine), to rely on the mere fact that clauses of an agreement that were implemented are likely to have anticompetitive effects and not on whether they had such effects, even though the Court of Justice has held that the burden of proving the anticompetitive effects of an agreement can be waived only in the case of a restriction of competition by object, which should concern only agreements so likely to have negative effects, in particular on the price, quantity or quality of the goods and services, that it may be considered redundant, for the purposes of applying Article 101(1) TFEU, to prove that they have actual effects on the market (judgment of 11 September 2014, CB v Commission, C‑67/13 P, EU:C:2014:2204, paragraph 51). If it were possible for the Commission to rely, in relation to agreements which have been implemented, solely on the effects that they are likely to have, in order to demonstrate that they had an anticompetitive effect, the distinction between restrictions of competition by object and by effect, established by Article 101(1) TFEU, would lose its relevance.
362 It follows from the foregoing that, since the agreements at issue were implemented and since, by the contested decision, the Commission found that an infringement had been committed, which enabled it to impose a fine on the parties to the agreements, the case-law just referred to in paragraphs 345 to 358 above, as regards taking the potential effects of agreements into account in relation to restrictions by effect, is not applicable.
363 It must also be noted, moreover, that the case-law mentioned in paragraphs 345 to 358 above must be distinguished from that concerning the taking into account of the effects on competition of a limitation of competition, including where that competition is only potential.
364 In that regard, in the judgment of 12 June 1997, Tiercé Ladbroke v Commission, T‑504/93, EU:T:1997:84, paragraphs 157 to 160), which is cited in recital 1217 to the contested decision, the General Court examined the legality of a Commission decision whereby the Commission had rejected a complaint on the ground, inter alia, that, in the absence of present competition on the relevant market, the agreement in question did not fall within the scope of Article 101(1) TFEU. The General Court held that the Commission had not examined with the required diligence all the matters of fact and of law brought to its attention by the applicant, because the agreement was likely to restrict potential competition. It therefore annulled the decision before it on that point.
365 It cannot be inferred from that case-law, which concerned a rejection of a complaint, that the mere fact that an agreement is ‘likely’ to restrict potential competition must necessarily lead to a finding of a restriction of competition by effect, but rather that the Commission cannot exclude from the outset the possibility of a restriction by effect where an agreement is only likely to restrict potential competition and not present competition.
366 Thus, when the Commission adopts a decision finding an infringement of Article 101(1) TFEU, which allows it to impose a fine on the infringers on the basis of that finding, the mere fact that the Commission has established the existence of potential competition and a limitation of the autonomy of a potential competitor, or even the elimination of that autonomy, does not release it from its obligation to demonstrate an analysis of the actual effects of the measure in question on competition if the case-law cited in paragraphs 345 to 358 is not applicable.
367 It must be borne in mind in that regard that the finding of anticompetitive effects of an agreement requires evidence that competition has, ‘in fact’, been prevented, restricted or distorted (judgment of 30 June 1966, LTM, 56/65, EU:C:1966:38, p. 249).
368 Thus, demonstrating that an agreement has anticompetitive effects requires that the Commission, in the light of the need to be realistic that arises from the case-law of the Court of Justice, take into account, in the context of the comparison referred to in paragraph 315 above, all the relevant factual developments, including those subsequent to the conclusion of the agreement, which took place before it adopts its decision.
369 In that regard, the Court of Justice has held that, when appraising the effects of an agreement between undertakings in the light of Article 101 TFEU, it is necessary to take into consideration the actual context in which the agreement in question is situated, in particular the economic and legal context in which the undertakings concerned operate, the nature of the goods or services affected, as well as the real conditions of the functioning and the structure of the market or markets in question (judgment of 11 September 2014, MasterCard and Others v Commission, C‑382/12 P, EU:C:2014:2201, paragraph 165).
370 It follows, as the applicant rightly submits, that the scenario envisaged on the basis of the hypothesis that the agreement in question was not concluded must, according to the Court of Justice, be ‘realistic’ (see, to that effect, judgment of 11 September 2014, MasterCard and Others v Commission, C‑382/12 P, EU:C:2014:2201, paragraph 166).
371 The Court of Justice indicated that taking into account developments that were likely to have occurred on the market in the absence of that agreement was necessary when examining the agreement’s restrictive effects on competition (see, to that effect, judgment of 11 September 2014, MasterCard and Others v Commission, C‑382/12 P, EU:C:2014:2201, paragraphs 167 to 169.
372 Moreover, the requirement of likelihood and realism applying to the description of the competition that would have occurred had an agreement not been concluded (one side of the comparison mentioned in paragraph 315 above) is consistent with the approach adopted by the Commission in several guidelines, which requires it to establish the sufficiently likely nature of the restrictive effects of the measures that it examines.
373 Thus, first, paragraph 24 of the Guidelines on the application of Article [101(3) TFEU], to which the Commission refers in recital 1219 of the contested decision, states that ‘for an agreement to be restrictive [of competition] by effect it must affect actual or potential competition to such an extent that on the relevant market negative effects on prices, output, innovation or the variety or quality of goods and services can be expected with a reasonable degree of probability’.
