delivered on 19 June 2025 (1)

Case C‑738/22 P

Google LLC,

Alphabet Inc.

European Commission

( Appeal – Competition – Abuse of a dominant position – Smart mobile devices – Decision finding an infringement of Article 102 TFEU and Article 54 of the EEA Agreement – Google Android operating system – Contractual restrictions – Tying of the search and browser applications Google Search and Chrome with the app store Play Store – Anti-fragmentation obligations – Exclusivity payments – Exclusionary effects – Hypothetically as-efficient competitor – Objective justification – Single and continuous infringement – Recalculation of the fine )

Table of contents

II. Background to the dispute

B. The decision at issue

III. Procedure before the Court of Justice and forms of order sought

(a) The causal link between the MADAs and their exclusionary effects (first, third and fourth parts of the first ground of appeal)

(1) Taking into account the RSAs when analysing the MADAs (first part of the first ground of appeal)

(2) The distinction between default setting and pre-installation (third part of the first ground of appeal)

(i) No new assessment of the facts on appeal

(ii) No assessment of the effects of a practice isolated from its context

(3) The alleged pro-competitive component of the MADAs and the requirement of a counterfactual assessment (fourth part of the first ground of appeal)

(b) MADAs’ capacity to foreclose as-efficient competitors (second part of the first ground of appeal and second ground of appeal)

(1) The assessment of the exclusionary effects of the MADAs as regards ‘as-efficient competitors’ and the reasons for users’ behaviour

(i) Starting point of the case-law on tying

(ii) Findings of the General Court regarding the MADAs’ exclusionary effects

–Creation of a ‘status quo bias’ by way of pre-installation

–The possibility for OEMs to pre-install competing search services and to set them as default

–Means of reaching users other than pre-installation

–Connection between usage shares and pre-installation

–Taking account of the context

(iii) Assessment

(2) The tying of the Google Search app and the Play Store

(3) The tying of Chrome with the Play Store and Google Search

(c) Conclusion on the assessment of the MADAs

(a) The scope of the abuse in relation to the AFAs and its effects (third ground of appeal)

(1) The establishment of the contested conduct in relation to the AFAs (first part of the third ground of appeal)

(2) The attribution of the alleged exclusionary effects (second part of the third ground of appeal)

(b) The justification of the AFAs (fourth ground of appeal)

(c) Conclusion on the assessment of the AFAs

3. The single and continuous infringement (fifth ground of appeal)

4. Conclusion on the establishment of the infringement

B. The fine (sixth ground of appeal)

VI. Conclusion

I.Introduction

As the General Court noted in its judgment in Google and Alphabet v Commission (Google Android), (2) the verb ‘to google’ refers to the action of ‘[using] the Google search engine to obtain information about someone or something on the World Wide Web’. As the General Court rightly points out, very few undertakings achieve such renown that their name gives rise to a verb. This alone attests to the importance that Google has acquired in everyday life.

Is the omnipresence of the Google search engine in the daily life of the majority of the world’s population, especially in the context of the use of smart mobile devices running its operating system (‘OS’) Android, from a legal point of view (also) due to the abuse of a dominant position?

This is, in essence, the question underlying the present appeal.

This question arises against the factual background of the conditions which, inter alia, the manufacturers of smart mobile devices were required to satisfy in order to obtain licences for Google’s most important applications (‘apps’) Play Store, Google Search and Chrome.

In Decision C(2018) 4761 final, (3) largely confirmed by the General Court, the European Commission considered that those conditions comprised several infringements of the prohibition of abuse of a dominant market position under Article 102 TFEU.

In view of the disputed findings of the General Court, the Court of Justice will be called upon to deal with, inter alia, the question of whether the proof of such abuse requires a counterfactual analysis of the state of competition without the contested behaviour as well as an analysis of the capacity of this behaviour to produce an exclusionary effect on undertakings considered to be as efficient as the dominant undertaking.

The recent case-law of the Court of Justice, especially in Servizio Elettrico Nazionale and Others, (4)Unilever Italia Mkt. Operations, (5)European Superleague Company, (6)Google and Alphabet v Commission (Google Shopping) (7) and Commission v Intel Corporation, (8) already provides considerable guidance for answering those questions.

II.Background to the dispute

The General Court set out the background to the dispute in paragraphs 2 to 72 of the judgment under appeal which, for the purposes of these appeal proceedings, can be summarised as follows.

A.Facts

Google LLC is a company established in Mountain View, California (United States), specialising in internet-related products and services and operating, inter alia, within the European Economic Area (EEA). Google is a subsidiary of Alphabet Inc. (together, ‘Google’ or ‘the appellants’).

Google derives most of its revenue from its flagship product, its search engine Google Search. Google’s business model is based on the interaction between, on the one hand, internet-related products and services offered for the most part free of charge to users and, on the other hand, online advertising services from which it derives the vast majority of its revenues. Those revenues are mostly generated by Google Search. Google’s business model is thus based first and foremost on increasing the numbers of users of its online search services so that it can sell its online advertising services. Furthermore, when users interact with its products and services, Google collects data that allow it to strengthen its ability to present relevant search responses and advertisements.

Google’s business model was initially developed in a personal computer (PC) environment where the web browser was the core entry point to the internet. Subsequently, however, Google wanted to take account of the development of the mobile internet and of the likely change in user behaviour to which that would lead as regards general searches performed online, in particular in view of the opportunities offered by geolocation. Against that background, Google developed a strategy to ensure that users would also perform their searches on mobile devices via Google Search.

In that context, in 2005 Google acquired the undertaking that had originally developed the Android OS for smart mobile devices. According to the Commission, in July 2018 approximately 80% of smart mobile devices used in Europe and worldwide were running Android.

When Google develops a new version of Android, it publishes the source code online under an open-source licence (Android Open Source Project licence; ‘the AOSP licence’) free of charge. This allows third parties to download and modify that code to create modified versions of Android, which are known as ‘forks’ (a fork being new software created from the source code of existing software).

The Android OS is part of an ‘ecosystem’ incorporating other elements such as all Google Mobile services (GMS bundle or Google Mobile Services; ‘the GMS suite’), which include, inter alia, the app store Play Store, the Google Search app and the Chrome browser.

The freely available Android source code covers the basic features of an OS, but not the Android apps and services owned by Google. Original equipment manufacturers (‘OEMs’) that wish to obtain Google apps and services must therefore enter into agreements with Google. Google also enters into such agreements with mobile network operators (‘MNOs’) that wish to be able to install Google’s proprietary apps and services on devices sold to end users.

Some of those agreements form the subject matter of the present case.

B.The decision at issue

On 18 July 2018, the Commission adopted the decision at issue, in which it imposed a fine on Google LLC and, in part, on Alphabet Inc. for an infringement of Article 102 TFEU and Article 54 of the EEA Agreement.

First of all, the Commission identified four types of relevant market, that is, (i) the worldwide market (excluding China) for licensable OSs for smart mobile devices, (ii) the worldwide market (excluding China) for Android app stores, (iii) the various national markets, within the EEA, for the provision of general search services, and (iv) the worldwide market for non-OS-specific internet browsers designed for mobile use. (9) According to the Commission, during the relevant period, Google held a dominant position on the first three of those markets. (10) Those findings are no longer disputed in the present appeal proceedings.

In the decision at issue, the Commission found that Google had abused its dominant position on those markets by imposing anticompetitive contractual restrictions on OEMs and MNOs (‘the restrictions at issue’). The Commission identified four distinct infringements that amount to three sets of contractual restrictions.

The first restrictions were contained in the Mobile Application Distribution Agreements (‘MADAs’) and constituted, according to the Commission, two instances of tying (first and second infringements): (11)

–on the one hand, the requirement that OEMs pre-install the Google Search app in order for them to be able to obtain a licence to use the app store Play Store (first bundle); by imposing this, Google abused its dominant position on the worldwide market (excluding China) for Android app stores from 1 January 2011 to the date of the decision at issue;

–on the other hand, the requirement that OEMs pre-install the Chrome browser in order for them to be able to obtain a licence to use the Play Store and Google Search (second bundle); by imposing this, Google abused its dominant position on the worldwide market (excluding China) for Android app stores from 1 August 2012 to the date of the decision at issue.