374 Secondly, in paragraph 19 of the Guidelines on the applicability of Article [101 TFEU] to horizontal cooperation agreements (OJ 2001 C 3, p. 2) (‘the 2001 Guidelines on horizontal cooperation agreements), it is indicated that many horizontal cooperation agreements do not have as their object a restriction of competition and that an analysis of the effects of each agreement is therefore necessary. It is added that, for that analysis, it is not sufficient that the agreement limits competition between the parties, but that the agreement must also be likely to affect competition in the market to such an extent that negative market effects as to prices, output, innovation or the variety or quality of goods and services can be expected.
375 Thirdly, the Commission confirmed that it maintained that approach in the Guidelines on the applicability of Article [101 TFEU] to horizontal cooperation agreements (OJ 2011 C 11, p. 1, ‘the 2011 Guidelines on horizontal cooperation agreements’). It thus states, in paragraph 28 of those guidelines, to which it refers in footnote 1733 of the contested decision, that restrictive effects on competition within the relevant market are likely to occur where it can be expected with a reasonable degree of probability that, due to the agreement, the parties would be able to profitably raise prices or reduce output, product quality, product variety or innovation.
376 Moreover, in the contested decision itself (recital 1218 of that decision), the Commission noted that restrictive effects on competition must be established with a sufficient degree of probability.
377 In the light of all the foregoing, it must be determined whether, in the present case, the Commission — despite the hypothetical approach that it adopted as regards the comparative step of the examination of restriction of competition by effect (see paragraphs 317 to 340 above) — established the sufficiently realistic and probable nature of the restrictive effects of the agreements concluded between Servier and Krka.
(c) The error of assessment
378 The Commission analysed the effects of the non-marketing and of the non-challenge clause contained in the settlement agreement between Servier and Krka, as well as the licensing of Krka’s technology to Servier, by examining, for each of those three measures, the competition that would have occurred in its absence (see, inter alia, recital 1825 to 1829 of the contested decision).
379 It is necessary to determine, for each of the three measures, whether the Commission was entitled to find a restriction of competition by effect.
(1) The non-marketing clause in the settlement agreement
380 It must be borne in mind that, in order to determine whether an agreement is to be considered to be prohibited by reason of the distortion of competition to which it gives rise, the competition in question should be assessed within the actual context in which it would occur in the absence of the agreement in dispute (see paragraph 315 above).
381 In the present case, the scope of the non-marketing clause is limited to that of the 947 patent, which is the subject of proceedings between Servier and Krka.
382 The actual context of the competition, absent the settlement agreement, consisted of the attempts of generic companies, including Krka, to enter the market by overcoming obstacles linked to Servier’s patents, in particular the 947 patent, and the patent litigation between those companies and Servier.
383 As noted in paragraph 143 above, the specific purpose of awarding a patent is, inter alia, to ensure that the patentee, in order to reward the creative effort of the inventor, has the exclusive right to use an invention with a view to manufacturing industrial products and putting them into circulation for the first time, either directly or by the grant of licences to third parties, as well as the right to oppose infringements (judgment of 31 October 1974, Centrafarm and de Peijper, 15/74, ECR, EU:C:1974:114, paragraph 9). When granted by a public authority, a patent is normally assumed to be valid and an undertaking’s ownership of that right is assumed to be lawful. The mere possession by an undertaking of such an exclusive right normally results in keeping competitors away, since public regulations require them to respect that exclusive right (judgment of 1 July 2010, AstraZeneca v Commission, T‑321/05, EU:T:2010:266, paragraph 362).
384 It is true that the at risk market entry of a generic company is not unlawful in itself (see, to that effect, judgment of 8 September 2016, Lundbeck v Commission, T‑472/13, under appeal, EU:T:2016:449, paragraph 122). However, such entry is much less likely where the generic company recognises the validity of the patent or believes that its chances of having that patent declared invalid are low.
385 A generic company’s recognition or non-recognition of the validity of the patent in question or its perception of the strength of that patent is therefore decisive when determining whether it is likely to enter the market at risk.
386 The Commission did not properly take into account the effects that the 947 patent and Krka’s recognition of its validity could have on the assessment, for the purpose of the comparison referred to in paragraph 315 above, of Krka’s likely behaviour in the absence of an agreement, in the section of the contested decision devoted to the examination of that conduct (recitals 1825 to 1834 of that decision).
387 Events decisive for assessing whether Krka recognised the validity of the 947 patent, or its perception of its prospects of success in having that patent declared invalid, such as the EPO decision of 27 July 2006 confirming the validity of the patent and the interim injunction against Krka issued by a court in the United Kingdom, are not mentioned in that section of the contested decision, despite the fact that they occurred even before the conclusion of the settlement agreement between Servier and Krka.
388 In addition, the Commission, in recitals 1828 to 1834 of the contested decision, in the analysis of Krka’s likely behaviour absent the agreements concluded with Servier, does not refer to the circumstance — which is significant in this context — that several pieces of evidence in the case file supported the finding that Krka might be infringing the 947 patent.
389 That confirms that the hypothetical approach adopted by the Commission (see paragraphs 316 to 341 above) led it not only to disregard the events that occurred after the conclusion of the agreements but, more generally, to disregard the actual course of events as it could have been observed when it adopted its decision.
390 The Commission’s reluctance to take into account in particular the effects of the 947 patent can be explained by the fact that, in its analysis of the restriction by object, it considered that the real basis for the settlement agreement between Servier and Krka was the inducement of the latter to comply with the restrictive clauses of that agreement and not a genuine recognition of the validity of the 947 patent. From that perspective, Krka could not in any way, according to the Commission, rely on the recognition of the validity of the 947 patent, since that recognition was vitiated in its very principle.
391 However, the finding of an inducement and of a restriction by object made by the Commission has been invalidated by the General Court as regards the settlement and licence agreements between Servier and Krka (see paragraph 268 above), which gives renewed relevance to the taking into account of Krka’s perception of the strength of the 947 patent or its recognition of the validity of that patent.
392 It must be borne in mind that there was, at the time the settlement and licence agreements were concluded, strong indications capable of leading the parties to believe that the 947 patent was valid (see paragraph 206 above). In the United Kingdom, that is to say one of the three countries (with France and the Netherlands) in which the Commission analysed and found a restriction by effect, Krka and Apotex, another competitor of Servier, were even the subject of an interim injunction.
393 Although the request for an interim injunction prohibiting the marketing of a generic version of perindopril placed on the market by Krka because of the infringement of the 947 patent, which was introduced by Servier in Hungary, was rejected in September 2006, it was a procedure which, unlike those mentioned in paragraph 392 above, did not concern any of the countries in which the Commission found a restriction by effect.
394 Moreover, although there were already meetings between Servier and Krka before the EPO decision of 27 July 2006 (see, inter alia, recital 837 of the contested decision), they had not resulted in an agreement (recitals 856 to 859 to the contested decision) and it was only after that decision that new negotiations began (recital 898 of the contested decision). The EPO decision of 27 July 2006 was therefore, at the very least, one of the catalysts which led to the settlement and licence agreements, which is further evidence that those agreements were based on the parties’ recognition of the validity of the patent (see paragraph 207 above).
395 It should also be added that, as noted above (see paragraph 183 above), the conclusion of a licence agreement, which, for any licensee, makes sense only if the licence is actually used, is based on the parties’ recognition of the validity of the patent (see paragraph 183 above). Thus, the very conclusion of the licence agreement, supported by a number of pieces of evidence (see paragraphs 235 and 237 above), confirms that Krka ultimately recognised the validity of the 947 patent.
396 It is even apparent from documents in the file that Krka seemed to consider that, in the absence of a licence agreement with Servier, entry to the 18 to 20 markets at risk was very unlikely, or even impossible (see paragraphs 235, 236 and 247 above).
397 Lastly, the Commission indicated, in the contested decision (recital 1693 of that decision), that Krka had ‘eventually ceased to consider entering at risk in the UK, France and other Western European markets in the aftermath of the [EPO decision of 27 July 2006]’.
398 In the light of the factors set out above, it must be concluded that it has not been established that, in the absence of an agreement, Krka would probably have entered at risk the markets of the 18 to 20 Member States, in particular the markets of France, the Netherlands and the United Kingdom.
399 The above conclusion is not called into question by the other elements in the file which might be relevant for the purposes of establishing that Krka would have entered the market had an agreement with Servier not been concluded. Those elements are contained primarily in the part of the contested decision devoted to the Commission’s demonstration that Krka was a potential competitor of Servier.
400 First, it must be borne in mind (see paragraph 262 above) that the fact that Krka continued to challenge Servier’s patents and to market its product even though the validity of the 947 patent had been upheld by the EPO’s Opposition Division can obviously be explained by Krka’s desire to strengthen its position in the negotiations that it might engage in with Servier with a view to reaching a settlement.
401 In addition, continuing to challenge Servier’s patent did not cause Krka to run any further risks in terms of infringement. It merely increased its litigation costs. With regard to the continued marketing of its product, the Commission confined itself to the five Central and Eastern Europe markets, in respect of which the Commission did not find a restriction of competition by effect. In addition, in five of the seven markets covered by the licence, the equivalents of the 947 patent had not yet been granted (recital 1755 of the contested decision). Thus, the risks incurred by Krka, in at least some of the markets in which it remained, were limited (see paragraph 263 above).
402 Krka’s decision to continue challenging Servier’s patent and to continue marketing its product therefore do not support the conclusion that Krka did not recognise the validity of the 947 patent and would therefore have probably entered the markets of the 18 to 20 Member States at risk or, at the very least, the three markets in respect of which the Commission found a restriction of competition by effect.
403 Secondly, although the comments made by Krka’s representatives show their surprise and discontent following the EPO decision of 27 July 2006 (recital 1688 of the contested decision), those comments cannot establish that, despite that decision, Krka would probably have entered the three national markets in respect of which the Commission found a restriction by effect.
404 Thirdly, the Commission devotes a section of the contested decision to Krka’s ‘intention to enter’ the market. That very short section is composed of only one recital, which is itself rather short: recital 1699 of that decision. In that recital, the Commission states that, ‘even’ after the EPO decision of 27 July 2006, Krka ‘appeared’ willing to support launches at risk by its partners and that it remained committed to supply its product ‘in case the patent barriers were overcome’. It is added in that recital that one of Krka’s commercial partners urged it to supply its product ‘in case the ‘947 patent was invalidated’ and that some of Krka’s partners entered the market with that product ‘once the ‘947 patent was invalidated in the [relevant] markets’.