As the General Court explained, those two product bundles are complementary. Thus, the Chrome-Play Store and Google Search bundle came to overlap with the Google Search-Play Store bundle to take account of changes in the MADA, which did not initially include the Chrome browser among the apps included in the GMS suite. According to the Commission, the objective of the two bundles was to enable Google to reach users, so that they would carry out their general searches using Google Search, either as a general search app or as the Chrome browser’s search engine. (12)

Second, in order to obtain a licence for the Play Store and Google Search, the OEMs had to enter into an Anti-Fragmentation Agreement (‘AFA’), requiring them not to sell devices running versions of Android that were not approved by Google (third infringement). Accordingly, entering into a MADA was possible only after having concluded an AFA. By requiring this, Google had abused its dominant position on the markets for Android app stores and for general search services from 1 January 2011 to the date of the decision at issue. (13)

Third, the Revenue Share Agreements (‘RSAs’) contained restrictions under which Google granted OEMs and MNOs a percentage of its advertising revenue, provided that those manufacturers or operators had agreed not to pre-install a competing general search service on any device within an agreed portfolio (‘portfolio-based RSAs’) (fourth infringement). By requiring this, Google had abused its dominant position on the national markets for general search services from 1 January 2011 to 31 March 2014. (14) Later, Google entered into device-based RSAs, pursuant to which the payment of revenue shares by Google was conditional on OEMs and MNOs not pre-installing any competing general search services on a given device for which revenue shares were paid. Those device-based RSAs were not found to be unlawful by the Commission.

According to the Commission, the objective of the restrictions at issue was to protect and consolidate Google’s dominant position on the national markets, within the EEA, for general search services and, therefore, the revenue obtained by Google through search advertisements. The common objective and the interdependence of the restrictions at issue led the Commission to classify them as a single and continuous infringement of Article 102 TFEU and Article 54 of the EEA Agreement.

In order to penalise what it considered to be abusive practices, the Commission imposed a fine of EUR 4 342 865 000 on Google LLC, of which EUR 1 921 666 000 jointly and severally with Alphabet Inc. Furthermore, the Commission required Google to bring to an end the contested behaviour within 90 days of notification of the decision at issue.

C.Procedure before the General Court and the judgment under appeal

1.Procedure before the General Court

By application lodged at the Registry of the General Court on 9 October 2018, Google brought an action against the decision at issue. Before the General Court, the form of order sought by Google was supported by the Application Developers Alliance (‘ADA’), the Computer & Communications Industry Association (‘CCIA’), Gigaset Communications GmbH (‘Gigaset’), HMD global Oy (‘HMD’) and Opera Norway AS, formerly Opera Software AS (‘Opera’).

The form of order sought by the Commission was supported by BDZV – Bundesverband Digitalpublisher und Zeitungsverleger eV (‘BDZV’), Bureau européen des unions de consommateurs BEUC (‘BEUC’), FairSearch AISBL, Qwant, Seznam.cz, a.s. (‘Seznam’) and Verband Deutscher Zeitschriftenverleger eV (‘VDZ’).

The General Court examined the case on the basis of a consolidated non-confidential version of the file. In the judgment under appeal, certain confidential data known to the main parties were replaced by the ranges used in the public version of the decision at issue available on the Commission’s website. (15)

2.The judgement under appeal

29.In its judgment of 14 September 2022, the General Court annulled the decision at issue in so far as it concerned the portfolio-based RSAs, that is, the fourth element of the single and continuous infringement.

30.Furthermore, the General Court set the amount of the fine imposed on Google LLC at EUR 4 125 000 000, with Alphabet Inc. jointly and severally liable for EUR 1 520 605 895.

31.The General Court dismissed the action as to the remainder and ordered all the parties to bear their own costs.

III.Procedure before the Court of Justice and forms of order sought

32.By application of 30 November 2022, Google lodged an appeal against the judgment under appeal.

33.Just as before the General Court, Google is supported in the appeal by ADA, CCIA, Gigaset, HMD and Opera.

34.The Commission is supported in the appeal, again just as before the General Court, by BDZV, BEUC, FairSearch, Qwant, Seznam and VDZ.

35.By orders of 19 January (16) and 18 April 2023, (17) the President of the Court of Justice ordered, at the request of the appellants, that the decision at issue as well as certain information contained in footnote 98 to the Commission’s response should be treated as confidential vis-à-vis the interveners. The elements concerned had previously been treated as confidential vis-à-vis those same parties in the proceedings at first instance.

36.Google claims that the Court of Justice should:

–set aside the judgment under appeal;

–annul the decision at issue;

–in the alternative, remand the case to the General Court;

–in the further alternative, set aside point 2 of the operative part of the judgment under appeal and fix the amount of the fine imposed in Article 2 of the decision at issue at a significantly lower amount; and

–order the Commission to pay Google’s costs and expenses in connection with the present proceedings and the proceedings before the General Court.

37.ADA, CCIA, Gigaset, HMD and Opera claim that the Court of Justice should grant the form of order sought by Google and order the Commission to pay the costs.

38.The Commission contends that the Court of Justice should:

–dismiss the appeal; and

–order Google to pay the costs.

39.BEUC, FairSearch and Seznam also contend that the Court of Justice should dismiss the appeal and order Google to pay the costs.

40.The parties presented oral argument and answered the questions put to them by the Court of Justice at the hearing which took place on 28 January 2025.

IV.Assessment

41.In support of its appeal, Google relies on six grounds, the first five of which concern the finding of the infringement (A), while the sixth criticises the way in which the General Court exercised its unlimited jurisdiction to vary the fine (B).

A.The finding of the infringement (first to fifth grounds of appeal)

42.As set out in points 19 to 23 of this Opinion, the single and continuous infringement found by the Commission in the decision at issue comprised four distinct infringements that amount to three sets of contractual restrictions:

–the MADA pre-installation conditions, according to which a licence for the Play Store could be obtained only if Google Search was pre-installed (first bundle) and a licence for Play Store and Google Search could be obtained only if Chrome was pre-installed (second bundle);

–the AFAs, according to which a licence for the Play Store and Google Search could be obtained only if the OEMs did not sell devices running versions of Android that were not approved by Google;

–and finally the portfolio-based RSAs, according to which a percentage of Google’s advertising revenue could be obtained if no competing search services were pre-installed.

43.The finding that the portfolio-based RSAs were restrictive of competition was annulled by the General Court in the judgment under appeal and is not contested by the Commission in the present appeal. Thus, the findings of the General Court in that regard are definitive.

44.By its first and second grounds of appeal, Google claims that the General Court committed errors of law when assessing the MADAs (1). The third and fourth grounds of appeal concern the characterisation of the AFAs (2). Finally, by its fifth ground of appeal, Google criticises the finding of a single and continuous infringement (3).

1.The assessment of the MADAs (first and second grounds of appeal)

45.By its first ground of appeal, Google challenges the assessment of the causal link between the pre-installation conditions of the MADAs and their alleged exclusionary effects (a). In the context of its second ground of appeal, it claims that the General Court erroneously confirmed the decision at issue although the Commission had not established the MADAs’ capacity to produce an exclusionary effect on undertakings considered to be as efficient as Google itself (b).

(a)The causal link between the MADAs and their exclusionary effects (first, third and fourth parts of the first ground of appeal)

46.According to Google’s first ground of appeal, the General Court erred in assessing the causal link between the pre-installation conditions of the MADAs and their capacity to foreclose competition.

47.In the context of this ground of appeal, Google submits four arguments, the second of which largely overlaps with the line of argument put forward by Google in its second ground of appeal. Thus, I will address that second argument in point 99 et seq. of this Opinion in the context of the second ground of appeal.

48.Within the remaining three arguments put forward in its first ground of appeal, Google contends, first, that the General Court erred in law when it took the RSAs into account as an element of context (1). Second, it erred in law when it claimed that it was not necessary to distinguish the effects of default setting from those of pre-installation (2). Finally, it did not take into account the competition (or the lack thereof) that would have existed without the MADA pre-installation conditions (3).

49.According to the Commission, the first ground of appeal is ineffective because it is based on the erroneous premiss that a counterfactual analysis is the only way to establish whether conduct is capable of restricting competition. However, such a proposition finds no support in the case-law of the Court of Justice. A counterfactual analysis is merely one way, but not the only way, to establish whether conduct is capable of restricting competition.

50.It is true that, according to the case-law, the actual or potential effects of practices examined can be established by reference to a range of evidence, without the Commission being required systematically to use any single tool to prove the existence of such a causal link. (18)

51.However, the arguments put forward by Google in the context of its first ground of appeal concern not only the assessment of the need for the General Court to conduct a counterfactual analysis, but also certain aspects taken into account by that court when assessing the abusive nature of the MADAs. It follows that those arguments are capable of calling into question the General Court’s findings underlying the operative part of the judgment under appeal and of leading, if they are well founded, to that judgment being set aside. (19)

52.Hence, the Commission’s claim of ineffectiveness must be dismissed.

(1)Taking into account the RSAs when analysing the MADAs (first part of the first ground of appeal)

53.By the first part of its first ground of appeal, Google contends that the General Court erred in law when taking into account, for the assessment of the effects of the MADAs, the effects of the device-based RSAs, which were not contested by the Commission, as well as the effects of the portfolio-based RSAs, the abusive character of which was not, according to the General Court, established. By doing so, the General Court failed to prove that the alleged exclusionary effects were caused by the MADA pre-installation conditions.

54.This line of argument raises, in essence, the question as to whether a practice that has been found to be abusive by the Commission, but not by the General Court, can nevertheless be taken into account as an element of context when analysing the effects of the other components of a single and continuous infringement, which have been found to be abusive.

55.As set out in points 23 and 42 of this Opinion, Google’s single and continuous infringement comprised, according to the Commission, the portfolio-based RSAs. Those agreements granted OEMs and MNOs a percentage of Google’s advertising revenue, provided that those manufacturers or operators had agreed not to pre-install a competing general search service on any device within an agreed portfolio. Accordingly, if the OEM or MNO concerned pre-installed a competing general search service on any device within a predefined and agreed portfolio, it had to forgo any share of the revenue in respect of the entire portfolio. (20)

56.According to the decision at issue, those portfolio-based RSAs implemented exclusivity payments, whose objective was to ensure that Google had exclusive rights to pre-install general search service apps on mobile devices. (21) In order to reach this conclusion, the Commission carried out, inter alia, an analysis of the capacity of the contested practice to foreclose as-efficient competitors, a test known as the ‘as-efficient competitor test’ (‘the AEC test’). (22)

57.However, the General Court reached the conclusion that the finding of the abusive nature of the portfolio-based RSAs was not proved to the requisite legal standard because of errors committed by the Commission when assessing their market coverage and implementing the AEC test. (23) As already set out in points 29 and 43 of this Opinion, the General Court thus annulled the decision at issue as far as the abusive nature of the portfolio-based RSAs was concerned. The Commission has not contested that finding in the present appeal. Thus, the conclusion that it did not establish the abusive nature of the portfolio-based RSAs is legally binding.

58.From 2013 onwards, Google gradually replaced portfolio-based RSAs with device-based RSAs. Under a device-based RSA, an OEM’s or MNO’s share of Google’s revenues depended on the number of devices sold that complied with the obligation not to pre-install competing general search services. Thus, the device-based RSAs enabled an OEM or an MNO to offer, for the same type of device, some that promoted Google’s general search service exclusively and others that also offered competing general search services. The Commission did not find those device-based RSAs to be abusive. (24)

59.Although, according to the findings of the Commission and the General Court, neither the device-based nor the portfolio-based RSAs were abusive when taken in isolation, the General Court took them into account when assessing the MADAs. (25) According to the General Court, this was justified because the decision at issue was correct in taking into account the device- and portfolio-based RSAs as aspects of the factual context for the purpose of assessing the exclusionary effects caused by the other aspects of the single and continuous infringement and thus also the MADAs. The General Court pointed out in particular that, irrespective of the characterisation of the RSAs in terms of competition law, the combined effects of the practices established by Google gave it the benefit, as regards Google Search, of exclusive pre-installation covering, at least until 2016, more than half of all devices marketed in the EEA running an Android-based OS. (26)

60.Before I examine whether this approach is well founded, it must be noted that, according to the established case-law of the Court of Justice, the finding that a practice constitutes an abuse of a dominant position in the sense of Article 102 TFEU must always be assessed having regard to all the relevant facts of the specific case. (27) On the other hand, a competition authority cannot rely on the effects that a practice might have produced if certain specific circumstances – which were not prevailing on the market at the time when that practice was implemented – had arisen. (28)

61.In that context, the taking into account of a factual element does not, as the Court of Justice made clear, for example, in the judgment in AstraZeneca v Commission, depend on whether this element amounts to a behaviour that in itself is characterised as abusive. (29) Likewise, the Court of Justice explained in the judgment in Sony Optiarc and Sony Optiarc America v Commission that, when a complex of practices is characterised as a single and continuous infringement, it cannot be deduced therefrom that each of those forms of conduct must, in itself and taken in isolation, necessarily be characterised as a separate infringement. (30) In the judgment in Google and Alphabet v Commission (Google Shopping), the Court of Justice found that two practices that were lawful when analysed in themselves and taken in isolation were found to be abusive because in their combination they deviated from competition on the merits and had (potential) anticompetitive effects. (31)

62.In the present case, the General Court reached the conclusion that the combined effects of the MADAs and the RSAs had to be taken into account because of the complementary nature of Google’s various practices. The result of the RSA-linked contractual obligation not to install anything other than Google Search for general searches was that the theoretical possibility of pre-installing a service competing with Google’s apps, although permitted in principle by the MADAs, was in fact excluded, from 2011 to 2016, for at least half of the Google Android devices sold in the EEA. In other words, the RSAs guaranteed to Google that only its own search service was installed on the devices concerned, which must be taken into account when assessing the anticompetitive effects of the MADAs (especially the question of whether competing search services could offset Google’s advantage by way of pre-installation alongside its own service). (32)

63.In that context, the RSAs were only one of the elements taken into account by the General Court in order to assess whether the competing search services could offset the advantage conferred on Google by the MADAs. (33) Hence, and contrary to what Google claims, the General Court cannot be criticised for not having considered that the RSAs did not cover all devices covered by the MADAs when it took into account the former for the evaluation of the latter.

64.Nevertheless, Google and Gigaset contend that the anticompetitive effects of the MADAs are not established because, in a counterfactual scenario without those agreements, the RSAs would have had the same exclusionary effects to the detriment of competing search services.

65.However, it is simply not possible to ignore the MADAs and to analyse the effects of the RSAs in isolation. This is because the OEM concerned had pre-installed the search service Google Search, which was the object of the RSAs, on a great number (34) of devices thanks to the MADAs in order to obtain access to the Play Store, which was indispensable for them and for which Google held a dominant position.

66.Because of this interplay of the MADAs and the RSAs, the other providers of general search services did not start from the same position as Google as regards the conclusion of an RSA. This is because Google’s search service Google Search was already pre-installed from the outset on a great number of smart mobile devices, whereas the other providers first had to make an effort to obtain pre-installation of their service alongside that of Google. On the other hand, because of the MADAs, no other provider of search services could obtain exclusivity on the devices concerned.

67.Contrary to what Google and Gigaset claimed during the hearing, the findings of the Court of Justice in its judgment in Google and Alphabet v Commission (Google Shopping) concerning the combined effects of such intertwined practices are also relevant here, despite the differences between the two cases. It is true that, in that case, the Commission did not contest the two practices at issue in isolation, but only in the combination of practices, whereas, in the present case, it found that both the MADAs and the portfolio-based RSAs were abusive. However, this does not mean that the effects of both practices must be analysed in isolation here, although, as a matter of fact, they emerged together.

Artificially dividing two so inextricably intertwined practices in order to analyse their effects would not be realistic, nor would it take into account the combined effects of both parts. (35) As the General Court pointed out, a distinction must be made in this respect between theoretical competition assumptions and the practical reality. (36) This is all the more true because, as the Commission explained during the hearing, in markets like those at issue here, which are characterised by dynamic network effects, it can be practically impossible to analyse how each contextual element in itself and taken in isolation specifically contributed to the exclusionary effects.

It follows that the General Court did not err in law when it found that, independently of their own characterisation as abusive or not, the RSAs could be taken into account in the analysis of the effects of the MADAs. Hence, the first part of the first ground of appeal must be dismissed.

(2)The distinction between default setting and pre-installation (third part of the first ground of appeal)

In the context of the third part of its first ground of appeal, Google contends that the General Court wrongly found that the evidence relating to default setting was relevant to the analysis of the MADA pre-installation conditions. However, only the latter (and not the MADAs as a whole, including the rules on default setting (37)) had been found to be abusive in the decision at issue. In so doing, the General Court validated the Commission’s analysis of the MADAs without requiring it to prove the causal link between the pre-installation conditions and their alleged exclusionary effects.

Default setting refers to the setting of a service as the default in a given app. Pre-installation signifies that apps are pre-loaded onto devices at the configuration stage before they are marketed. (38)

The General Court assessed Google’s arguments relating to the distinction between pre-installation and default setting in paragraphs 320 to 418 of the judgment under appeal. In those paragraphs, it analysed Google’s criticism that the Commission had not established in the decision at issue that the pre-installation conditions gave rise to a ‘status quo bias’. As set out, for example, in paragraphs 321 and 331 of the judgment under appeal, that term signifies that users tend to use what is already pre-installed on their device and is offered to them.

In this context, the General Court found, first, in paragraphs 327 to 335 of the judgment under appeal, that the distinction suggested by Google between default setting and pre-installation lacked practical relevance. It was not easy to make such a distinction, because both Google and other actors used the term ‘default’ in a broad sense and conflated the concepts of default setting and pre-installation. Furthermore, it was not disputed that the pre-installation of an app in itself confers an advantage over competing apps. Lastly, the MADA pre-installation conditions also included provisions concerning premium placement or default setting.

Against that background, the General Court found that the evidence criticised by Google could be relied on in order to establish the existence of a general tendency to ‘freeze’ the situation, regardless of whether, strictly speaking, the evidence concerned default setting or pre-installation or premium placement. Consequently, it was not necessary at the outset, for the purpose of establishing whether there was a ‘status quo bias’, to distinguish precisely, as Google wished, the effects of default setting from the effects of pre-installation.