405 The extracts cited in paragraph 404 above attest less to Krka’s intention to enter the three national markets in respect of which the Commission found a restriction by effect than to the importance attached to the ‘patent barrier’ represented by the 947 patent after the EPO decision of 27 July 2006, both for Krka and its commercial partners.
406 In view of all the elements set out above, it has not been established that, had the settlement and licence agreements not been concluded, Krka would probably have entered the three national markets in respect of which the Commission found a restriction of competition by effect.
407 Nor was it established by the Commission in the contested decision that, absent those agreements, Krka would have probably entered the markets concerned before the date on which the infringement came to an end, namely before 6 July 2007 for the United Kingdom, 12 December 2007 for the Netherlands and 16 September 2009 for France.
408 The hypothetical approach adopted by the Commission (see paragraphs 317 to 341 above) led it to pay little attention to the actual course of events — in particular those that occurred after the conclusion of the agreements — and therefore to possible changes in Krka’s perception of the validity of the 947 patent as a result of those events.
409 However, it is not for the Court, as regards the appraisal of the constituent elements of an infringement — which do not fall within the scope of its unlimited jurisdiction, but rather the review of legality — to substitute its own reasoning for that of the Commission (judgment of 21 January 2016, Galp Energía España and Others v Commission, C‑603/13 P, EU:C:2016:38, paragraphs 73 and 75 to 77).
410 Therefore, it is not for the Court to examine, for the first time, on the basis of the evidence in the file, whether a restriction of competition by effect could have arisen during the period after the conclusion of the agreements as a result of a decline in Krka’s recognition of the validity of the 947 patent.
411 In any event, the evidence in the file does not support the conclusion that, had the agreements not been concluded, Krka would probably have entered the three national markets concerned during the period between the conclusion of the agreements and the end of the infringement.
412 Moreover, it should be emphasised that the Commission does not even allege that Krka would have probably entered the market in the absence of an agreement. In the section of the contested decision entitled ‘Likely behaviour absent the Krka Agreements’, the Commission does not rely, at least explicitly, on a hypothesis of an early market entry by Krka on the three markets concerned in the absence of an agreement, but only on the hypothesis that it would have remained a ‘competitive threat’ on those markets (see paragraph 334 above).
413 Thus, according to the Commission, ‘Krka would have remained a competitive threat as a potential generic entrant with perindopril in the UK, France and the Netherlands’ (recital 1825 of the contested decision). The Commission notes that Krka would, inter alia, have continued to be a threat as a supplier to local distribution partners (recital 1828 of the contested decision).
414 The Commission also indicates that Servier and Krka could have, in the absence of an inducement, negotiated a less restrictive agreement granting Krka earlier entry or a licence for the entire EU territory (recital 1831 of the contested decision).
415 The Commission concludes by indicating that, ‘in the absence of the restrictions in the ... Agreements, Krka would have remained a prominent potential competitor to Servier’ (recital 1834 of the contested decision).
416 It should be noted that, by merely invoking the ‘competitive threat’ that Krka would have continued to represent for Servier, even though the procompetitive effects of a mere ‘threat’ are not — unlike those of the market entry of a generic company — evident and, moreover, the effects of that ‘threat’ were, in the present case, largely mitigated by the presence of the 947 patent and the confirmation of its validity by the competent authorities (see paragraphs 380 to 407 above), the Commission failed to establish that the competition that would have occurred in the absence of the settlement agreement would probably have been more open.
417 It may be noted, in that respect, that the Commission should have specified the probable effects, in particular on prices, production, quality, diversity of products or innovation (see paragraphs 373 to 375 above), of the ‘competitive threat’ that Krka would have continued to represent for Servier in the absence of the settlement agreement, which it could have done, for example, by demonstrating that, because of the absence of a threat, Servier had limited its research and development expenditure.
418 It should be noted that, although the Commission’s analysis of Servier’s market power, and the structure of the relevant market, characterised by a lack or shortage of sources of competition, could have supported a finding of restrictive effects of an agreement preventing the market entry of a potential competitor, it is not enough to make probable and concrete the restrictive effects of an agreement undermining the existence of a ‘competitive threat’.
419 Irrespective of the structure of the market, the anticompetitive effects of the non-marketing clause remain largely hypothetical if it is likely, given the actual course of events as it could have been observed when the Commission adopted its decision, that, even in the absence of that clause, the potential competitor concerned might have behaved similarly to how it did in the presence of the clause, that is to say, in the present case, that Krka would have remained outside the three markets in respect of which the Commission found a restriction by effect.
420 The credibility of the hypothesis that, in the absence of the settlement and licence agreements between Servier and Krka, and, in particular, of the inducement that, according to the Commission, they contained, another agreement allowing the early entry of Krka or granting it a licence for the European Union in its entirety would have been concluded (see paragraph 414 above and recital 1142 of the contested decision), is in no way established, especially since, as is apparent from the examination of the plea alleging the absence of restriction of competition by object, the existence of an inducement has not been established by the Commission.