On the basis of that premiss, the General Court assessed, in paragraphs 340 to 394 of the judgment under appeal, the evidence relied on in the decision at issue stemming from Google itself, HP, Nokia, Amazon and Mozilla, an analysis from Yandex as well as the pre-installation agreement between Microsoft and Verizon; in paragraphs 395 to 408, certain comparisons made in the decision at issue using a FairSearch study, data provided by Microsoft and NetMarketShare data as well as the comparison of Google’s revenues from Android and iOS devices; and, in paragraphs 409 to 418, certain matters relating to Chrome by reference to the comparison of Google’s revenues via Safari and via Chrome, and an Opera survey.

In paragraph 418 of the judgment under appeal, the General Court concluded that the various arguments put forward by Google to refute the advantage conferred by the pre-installation of the Google Search and Chrome apps on Google Android devices could not call into question the conclusions drawn by the Commission from the various points set out in the decision at issue in that regard.

In criticising those findings, Google contends in its appeal that the General Court went beyond the findings of the decision at issue when it found, in paragraph 349 of the judgment under appeal, that the arguments assessed therein were valid in the context of both terms at issue. However, in Google’s view, the General Court’s reasons, set out in paragraphs 329 to 334 of the judgment under appeal (recapitulated in points 73 and 74 of this Opinion), for it not being necessary to distinguish precisely between the two terms are not convincing. Against that background, the General Court’s assessment described in point 75 of this Opinion is erroneous.

Google’s argument is partly inadmissible as it is aimed at securing a new assessment of the facts by the Court of Justice (i). On the other hand, the argument cannot succeed as it is contrary to the principles set out above according to which the anticompetitive effects of a practice must be assessed in its context (ii).

(i)No new assessment of the facts on appeal

On the one hand, Google’s arguments are inadmissible in so far as Google seeks, in reality, to obtain a fresh assessment of the evidence produced before the General Court. However, that court has exclusive jurisdiction to assess the evidence, (39) save where the evidence adduced before it has been distorted. Such a distortion must be substantiated by the appellant and obvious from the documents in the file, without there being any need to carry out a new assessment of the facts and the evidence. (40)

However, in its criticism recapitulated in point 77 of this Opinion, Google does not contend that the General Court distorted the evidence assessed by it, and nor is such a distortion obvious from the file. Rather, the question of whether the facts and the evidence concerning default setting are also relevant to pre-installation and how the different actors conflated those terms are matters that fall within an assessment of the facts and the evidence. Such an assessment is different from the review of a distortion which aims to examine whether the General Court misconstrued the evidence in a manner manifestly at odds with its wording. (41) In that respect, Google has also failed to point to any breach of the rules relating to the burden of proof and the taking of evidence that would be subject to the jurisdiction of the Court of Justice on appeal.

(ii)No assessment of the effects of a practice isolated from its context

On the other hand, Google’s arguments concern the question of whether the Commission had an obligation to assess the MADA pre-installation conditions in isolation from their context, in particular the obligations on default setting and premium placement. This question relates to the legal characterisation of those facts by the General Court and the legal conclusions which it has drawn from them, in respect of which the Court of Justice has jurisdiction upon appeal. (42)

However, as already set out in points 60, 61, 67 and 68 of this Opinion in the context of the first part of this ground of appeal, Google’s premiss that, in order to establish the anticompetitive effects of a practice, it must be analysed in isolation from its context is erroneous. As explained in those points, the characterisation of a practice as an abuse of a dominant position in the sense of Article 102 TFEU must always be assessed having regard to all of the relevant facts of the specific case. In so doing, the taking into account of a factual element does not depend on whether it derives from a practice that is in itself characterised as abusive. This is especially the case when different practices – both abusive and legal, as the case may be – are so closely intertwined that their effects unfold in combination with each other, and cannot be analysed in isolation from each other.

In view of this as well as of the autonomous assessment of the evidence by the General Court, which has not been validly called into question by Google, that court was right not to require the Commission to distinguish precisely between pre-installation and default setting when assessing the evidence in order to analyse whether the MADA pre-installation conditions created a ‘status quo bias’.

It follows that the third part of the first ground of appeal must also be dismissed.

(3)The alleged pro-competitive component of the MADAs and the requirement of a counterfactual assessment (fourth part of the first ground of appeal)

By the fourth part of the first ground of appeal, Google criticises the fact that the General Court did not require the Commission to carry out a counterfactual analysis to assess whether an open licence model like that of Android, the pro-competitive component of which it had acknowledged, would have been possible without the contested rules of the MADAs.

Google submits that throughout the administrative investigation and before the General Court, it had explained that the MADA pre-installation conditions were the non-monetary consideration that it received from OEMs in exchange for providing the Android OS, the Play Store, the GMS suite and other related services free of charge. This non-monetary exchange enabled more distribution opportunities for rivals and more competition than any realistic alternative licensing arrangement.

Google criticises the fact that the Commission had banned this non-monetary exchange enabled by the MADA pre-installation conditions without ever examining whether any realistic alternatives would have resulted in equal or greater distribution opportunities for rivals.

According to Google, when endorsing that approach, the General Court erroneously assessed the question of whether the MADA pre-installation conditions had anticompetitive effects. Indeed, this would have required an analysis of how competition would have unfolded without the contested behaviour. Here, this would have meant examining whether an open and free-of-charge licensing system with all its advantages would have been realistic without the non-monetary exchange implied by the MADA pre-installation conditions.

In paragraphs 590 to 594 of the judgment under appeal, contested by Google in that context, the General Court examined Google’s criticism that the Commission had not taken into account all relevant circumstances for the assessment of the alleged effects of the behaviour at issue.

By contrast, Google does not contest paragraphs 599 to 619 of the judgment under appeal, which deal with its arguments concerning the objective justification of the MADAs.

In paragraphs 590 to 594 of the judgment under appeal, the General Court explained that the Commission had dealt with Google’s arguments in the decision at issue but had dismissed those arguments because the behaviour considered to be abusive did not relate to the development and maintenance of the Android platform nor to the fact that it is open and free of charge. Rather, only one aspect of the MADAs was concerned, the effects of which restricted competition, that is, the pre-installation conditions. Those conditions conferred a competitive advantage on Google because of the ‘status quo bias’ created by them that could not be offset by competitors and restricted competition on the merits to the detriment of consumers.

Contrary to what Google contends on appeal, the General Court did not err when, in the light of those findings, it did not require the Commission to examine in the context of a counterfactual analysis how competition could have unfolded on the relevant markets without the conditions at issue.

It is true that the Commission has to establish the causal link between the contested abusive behaviour and its anticompetitive effects. However, in that context, it is permissible for the Commission to rely on a range of evidence. It is not required systematically to use any single tool, especially a counterfactual analysis, to prove the existence of such a causal link. (43) This is particularly true when such an analysis is not required in the circumstances of the actual case to establish the (especially potential) effects of the behaviour at issue.

In order to establish the abuse of a dominant position it is sufficient to prove that the behaviour at issue is capable of producing exclusionary effects. It is not, however, required that its actual effects be proved. Hence, the circumstance that – unlike its competitors – only the dominant undertaking was able to influence the conduct of customers or users in a discriminatory way to the detriment of those competitors may be sufficient to establish that the behaviour at issue was capable of impairing effective, undistorted competition. (44)

As set out in point 91 of this Opinion, the General Court found that the MADA pre-installation conditions conferred a competitive advantage on Google that derived from the ‘status quo bias’ stemming from those conditions and which could not be offset by competitors.

Subject to the examination of the second part of the first ground of appeal as well as the second ground of appeal in point 99 et seq. of this Opinion, the General Court thus established that the users’ decision to use Google Search and Chrome rather than competing apps had been influenced in a discriminatory way by the ‘status quo bias’ which competitors could not offset. In those circumstances, it was right not to require a further counterfactual analysis.

Furthermore, contrary to what Google contends, the Commission cannot be required to reflect, in the context of such an analysis, on how the dominant undertaking might have behaved if it had not adopted the contested abusive conduct. The counterfactual scenario, although hypothetical, must nevertheless be realistic and credible. (45) On the one hand, in a case such as the present one, the functioning of the relevant markets is characterised by variables such as innovation, access to data, multi-sidedness, user behaviour and network effects. (46) On the other hand, this functioning and, in particular, the behaviour of users have already been substantially influenced by Google’s strategies. Thus, it does not appear to be possible to establish realistic or plausible prognostics of options that Google could have chosen instead of the MADAs, or how those options would have influenced the relevant markets.

It follows that the fourth part of the first ground of appeal must also be dismissed.