421 Lastly, it must be highlighted that the specific context of the settlement and licence agreements between Servier and Krka, which was characterised by the presence of a patent the validity of which was confirmed by the EPO (see paragraph 382 above), is different from that at issue in the case that gave rise to the judgment of 14 April 2011, Visa Europe and Visa International Service v Commission, T‑461/07, EU:T:2011:181, paragraphs 187 and 191), which is cited by the Commission, in particular, in recital 1219 of the contested decision. In the absence of background factors comparable to those, relating to the existence of a patent and the recognition of its validity, which have been set out above (see, inter alia, paragraphs 383 to 397 above) and which are decisive in the present proceedings, the Court considered, in that judgment, on the basis of the sole fact that an undertaking subject to an exclusion clause by the measure at issue was a potential competitor, that the Commission had been entitled to conclude that that undertaking would have entered the market in the absence of the exclusion clause.
422 It should also be noted that, in the judgment of 14 April 2011, Visa Europe and Visa International Service v Commission (T‑461/07, EU:T:2011:181), the Court did not confirm a decision-making practice of the Commission whereby it could, in cases involving the elimination of a potential competitor, ignore the actual course of events as it could have been observed when it adopted its decision.
423 Moreover, such a practice, if it were valid, could lead to an inconsistent outcome in some cases, for example, where the only potential competitor, which is eliminated by an agreement, has been wound up by the time it is implemented, as a result of insolvency for example, which would obviously neutralise the exclusionary effects of the agreement, except if those effects were envisaged in a hypothetical manner, and not realistically, as required by the case-law (see paragraphs 367 and 370 above).
424 A restriction of competition by effect therefore cannot be found in the present case by reference to the judgment of 14 April 2011, Visa Europe and Visa International Service v Commission (T‑461/07, EU:T:2011:181).
425 It follows from the foregoing that the competition-restricting effects of the non-marketing clause in the settlement agreement have not been established by the Commission.
(2) The non-challenge clause in the settlement agreement
426 As a preliminary point, it should be observed that, in the section of the contested decision entitled ‘Likely behaviour absent the Krka Agreements’, the Commission does not mention anything relating to Krka’s likely behaviour as regards the 340 patent, in respect of which the settlement agreement also contains a non-challenge clause.
427 Consequently, in the stage of the analysis of restriction by effect consisting in a comparison between the competition in the presence of the agreements and the competition that would have existed in their absence (see paragraphs 315 above), the Commission limited its analysis to the 947 patent.
428 Furthermore, the Commission indicated, also in the section of the contested decision entitled ‘Likely behaviour absent the Krka Agreements’, that ‘it appears plausible that, absent the non-challenge obligation, Krka would have remained a challenger to the validity of the ‘947 [patent] before the UK courts and the EPO’ (recital 1827 of the contested decision).
429 The Commission therefore based its finding of a restriction by effect on the fact that, in the absence of the non-challenge clause, Krka would have continued the proceedings in which it was engaged before the United Kingdom courts and the EPO.
430 In that respect, it should be noted that a non-challenge clause is, by itself, restrictive of competition, since it undermines the public interest in eliminating any obstacle to economic activity which may arise where a patent was granted in error (see, to that effect, judgment of 25 February 1986, Windsurfing International v Commission, 193/83, EU:C:1986:75, paragraph 92).
431 It is therefore necessary to determine whether the application of the non-challenge clause and, in particular, Krka’s withdrawal from the proceedings in which it was engaged, had an effect as regards the elimination of the 947 patent.
432 It should be borne in mind that, at the time the settlement and licence agreements were concluded, Krka and Servier were engaged in two sets of proceedings against one another and that it was the settlement agreement that led Krka not to pursue those proceedings.
433 Thus, in the United Kingdom, on 2 August 2006, Servier had brought an action for infringement of the 947 patent against Krka before the High Court of Justice (England and Wales), Chancery Division (Patents Court). It had also made an application for an interim injunction. On 1 September 2006, Krka had lodged a counterclaim for the annulment of the 947 patent. On 3 October 2006, the High Court of Justice (England and Wales), Chancery Division (Patents Court), had upheld Servier’s application for interim relief and had refused the motion for summary judgment lodged by Krka on 1 September 2006, seeking a declaration of invalidity of the 947 patent. On 1 December 2006, the ongoing proceedings were discontinued as a result of the settlement reached between the parties and the interim injunction was lifted.
434 As regards the proceedings before the EPO, in 2004, ten generic companies, including Krka, filed opposition proceedings against the 947 patent before the EPO seeking the revocation of that patent on grounds of lack of novelty, lack of inventive step and insufficient disclosure of the invention. On 27 July 2006, the EPO’s Opposition Division had confirmed the validity of that patent following minor amendments to Servier’s original claims. Seven companies had brought an appeal against that decision. Krka withdrew from the opposition procedure on 11 January 2007, pursuant to the settlement agreement reached between the parties.
435 It must be borne in mind, however, that, on 1 August 2006, Servier had also brought an action for infringement before the High Court of Justice (England & Wales), Chancery Division (Patents Court), against the company Apotex, claiming infringement of the 947 patent, since Apotex had launched a generic version of perindopril in the United Kingdom on 28 July 2006. Apotex had brought a counterclaim for annulment of that patent. An interim injunction prohibiting Apotex from importing, offering to sell or selling perindopril had been obtained on 8 August 2006.
436 On the basis of the counterclaim lodged by Apotex, the High Court of Justice (England & Wales), Chancery Division (Patents Court), ruled, on 6 July 2007, that the 947 patent was invalid because it lacked novelty and inventive step over the 341 patent. Consequently, the injunction was lifted immediately and Apotex was able to resume selling its generic version of perindopril on the United Kingdom market.
437 The Commission considered that the infringement concerning the agreements concluded between Servier and Krka had ended on that date in the United Kingdom.
438 Furthermore, as regards the litigation before the EPO, on the basis of the proceedings initiated by Krka, amongst others, the EPO’s Technical Board of Appeal, by decision of 6 May 2009, annulled the EPO decision of 27 July 2006 and dismissed the 947 patent.
439 The Commission found that the infringement concerning the agreements concluded between Servier and Krka, in so far as it was still taking place in certain Member States, had ended on that date.
440 In the light of the proceedings relating to the 947 patent, which continued after Krka’s withdrawal from the proceedings to which it was a party, as mentioned above, it cannot be considered that, in the absence of the settlement agreement reached between the parties, Krka’s continuation of the proceedings would probably, or even plausibly, have allowed a faster or more complete invalidation of the patent.
441 However, the Commission has not established, or even alleged, in the contested decision, that the invalidation of the 947 patent would have been more rapid or more complete if Krka had not agreed to the non-challenge clause in the settlement agreement.
442 The fact that ‘Krka [had] previously considered that its patent case was amongst the best ones, and that it was a particular threat to the ‘947 patent’ or that the courts of the United Kingdom, despite their rejection of the motion for summary judgment lodged by Krka, had considered that it had a ‘powerful base’ to challenge the validity of the 947 patent (recital 1827 of the contested decision) does not support the conclusion that Krka’s participation in the proceedings in question would have led to the faster or more complete invalidation of the patent.
443 Likewise, to note, as the Commission does in recital 1712 of the contested decision (which is referred to in footnote 2445 of that decision), that ‘eliminating a strong challenger may impact the final outcome of the litigation/opposition’ does not justify a finding that the effects of the non-challenge clause that applied to Krka are probable, or even plausible.
444 It was for the Commission to demonstrate, in a sufficiently precise and substantiated manner, how Krka’s arguments or its particular position as regards the litigation could, if it had continued the proceedings in which it was engaged, have had a decisive impact, not on the outcome of disputes, since two of those cases — namely that before the EPO, which continued after Krka’s withdrawal and that between Servier and Apotex before the High Court of Justice (England and Wales), Chancery Division (Patents Court) — had, in any event, resulted in the invalidation of the 947 patent, but on the period in which that invalidation occurred or its scope.
445 Moreover, it is not for the Court, as regards the appraisal of the constituent elements of an infringement — which do not fall within the scope of its unlimited jurisdiction, but rather the review of legality — to substitute its own reasoning for that of the Commission (judgment of 21 January 2016, Galp Energía España and Others v Commission, C‑603/13 P, EU:C:2016:38, paragraphs 73 and 75 to 77).
446 Therefore, it is not for the Court to examine for the first time, on the basis of elements in the file other than those relied upon by the Commission in order to establish the restrictive effects of the non-challenge clause, whether Krka’s continued participation in the ongoing litigation would have resulted in the faster or more complete invalidation of the 947 patent.
447 In any event, in the light of the considerations set out in paragraph 444 above, the elements in the file that might be relevant do not lead to that conclusion, whether they concern the fact that ‘Krka believed [it had] a strategic advantage due to its superior evidence’ (recitals 851 and 1685 of the contested decision), that another company continuing to market generic perindopril had indicated that Krka’s nullity suit was ‘the most promising one’ (recital 867 of the contested decision), or the fact that, according to Krka, Servier believed that Krka had some of the best and most comprehensive evidence for the opposition before the EPO and the revocation in the UK (recitals 912 and 1688 of the contested decision).
448 It should also be added that, irrespective of the structure of the relevant market, including where, as in this case, it is characterised, according to the Commission, by a lack or shortage of sources of competition, the anticompetitive effects of a non-challenge clause remain largely hypothetical if it is likely, given the actual course of events as it could have been observed when the Commission adopted its decision, that, in the absence of that clause, the patent in question, namely, in the present case, the 947 patent, would have been invalidated at the same time and in equal measure (see paragraph 419 above).
449 Moreover, the Commission has not demonstrated, contrary to what recital 1712 of the contested decision suggests, that the proceedings between Servier and Krka before the courts of the United Kingdom could have established that Krka’s technology was non-infringing. The proceedings concerning Krka and Apotex consisted of actions for infringement brought by Servier and counterclaims for annulment of the 947 patent lodged by those two generic companies in response. These proceedings were therefore similar. The proceedings concerning Apotex were brought to an end in their entirety by the invalidation of the 947 patent and, thus, without it being necessary to determine whether its technology was infringing. It is plausible, given the similarity of the proceedings and in the absence of evidence to the contrary adduced by the Commission, that the same would have happened to Krka.