(b)MADAs’ capacity to foreclose as-efficient competitors (second part of the first ground of appeal and second ground of appeal)

In the context of the second part of its first ground of appeal, Google contends that the General Court committed an error of law when it confirmed the Commission’s assessment that the finding that users did not frequently download competing search and browser apps was sufficient in order to establish the anticompetitive effects of the MADAs’ pre-installation conditions. However, both the Commission and the General Court had acknowledged that there had not been any technical or financial constraints on users downloading rival services. Thus, they should have examined whether the decision of users not to download rival apps more frequently was due to preference and the superior quality of Google’s apps rather than the ‘status quo bias’ allegedly created by the pre-installation.

By its second ground of appeal, Google claims that the General Court wrongly confirmed the decision at issue although it had not established MADAs’ capacity to foreclose as-efficient competitors.

As the arguments put forward by Google in these two parts of its appeal partially overlap, I shall address them together.

First, it must be examined whether, in the light of the circumstances of the present case, the General Court had to require the Commission to expose not only MADAs’ capacity to restrict competition, but also their particular capacity to foreclose competitors as efficient as Google. In that context, I shall first assess Google’s argument criticising the General Court for failing to examine whether the decision of users not to download competing search and browser apps more frequently was due to users’ preferences rather than to the MADA pre-installation conditions (1). I shall then examine Google’s specific arguments put forward in the context of its second ground of appeal concerning the tying of the Google Search app and the Play Store (2) and the tying of Chrome with the Play Store and Google Search (3).

(1)The assessment of the exclusionary effects of the MADAs as regards ‘as-efficient competitors’ and the reasons for users’ behaviour

According to the Commission, Google’s arguments cannot succeed as they are based on the premiss that the establishment of a restriction of competition requires in all cases an analysis of the efficiency of (hypothetical or actual) competitors. In instances of tying such as those at issue here, which fall outside the scope of competition on the merits, there is no need to assess whether they are capable of producing exclusionary effects on as-efficient competitors, but only whether they are capable of restricting competition.

It is common ground that Article 102 TFEU does not aim to protect less-efficient competitors. (47) Does that mean, a contrario, that in a case such as the present one the establishment of the abuse of a dominant position always requires an examination as to whether the practice at issue is capable of producing exclusionary effects on as-efficient competitors? And, if so, how should such an assessment be carried out?

In order to find an answer to these questions it is useful, first, to recall the starting point of the case-law on tying (i). Second, it is necessary to recapitulate the General Court’s responses to Google’s criticism that the Commission had not established that the MADA pre-installation conditions had restricted competition (ii). Finally, against that background, Google’s criticism concerning the General Court’s proceeding must be assessed (iii).

(i)Starting point of the case-law on tying

The starting point of the case-law on instances of tying by a dominant undertaking is that they are in principle to be classified as abusive. (48)

) A practice consisting of making the acquisition of a tying product – stemming from a dominated market – dependent upon the acquisition of a tied product, stemming from a neighbouring or up- or downstream market, does indeed fall outside the scope of competition on the merits. This is because customers generally do not choose the tied product because of its intrinsic value or free choice, but buy it only to obtain the tying product that is important for them. Indeed, from the point of view of those customers, in the light of the dominant position of the provider as well as the already weakened competition on the relevant market, there are often no acceptable alternative products. This suggests that such behaviour of an undertaking that holds a dominant position on the market for the tying product distorts competition on the market for the tied product. As efficient as the (actual or potential) competitors of the dominant undertaking on the latter market may be, they can hardly offset the leverage effect arising from the tying of the tying product from the other market. Accordingly, in such a case, at least potentially anticompetitive effects of the behaviour at issue can be presumed. (49)

However, in its (definitive) judgment in Microsoft v Commission, the General Court found, in a case similar to the present one, that the Commission had been right, in the light of the particular circumstances of that case, not to restrict itself to establishing that the behaviour at issue was an instance of tying, but to examine further whether it had been capable of restricting competition. Indeed, although Microsoft made the acquisition of its Windows-OS dependent on the simultaneous acquisition of the integrated app Media Player, users were theoretically able to obtain third-party media players through subsequent downloading. Nevertheless, the General Court confirmed the Commission’s finding that, especially because of users’ inertia, this remained a largely theoretical possibility and that the competing media players, irrespective of their quality, could not offset the competitive advantage obtained by Microsoft thanks to the pre-installation. (50)

That is the approach followed by the Commission in the present case.

(ii)Findngs of the General Court regarding the MADAs’ exclusionary effects

As the General Court pointed out in paragraph 292 of the judgment under appeal, in the present case, the parties were not arguing about the possibility of users being able, under the MADAs, to obtain general search and browser apps competing with those that were subject to tying. Indeed, they all recognised the theoretical existence of this possibility which was not forbidden by MADA rules. Rather, they argued about the incentives users might have to do so.

In those circumstances, the General Court held in paragraph 295 of the judgment under appeal that the Commission was right to argue that examination of the effects of the instances of tying at issue was required before it could be concluded that the tying at issue was harmful to competition.

As the General Court recapitulated in paragraphs 304 to 312 of the judgment under appeal, the Commission reached the conclusion that the competing search and browser apps could not offset the advantage conferred on Google by the tying of Google Search and the Play Store as well as of Chrome, the Play Store and Google Search.

In paragraphs 317 to 596 of the judgment under appeal, the General Court dismissed, in the context of the assessment of the second plea, Google’s argument that the Commission had not established that the MADA pre-installation conditions restricted competition. In so doing, the General Court examined Google’s criticism relating to five points, that is, the creation of a ‘status quo bias’ by the pre-installation (paragraphs 320 to 418), the OEM’s possibility of pre-installing and default setting competing general search services (paragraphs 419 to 538), competitors’ other means of reaching users (paragraphs 539 to 567), the causality between Google’s usage shares and pre-installation (paragraphs 568 to 584) and, finally, consideration of the entire context (paragraphs 585 to 596).

In that context, the General Court dealt several times with arguments pointing out the allegedly higher quality of Google’s services as a reason for users’ behaviour and Google’s market shares. In the light of those considerations of the General Court, it is clear why, in its view, no further examination of the MADAs’ effects, especially in view of possible exclusionary effects on Google’s as-efficient competitors, was necessary. It is useful for the purposes of further examination to recapitulate briefly the relevant findings of the judgment under appeal.

–Creation of a ‘status quo bias’ by way of pre-installation

In paragraph 331 of the judgment under appeal, the General Court pointed out that Google acknowledged that the pre-installation of an app in itself confers an advantage over competing apps because the app is available on the device when it is first used and does not need to be installed before use. This increases the likelihood of users trying the apps that are pre-installed.

In paragraph 418 of the judgment under appeal, the General Court found that the various arguments put forward by Google to refute the advantage conferred by the pre-installation of the Google Search and Chrome apps on Google Android devices could not call into question the conclusions drawn by the Commission as regards the creation of a ‘status quo bias’ from the various points set out in the decision at issue in that regard.

–The possibility for OEMs to pre-install competing search services and to set them as default

In paragraphs 426 to 428 of the judgment under appeal, the General Court pointed out in relation to the possibility of pre-installing other search apps that, in the light of the combined effects of the MADAs and the RSAs and in view of the market shares for Google Search and Chrome (in which Search had to be set as the default search app (51)) and their evolution, the debate on the options available for competitors to offset the competitive advantage granted by the MADA pre-installation conditions remained largely theoretical. In practice, competing app providers were not in a position to offset via pre-installation agreements the competitive advantage that Google ensured for itself through the pre-installation of Google Search and Chrome on practically all Google Android devices sold in the EEA.

After examining Google’s complaints against that background, the General Court concluded in paragraph 537 of the judgment under appeal that the Commission had been in a position to conclude that providers of competing general search services were free to provide OEMs and MNOs with the same pre-installation as that provided in respect of the Google Search app and Chrome on Google Android devices sold in the EEA. However, that did not happen for much of the infringement period and, at the very least, part of the explanation for the lack of such pre-installation lies in the combined effects of the MADAs, the RSAs and the AFAs.

In that context, the General Court dismissed, in paragraph 514 (referring to paragraphs 294 and 483, fourth indent) of the judgment under appeal, Google’s assertion that the Commission’s finding that the OEMs would not have an interest in pre-installing rival apps because most search usage would go to Google implies that those apps are less attractive. According to the General Court, the decision at issue set out the reasons why such an assumption could not be made in the present case, given the relevance of the various technical solutions proposed by Google’s competitors for users or innovation (especially specialised search services in a particular language or targeting a specific group of users).

–Means of reaching users other than pre-installation

The General Court’s conclusion concerning the possibility of reaching users especially via the downloading of competing apps and the access to competing search services through the browser, and concerning the criticism that the Commission confounded competitive advantage and anticompetitive eviction, is to be found in paragraph 567 of the judgment under appeal. After this, the Commission was justified in taking the view that, even if users remained free to download apps competing with the Google Search app and Chrome or to change the default settings, or mobile web browser developers were able to offer their apps to OEMs, that was not sufficiently the case for much of the infringement period because of the MADA pre-installation conditions.

–Connection between usage shares and pre-installation

As regards the establishment of the connection between usage shares and pre-installation, the General Court held in paragraph 575 of the judgment under appeal that, in a situation such as this, the Commission was not required to determine precisely whether Google’s usage shares could be explained not only by pre-installation – as it believes – but also, or rather, by the superior quality claimed by Google. Indeed, pre-installation was not disputed, so that all Google Android devices had the Google Search app and Chrome, whereas the impact of quality on the lack of pre-installation or downloading of a competing app was merely asserted by Google, with the evidence submitted in that regard being neither sufficient nor particularly relevant.

Furthermore, as the General Court made clear in paragraphs 577 and 578 of the judgment under appeal, even assuming that Google Search and Chrome were superior in terms of quality to the services offered by rivals, that would not have been decisive since it was not claimed that the various services offered by the rivals were not technically capable of meeting consumer needs. In addition, as was apparent from the material in the file, the needs of consumers were not necessarily met by what is qualitatively the best solution. Even assuming that Google might claim that its services represented such a solution, variables other than technical quality, such as the protection of privacy or the account taken of specific linguistic features of search queries, also played a role.

Finally, the General Court concluded in paragraph 582 of the judgment under appeal, in view of the Play Store ratings of Google’s apps and those of its competitors relied on by the Commission, that the quality assessment of the various competing services remained similar. That could therefore, according to the General Court, be taken into account in its finding that the respective quality of the various competing search and browser services was not a decisive criterion in their use, since they all offered a service capable of meeting demand.

Thus, as the General Court found in paragraph 583 of the judgment under appeal, in view of the tendency to freeze the situation associated with the MADA pre-installation conditions and in the absence of proof of the precise impact of the superior quality claimed by Google in respect of its general search and browser apps, the Commission correctly considered that Google’s usage shares corroborated the ‘status quo bias’ linked to pre-installation.

–Taking account of the context

Last but not least, the General Court concluded in paragraph 596 of the judgment under appeal that Google had not established that the Commission had not taken into account all the relevant circumstances when assessing the contested behaviour.

(iii)Assessment

Contrary to what Google claims, in view of those findings, which are not validly called into question by Google, the Commission was not obliged additionally to examine not only whether the contested pre-installation conditions were capable of restricting competition, but also whether they were capable of producing exclusionary effects, especially on competitors as efficient as Google.

Without prejudice to the question of whether instances of tying of the type described above in point 106 of this Opinion generally require proof of the existence of (potential) anticompetitive effects, it is in principle necessary, in order to demonstrate the abusive nature of an exclusionary practice, first, that that practice relied on the use of means falling outside the scope of competition on the merits and, second, that the practice was capable of producing exclusionary effects. (52)

Although both of those aspects are conditions that come under the assessment of whether a practice is abusive, (53) in certain circumstances the use by an undertaking in a dominant position of resources other than those governing competition on the merits may be sufficient to establish the existence of an abuse. (54)

The analysis of the capacity of a hypothetical, (55) as-efficient but not dominant competitor to replicate the conduct at issue was developed for cases in which such an analysis makes sense in order to examine whether that conduct must be regarded as being based on the use of means which come within the scope of normal competition and/or whether it is capable of producing anticompetitive effects. (56) That analysis, initially developed in the form of the AEC test for price-related practices, (57) can, according to the Court of Justice, be relevant for price-related as well as for non-price-related practices, depending on the specific circumstances. (58)

According to the case-law, where an undertaking provides a competition authority, during the administrative procedure, with evidence to that effect, that authority must assess that evidence and, as the case may be, state the reasons why it does not find it relevant. (59) In the same vein, when the Commission has relied on an AEC test, the General Court has to assess the arguments put forward by the undertaking in that regard. (60) However, in the present case, it does not appear that Google has put forward any such evidence in relation to the MADAs during the administrative procedure and that the Commission has not properly assessed that evidence, nor that the Commission has in itself relied on an AEC test that has not been properly taken into account by the General Court.

Moreover, the case-law has already acknowledged that the analysis of a practice in view of its effects in relation to as-efficient competitors is only one of a number of methods for assessing whether a practice is capable of producing exclusionary effects. Consequently, the competition authorities cannot be under a legal obligation to use that method in each case. (61)

Even more, it is apparent from the case-law that there may be cases in which it just does not make sense to base the analysis as to whether a practice is anticompetitive on the question of whether a hypothetical competitor could replicate that conduct. This is especially the case when such an analysis is not appropriate because the market structure makes the entry or ongoing presence of an as-efficient competitor or the replication of the conduct at issue by that competitor practically impossible.

In other words, if there cannot be a hypothetical competitor that would be as efficient as the dominant undertaking or if a competitor would not be able to compete from the outset, however efficient it might be, then it does indeed make no sense to analyse the harmfulness of the conduct of the dominant undertaking by reference to the efficiency of its (hypothetical) competitors. Such a situation may exist, for example, in the case of a market where the dominant undertaking holds a very high market share, has substantial structural advantages, or where there are high barriers to entry. (62)

As the General Court pointed out, in the present case, the relevant markets fall within the digital economy, where variables such as innovation, access to data, multi-sidedness, user behaviour or network effects play an important role. In such a digital ‘ecosystem’, which brings together several categories of supplier, customer and consumer and causes them to interact within a platform, the products or services which form part of the relevant markets that make up that ecosystem may overlap or be connected to each other on the basis of their horizontal or vertical complementarity. (63) Those markets are characterised by high barriers to entry and complex interactions that influence and determine each other. (64)

It follows that, in the present case, it is not realistic to compare the situation of Google with that of a hypothetical as-efficient competitor. Indeed, Google held a dominant position in several markets of the Android ecosystem (65) and could thus benefit from network effects that enabled it to ensure that users used Google Search. As a result, it obtained access to data that enabled it in turn to improve its service. No hypothetical as-efficient competitor could have found itself in such a situation.

To require, in those circumstances, a comparison of Google with a hypothetical as-efficient competitor would be to undermine the prohibition of the abuse of a dominant position under Article 102 TFEU. Indeed, as the Court of Justice has held, that provision also prohibits practices that have the actual or potential effect or object of impeding potentially competing undertakings at an earlier stage from even entering the market and, in so doing, preventing the growth of competition therein. Such a practice may in particular consist in the placing of obstacles to entry or the use of other blocking measures or other means different from those which govern competition on the merits. (66)

This also applies to competitors that are already active on the market but are not (yet) as efficient. Indeed, especially on markets with high barriers to entry, where competition is already weakened because of the presence of the dominant undertaking, the latter can play an important role in serving the object of maintaining competition. (67)

Finally, the Court of Justice has already made clear that there may be cases in which it is not possible to analyse whether the behaviour of customers or users is attributable to the qualitative characteristics of the services or products of the dominant undertaking. This applies when the users’ decision in favour of the dominant undertaking is already influenced by the conduct of the latter that does not fall under the scope of competition on the merits, and is thus distorted.

Accordingly, in the judgment in Servizio Elettrico Nazionale and Others, the Court of Justice explained that in a case in which it is established to the requisite legal standard that the manner in which the customers or users made their choice was biased in order to favour a dominant undertaking to the detriment of its competitors, the existence of such bias precluded the finding that this choice was attributable to the higher performance of the dominant undertaking. Indeed, the mere existence of that bias would, by definition, make it impossible to establish that there were objective reasons for the differences. In a case in which the distortion is attributable to the conduct of the dominant undertaking, the difference as regards the number of customers or users must thus be attributed to the dominant undertaking. (68)

It follows that in a case in which it is established that the customers’ behaviour has been influenced in a discriminatory manner, this is sufficient to establish that the conduct at issue is capable of producing exclusionary effects. (69) That must apply a fortiori in the case of a tying strategy such as the present one.

140.Hence, in such a case, in order to establish the anticompetitive character of the conduct at issue, it is neither necessary nor possible nor expedient to analyse whether that conduct could also have been displayed by a hypothetical as-efficient competitor.

141.Thus, in the present case, as notably BEUC convincingly pointed out at the hearing, no hypothetical competitor would have been able to be as ‘efficient’ as Google as regards general search services, because nobody could offset the advantage of pre-installation. Equally, hardly any competitor could obtain pre-installation, as nobody had a market position comparable to Google on the market for app stores like the Play Store.

142.Accordingly, the Court of Justice has made clear that a practice is to be considered as falling outside the scope of competition on the merits if it relies on the use of resources inherent to the holding of a dominant position. Indeed, no hypothetical competitor could adopt such a practice, even if it was as efficient, because it does not hold a dominant position on the relevant market. (70)

143.It follows from the findings recapitulated in points 109 to 124 of this Opinion that Google does not call into question, in a substantiated way, the notion that the manner in which users decide to use Google’s services rather than competing services has been distorted because of the advantage conferred on Google by pre-installation that could not be offset by its competitors. The pre-installation of Google’s apps, in turn, was implemented by the OEMs because this was necessary in order to obtain a licence for the indispensable Play Store, in respect of which Google held a dominant position.