450 A fortiori, it has not been shown that the procedure before the EPO could have established that Krka’s technology was a non-infringing technology since that procedure concerned only the validity of the 947 patent.
451 It follows from the foregoing that the restrictive effects on competition of the non-challenge clause in the settlement agreement have not been established by the Commission.
(3) The licensing of Krka’s technology
452 As regards the licence agreement by which Krka sold its technology to Servier, the Commission merely noted that, absent that agreement, ‘Krka would have retained the freedom to sell or license out the rights to its perindopril technology’ (recital 1829 of the contested decision), which is not sufficient — in relation to a mere transfer of property accompanied by a licence agreement and not an exclusionary measure as a non-marketing clause may be — to establish the existence of probable effects, in particular on prices, production, quality, diversity or innovation (see paragraphs 373 to 375 above). The existence of anticompetitive effects is even less established since Krka’s technology did not make it possible to circumvent the 947 patent, which, in view of the serious indications suggesting that that patent was valid, renders implausible, as the applicant rightly submits, the assumption that generic companies competing with Servier would, absent the assignment agreement, have sought to acquire Krka’s technology.
453 It follows from the foregoing that the restrictive effects on competition of the licensing of Krka’s technology have not been established by the Commission.
454 It follows from all the foregoing that the Commission has not established the existence of a restrictive effect on competition resulting from the settlement agreement or the assignment agreement which is sufficiently likely as to be able to support the finding of a restriction by effect. It should be added that such a restrictive effect is no more likely to be found if the two agreements are considered as a whole.
455 The complaint alleging an error of assessment must therefore be upheld, and the applicant’s plea alleging that the Commission erred in finding a restriction by effect as a result of the agreements between Servier and Krka, may, on that basis alone, be declared well founded in its entirety.
456 It is also necessary to determine whether the Commission has, in addition, as the applicant submits, vitiated its decision by an error of law.
(d) Error of law
457 As mentioned (see paragraphs 330 and 340 above), the Commission considered that, since it had established that the settlement agreement excluded a potential competitor of Servier, it was not required, in order to demonstrate the competition that would have occurred had an agreement not been concluded (one side of the comparison mentioned in paragraph 315 above), to take into account the actual course of events which it could have observed at the time it adopted its decision. Rather, the Commission considered — relying on its usual practice in taking into account the potential effects of an agreement, according to which it suffices to demonstrate that that agreement is ‘likely’ to have anticompetitive effects (see paragraphs 319 and 324 above) — that it could base its description of the competition that would have occurred had an agreement not been concluded on hypotheses or possibilities.
458 As is apparent from the above examination of the complaint alleging an error of assessment, some of the events that the Commission did not take into account were not only relevant, but also decisive for the purposes of the comparison referred to in paragraph 315 above.
459 Thus, as regards the non-marketing clause, although the Commission took into account the EPO decision of 27 July 2006 and the interim injunctions issued by the United Kingdom courts against Krka and Apotex in order to establish that Krka was a potential competitor, it did not take due account of those events for the purpose of determining whether Krka would probably have entered the market absent an agreement, merely stating in that regard that, absent an agreement, the ‘competitive threat’ from Krka would have persisted.
460 As regards the non-challenge clause, the Commission did not take into account the outcome of the proceedings brought against the 947 patent by other generic companies, which continued despite the fact that Krka had ceased any challenge.
461 As regards, lastly, the market structure, a cross-cutting issue which concerns both the non-marketing clause and the non-challenge clause, the Commission merely identified the remaining sources of competition at the time the last of the settlement agreements referred to in the contested decision was concluded and indicated that there was a ‘strong possibility’ that those sources would be removed from competition by an agreement or otherwise, without taking account of the fact that that possibility did not occur during the infringement period (recital 1846 of the contested decision).
462 That reasoning is expressly apparent from footnote 2445 of the contested decision, in which the Commission relies — in order to prove that the non-challenge clause had restrictive effects — on the fact that there were few companies competing with Servier likely to pursue the ongoing proceedings or to launch new ones and that ‘it was plausible that Servier would consider reaching settlements with these companies’, which would have eliminated any possibility that proceedings against the 947 patent would continue or be launched. While it is true that Servier approached those companies, it did not reach a settlement with them and in particular with one of them which ultimately obtained the annulment of the 947 patent at the very time the non-challenge clause was applied by Krka.
463 The limited nature of the review undertaken by the Commission cannot be justified in the light of the case-law of the EU Courts. The case-law on taking into account the potential effects of agreements, examined in paragraph 345 to 358 above, was not applicable in the present case (see paragraph 362 above).
464 The same is true, for the reasons indicated in paragraphs 421 to 424 above, as regards the applicability of the solution adopted, in relation to agreements eliminating potential competition, in the judgment of 14 April 2011, Visa Europe and Visa International Service v Commission (T‑461/07, EU:T:2011:181) (see paragraphs 421 to 424 above).
465 It must therefore be concluded that the Commission carried out an incomplete examination of the situation that it was required to assess in order to determine whether the agreements between Servier and Krka were restrictive of competition by effect, and that the incompleteness of the Commission’s examination shows a misapplication of the case-law of the EU courts and thus an error of law.