144.Thus, it has been established to the requisite legal standard that the users’ choice was distorted because of the pre-installation conditions, which Google obtained by using a means that did not fall within the scope of competition on the merits. This is sufficient, on the one hand, to establish that the conditions at issue were capable of producing abusive exclusionary effects, and, on the other hand, to make it impossible to find out whether the users’ choice was (also) due to the alleged higher quality of Google’s apps.

145.Google’s claim that the General Court erred in not requesting proof that the MADAs’ exclusionary effects were not due to the higher quality of Google and that they were capable of foreclosing as-efficient competitors (second part of the first ground of appeal and second ground of appeal) must thus be rejected.

(2)The tying of the Google Search app and the Play Store

146.In the light of the considerations that have just been developed, Google’s specific claims relating to the alleged error committed by the General Court in not analysing whether the Google Search-Play Store bundle was capable of foreclosing as-efficient competing general search services has no prospect of success. Consequently, I shall only briefly address these arguments in the alternative.

147.First, Google contends that the General Court committed an error of law when it failed to take into account the market coverage of the practice at issue. A market coverage analysis is required in any case of exclusionary abuse. The General Court recognised this with regard to the RSAs. In so doing, it concluded illogically that the MADAs’ market coverage was significant, while considering similar market coverage insufficient in the case of the RSAs.

148.However, this line of argument cannot cast doubt on the validity of the General Court’s approach. Indeed, as the instances of tying at issue fell undoubtedly outside the scope of competition on the merits because of their characteristics as set out by the General Court, the analysis of their market coverage, which has in any case been developed for price-related practices, was not required in the present case to establish the abusive nature of those practices. (71)

149.Second, Google claims again that the General Court erred in law when it confirmed, especially in relation to Google Search, the Commission’s approach according to which it was not necessary to establish the exclusionary effect of the pre-installation conditions by reference to as-efficient competitors.

150.This argument must be dismissed for the reasons set out in points 125 to 145 of this Opinion. Moreover, Google does not, in any case, call into question in its appeal the General Court’s factual findings, recapitulated in points 118, 121 and 122 of this Opinion, that, for users, it was not only quality that played a role and that, in addition, the quality assessment of the various competing services remained similar. The same holds true for the factual findings set out in points 115 to 117 of this Opinion, according to which competing search services could not offset the ‘status quo bias’.

151.Lastly, in Google’s view, the General Court unjustly failed to analyse whether the different opportunities available for reaching users (pre-installation alongside Google’s app as well as downloads and access via the browser), not only in isolation, but also in combination, could have allowed as-efficient competitors to stay competitive.

152.However, this argument also amounts to an attempt to obtain a fresh assessment of the factual findings of the General Court, which are not as such called into question by Google. Moreover, Google does not explain what, specifically, would have been the added value of an analysis of the remaining opportunities ‘in combination’ as compared to the analysis of those opportunities carried out by the General Court.

(3)The tying of Chrome with the Play Store and Google Search

153.The specific arguments put forward by Google in relation to the opportunities for as-efficient competitors to offset the advantage created by the tying of the Chrome browser with the Play Store and Google Search cannot call into question the findings set out above either.

154.First, according to Google, the General Court analysed the legality of the tying of Chrome by reference to a geographic market other than the one defined in the decision at issue. In so doing, it analysed neither the market coverage of the practice at issue, nor the advantage conferred on Google by the pre-installation, nor the question of whether the remaining opportunities available to competitors ruled out the finding of an exclusion.

155.This argument has no prospect of success because it relies on the premiss that the market coverage of the Chrome-Play Store and Google Search bundle would have had to have been established in order for the abusive nature of that practice to be established. However, as set out in point 148 of this Opinion, this was not necessary in the present case.

156.Second, Google contends that the General Court erroneously transposed the conclusions relating to the Google Search tie by analogy to the Chrome tie.

157.However, the arguments put forward by Google in support of this criticism are inadmissible because they are aimed at securing a fresh assessment by the Court of Justice of the factual differences between the implications and the effects of the pre-installation of Google Search and Chrome. Moreover, Google demonstrates neither a distortion nor an error of law in the General Court’s assessments set out in paragraphs 439 to 465 of the judgment under appeal, according to which competing browser services could not offset the advantage conferred by the pre-installation of Chrome although there had been more competition relating to browsers than to general search services.

(c)Conclusion on the assessment of the MADAs

158.It follows from the foregoing that Google has not succeeded in pointing out errors of law committed by the General Court when assessing the arguments it put forward at first instance in relation to the MADA pre-installation conditions. Consequently, the first two grounds of appeal must be dismissed.

2.The assessment of the AFAs (third and fourth grounds of appeal)

159.In the context of its third ground of appeal, Google contends that the General Court rewrote the reasons for the decision at issue in relation to the AFAs and erroneously assessed the link between those agreements and their alleged effects (a). According to the fourth ground of appeal, the General Court wrongly confirmed the Commission’s decision not to accept Google’s justifications for the obligations contained in the AFAs that have been found to be abusive (b).

(a)The scope of the abuse in relation to the AFAs and its effects (third ground of appeal)

160.By its third ground of appeal, Google criticises the General Court for having reinterpreted the abusive conduct established by the Commission in relation to the AFAs (see point 22 of this Opinion) (1) and for having attributed the alleged exclusionary effects to a conduct that had not been found by the Commission to be abusive (2).

(1)The establishment of the contested conduct in relation to the AFAs (first part of the third ground of appeal)

161.In the context of the first part of its third ground of appeal, Google claims that the General Court expanded the finding of abuse of the AFAs and committed an error of law by including a legitimate practice in the conduct found to be abusive.

162.As set out in point 22 of this Opinion, in order to obtain a licence for the Play Store and Google Search, the OEMs had to enter into an AFA. In so doing, they committed not to sell devices running versions of Android that were not approved by Google. Accordingly, entering into a MADA was possible only after having entered into an AFA.

163.The General Court explained in paragraphs 806 to 808 of the judgment under appeal that the AFA prescribed the obligation to respect a minimum compatibility standard for the implementation of the Android source code.

164.That obligation covered all devices sold by an OEM that had entered into an AFA if those devices were running Android or an Android fork. (72) To prove their compatibility with the prescribed standards, the devices had to pass tests. Android forks that had passed those tests were designated as ‘Android-compatible forks’ by the General Court, whereas Android forks that had not been tested or that had not passed the tests were designated as ‘non-compatible Android forks’. (73)

165.As the General Court made clear in paragraphs 810, 811 and 828 of the judgment under appeal, the AFAs were found to be abusive in the decision at issue only in so far as they required OEMs to guarantee the compatibility with the prescribed standards of all devices sold by them and whose OS was Android or an Android fork, including devices on which Google’s apps were not pre-installed. In other words, the AFAs were considered to be abusive only in so far as they prohibited the commercialisation of smart mobile devices with non-compatible Android forks as OS even if no Google apps were pre-installed on those devices.

166.According to the General Court, this followed from the fact that the Commission had found compatibility obligations justified in the case of smart mobile devices on which Google’s apps were pre-installed.

167.Google now criticises the General Court for having rewritten those Commission findings and for having extended them to a practice that had not at all been judged abusive by the Commission.

168.Indeed, Google contends that, in the decision at issue, the Commission had classified as abusive only the OEMs’ obligation to respect the AFAs’ compatibility obligations in order to obtain a licence for the Play Store and Google Search. By contrast, the General Court had referred to the abusive practice, for example in paragraphs 828 and 864 of the judgment under appeal, as the practice ‘of preventing the development and market presence of devices running a non-compatible Android fork’.

169.By this amplification of the contested abuse, the General Court had impermissibly included Google’s legitimate practice of granting licences for its proprietary application programming interfaces (‘APIs’) (74) only when they were used on compatible Android devices. In so doing, the General Court committed an error of law when assessing the causal link between the alleged abuse and its alleged effects.

170.However, as the Commission objects, this criticism is based on an inaccurate recapitulation of the judgment under appeal.

171.Indeed, in paragraphs 812, 815 and 816 of the judgment under appeal, the General Court recalled, notably with reference to recital 1036 of the decision at issue, that the Commission had accused Google, against the background of the judgment in Microsoft v Commission (75) and the conditions under which it may be established that a bundle of products or obligations is abusive, (76) in essence, of having adopted an anticompetitive practice aimed at depriving non-compatible Android forks of commercial markets. This was, notably, due to the fact that, according to the decision at issue, non-compatible Android forks constituted a credible competitive threat to Google, the anti-fragmentation obligations hindered the development of non-compatible Android forks, and Android-compatible forks did not constitute a credible competitive threat to Google. Moreover, the capacity of the obligations at issue to restrict competition was reinforced by the unavailability of Google’s proprietary APIs to non-compatible Android fork developers, which reduced the incentive for developers to design apps intended to function on such OSs.