466 Moreover, according to the Commission’s approach, it had only to demonstrate the elimination of a potential competitor in order to be able to find — in the context of a market structure characterised by a lack or shortage of sources of competition and by market power on the part of the originator company — a restriction of competition by effect.
467 If it were accepted, that approach would allow the Commission, in cases such as the present case concerning restrictive clauses linked to a settlement agreement in relation to medicinal product patents, to find a restriction of competition by effect by, in essence, ensuring that only two of the three conditions required in order to find a restriction by object, namely the existence of potential competition and the presence of clauses restrictive of competition, are met.
468 Since demonstrating that the third condition, namely the existence of an inducement, is met is, as is clear from the examination of the plea on restriction by object, particularly difficult, the Commission’s task would be made significantly easier.
469 In the light of the higher evidential requirements that apply to the demonstration of a restriction of competition by effect (see paragraphs 361 and 366 to 377 above), that solution — which runs counter to the spirit of the distinction established by the Treaty between restrictions of competition by object and restrictions of competition by effect — cannot be accepted.
470 It follows from the foregoing that the complaint alleging an error of law must be upheld, and the applicant’s plea alleging that the Commission erred in finding a restriction by effect resulting from the agreements between Servier and Krka may, on that basis alone, be declared well founded in its entirety.
IV. Overall conclusion
471 Since the pleas relating to the absence of a restriction of competition by object, as well as that relating to the absence of a restriction of competition by effect, are well founded, it must be concluded, without there being any need to examine the other pleas in law, that the Commission erred in finding an infringement under Article 101(1) TFEU as regards the agreements between Servier and Krka.
472 It is therefore appropriate to annul Article 4 of the contested decision in so far as, by that provision, the Commission found that Krka had participated in an infringement of Article 101(1) TFEU as regards the agreements between Servier and Krka.
473 In order to take account of the annulment of Article 4 of the contested decision in so far as the Commission found in that provision that Krka had participated in an infringement of Article 101(1) TFEU, it is necessary to annul Article 7(4)(a) of the contested decision, by which the Commission imposed a fine totalling EUR 10 million on Krka in respect of that infringement. It is also appropriate to annul Articles 8 and 9 of the contested decision in so far as they relate to Krka.
V. Costs
474 Under Article 134(1) of the Rules of Procedure, the unsuccessful party is to be ordered to pay the costs if they have been applied for in the successful party’s pleadings. Since the Commission has been unsuccessful, it must be ordered to pay the costs, in accordance with the form of order sought by the applicant.
On those grounds,
THE GENERAL COURT (Ninth Chamber)
hereby:
1. Annuls Article 4 of Commission Decision C(2014) 4955 final of 9 July 2014 relating to a proceeding under Article 101 and Article 102 TFEU (Case AT.39612 — Perindopril (Servier)) in so far as it finds that Krka Tovarna Zdravil d.d. participated in the agreements referred to in that article;
2. Annuls Article 7(4)(a) of Decision C(2014) 4955 final;
3. Annuls Articles 8 and 9 of Decision C(2014) 4955 final, in so far as they concern Krka Tovarna Zdravil;
4. Orders the Commission to pay the costs.
Gervasoni | Madise | da Silva Passos |
Delivered in open court in Luxembourg on 12 December 2018.
Table of contents
I. Background to the dispute
A. Perindopril
1. The compound patent
2. Secondary patents
B. The applicant
C. Disputes relating to perindopril
1. Disputes before the EPO
2. Disputes before the national courts
(a) Dispute between Servier and Krka
(b) Dispute between Servier and Apotex
D. Patent dispute settlements
E. The Sector Inquiry
F. The administrative procedure
II. Procedure and forms of order sought
III. Law
A. Third plea in law, alleging that there was no restriction of competition by object as regards the settlement and licence agreements
1. Arguments of the parties
(a) The absence of restriction by object since the applicant was not induced to abandon its efforts to enter several EU markets with a generic product
(b) The absence of restriction by object as a result of the lawful nature of the use of a settlement
(1) The allegation that there were genuine disputes between the parties and that the conclusion of the settlement agreement was based on the parties’ evaluation of those disputes
(2) The allegation that the non-challenge clause cannot be classified as an infringement of competition law
(3) The allegation that the non-marketing clause cannot be classified as an infringement of competition law
(4) The allegation that the principles relating to the restriction of competition by object were applied incorrectly
2. Findings of the Court
(a) Restrictions of competition by object
(b) Intellectual property rights and, in particular, patents
(c) Patent dispute settlement
(d) The reconciliation of patent settlement agreements and competition law
(e) Side deals
(f) Licence agreements
(g) The facts of the case
B. Fourth plea in law, alleging that there is no restriction of competition by object as regards the assignment agreement
1. Arguments of the parties
2. Findings of the Court
C. Fifth plea in law, alleging that there is no restriction of competition by effect
1. Arguments of the parties
2. Findings of the Court
(a) The approach adopted by the Commission
(b) The relevant case-law in the present case
(c) The error of assessment
(1) The non-marketing clause in the settlement agreement
(2) The non-challenge clause in the settlement agreement
(3) The licensing of Krka’s technology
(d) Error of law
IV. Overall conclusion
V. Costs