172.The General Court continued in paragraph 828 of the judgment under appeal to note that the Commission thus criticised Google for making the licensing of the Play Store and Google Search conditional on a set of obligations that restrict the freedom of OEMs wishing to obtain those licences, specifically in so far as they prohibit OEMs from marketing any other device running a non-compatible Android fork. The Commission did not dispute Google’s right to impose compatibility requirements in respect of devices on which its apps were installed. However, it considered that Google’s practice of preventing the development and market presence of devices running a non-compatible Android fork was abusive. It was therefore necessary to examine whether the Commission had succeeded in establishing that, as it found in the decision at issue, Google had implemented a practice designed to exclude non-compatible Android forks, and whether that practice might be classified as anticompetitive for the purposes of Article 102 TFEU.

173.Contrary to Google’s submissions, these considerations do not reveal any modification of the statements of the decision at issue.

174.In criticising those parts of the judgment under appeal, Google accuses the General Court, in essence, of having designated the exclusion of non-compatible Android forks as the objective of the compatibility obligations contained in the AFAs that had been found by the Commission to be abusive. However, in so doing, the General Court did not impermissibly replace the reasoning of the decision at issue, but merely summarised the latter’s essential criticism in relation to the abusive nature of the AFAs. (77)

175.It follows from the recapitulation of recital 1036 of the decision at issue by the General Court, reproduced in point 171 of this Opinion, the correctness of which is not disputed by Google, that the Commission did indeed set out there (and in the following sections of the decision at issue referred to in recital 1036) that the obligations of the AFAs at issue were precisely capable of restricting competition because they hindered the development of non-compatible Android forks. The Commission explained therein that only the latter constituted a threat to Google and that their development was restricted because the OEMs did not commercialise them any more since otherwise they would no longer have obtained a licence for Google’s apps.

176.This implies, purely logically, that the Commission accused Google of having adopted a practice that was aimed at excluding non-compatible Android forks because they constituted a threat to it. In analysing whether the Commission had established that fact to the requisite standard, the General Court thus did not rewrite the decision at issue.

177.Moreover, upon closer examination, it appears that the essence of Google’s criticism is in fact aimed at the General Court’s assessment in paragraphs 853 to 856 and 863 of the judgment under appeal. There, the General Court included Google’s practice of reserving its proprietary APIs to Android-compatible forks as a contextual element in its assessment of the AFAs.

178.According to Google, the fact that the OEMs commercialised only compatible Android devices is due not to the contested obligations of the AFAs for obtaining a licence for the Play Store and Google Search, but to the OEMs’ desire to have access to Google’s APIs. However, Google’s practice of reserving those APIs to compatible devices had not been criticised by the Commission or by the General Court.

179.However, as I have set out in points 60, 61, 67 and 68 of this Opinion, it is correct to assess a practice criticised as abusive in the light of its full context. It would indeed be artificial and unrealistic if a practice were analysed in isolation from its concrete legal and factual context. In so doing, the inclusion of a contextual element in this assessment does not depend on whether the element itself has been found to be abusive. As FairSearch convincingly explained at the hearing, especially in the context of the presence of a dominant undertaking such as Google, it would not be appropriate to ignore market conditions created by that undertaking itself.

180.It follows that the first part of the third ground of appeal is unfounded.

(2)The attribution of the alleged exclusionary effects (second part of the third ground of appeal)

181.In the context of the second part of its third ground of appeal, which follows the arguments that have just been assessed, Google contends that the General Court erred in attributing the alleged exclusionary effects to the contested compatibility obligations of the AFAs. Rather, the circumstance that the OEMs had no incentive to commercialise non-compatible Android forks was attributable to the fact that they could not access Google’s strategically important proprietary APIs.

182.If the General Court had carried out a counterfactual analysis, it would have found out that, had the AFA obligations not existed, non-compatible Android forks would have been in exactly the same situation: because of Google’s (legitimate) practice in relation to its APIs, they would equally have been confronted with the problem that the OEMs did not wish to commercialise them.

183.Moreover, the General Court had acknowledged that fact in its explanations concerning Google’s dominant position in paragraphs 226 to 233 of the judgment under appeal. Thus, it found in paragraph 229 that an alternative version of Android might not initially have been a credible competitor, notably because replicating Google’s proprietary apps (which were not available for such a version) would have required time and significant investment.

185.Accordingly, in judging in paragraph 893 of the judgment under appeal that a counterfactual analysis was not necessary because the Commission had established the AFAs’ anticompetitive effects to the requisite standard, the General Court committed an error of law.

186.This argument has no prospect of success.

187.As set out in points 93, 94 and 97 of this Opinion, the Commission is not required to carry out a counterfactual analysis in any case when assessing the potential or actual anticompetitive effects of a practice. This is notably not required when those effects are otherwise established to the requisite standard. Moreover, especially in the present context of highly complex digital markets, it may prove to be inappropriate to speculate about hypothetical options for action that the market participants could have taken in the absence of the contested conduct of the dominant undertaking. Last but not least, as the Commission itself contends and as the General Court set out, for example, in paragraph 850 of the judgment under appeal, the Commission does not have to prove that this conduct was the <i>only</i> reason for an exclusionary effect; it is sufficient if it is established to the requisite standard that it has <i>contributed</i> to that effect.

188.Moreover, the General Court’s finding in relation to Google’s dominant position that there was a risk that alternative Android versions might not initially have been credible competitors, so that the possible existence of such versions did not call into question the findings in relation to market dominance, does not signify that those versions did not exert any competitive pressure on Google. Rather, the General Court found in paragraphs 844 to 847 that non-compatible Android forks constituted a competitive threat to Google, without Google casting doubt on the evidence set out therein.

189.It follows that the second part of the third ground of appeal must also be dismissed as, accordingly, must the third ground of appeal in its entirety.

(b)The justification of the AFAs (fourth ground of appeal)

190.By its fourth ground of appeal, Google claims that the General Court erred when it failed to recognise the objective justifications for the AFAs.

191.As the General Court set out in paragraph 876 of the judgment under appeal, conduct is not abusive if it is justified by pro-competitive advantages or serves legitimate interests, it being for the dominant undertaking to prove that this is actually the case. (78)

192.However, the arguments put forward by Google in order to substantiate errors committed by the General Court when assessing this are for the most part inadmissible because they are aimed at obtaining a fresh assessment of the facts by the Court of Justice without any allegation of distortion of those facts. (79) Rather, an assessment of these arguments as well as of the findings of the General Court contested therein shows that Google was simply unable to establish that the contested anticompetitive components of the AFAs were objectively necessary and justified.

193.First, according to Google, in paragraph 880 of the judgment under appeal, the General Court erroneously relied on the actual growth of Android while the AFAs were in place to counter the argument that the obligations imposed by the AFAs were necessary for the survival of that OS on the market. Rather, the General Court should have examined how Android could have developed without the obligations at issue.

194.However, as has already been mentioned several times, it is not for the Commission and the General Court to consider how the market could have evolved without the contested behaviour. (80) Given the lack of sustainable arguments from Google concerning the concrete necessity of the obligations at issue, the General Court was right to rely on Google’s dominant position and the growth of Android in finding that the examples put forward by Google of other OSs run as ‘open source’ did not make the alleged threat to Android’s survival credible.

195.Second, according to Google, the General Court wrongly failed to consider Google’s legitimate interest in protecting not only the Android versions that were compatible with Google apps from incompatibilities and fragmentation, but the entire Android ecosystem, that is, also the versions without Google apps. In so doing, the General Court failed to consider Google’s related arguments.

196.This criticism is unfounded, as the General Court assessed Google’s arguments in that regard notably in paragraphs 882 to 884, 886 and 889 to 891 of the judgment under appeal. In so doing, the General Court considered, notably in paragraph 891, that Google had not proved the necessity of the exclusion of non-compatible Android forks for the survival of the ‘Android ecosystem’.

197.Third, Google alleges that the General Court misjudged that the AFA’s obligations had been a <i>conditio sine qua non</i> for offering an open-source OS under a licence free of charge. In so doing, it deprived dominant undertakings for the future of the possibility of launching such a ‘managed open-source model’; henceforth, those undertakings could only choose between a closed- and a fully open-source model.

198.As the Commission rightly objects, this criticism is also unfounded, however. Indeed, it follows from the findings of the General Court, notably in paragraphs 878, 882 and 890 of the judgment under appeal, that a dominant undertaking remains perfectly entitled to operate a ‘managed open-source model’ with proportionate and justified restrictions. An example in that regard is the AFAs’ compatibility obligations for devices <i>with</i> Google apps, which have not been contested by the Commission. However, as the General Court pointed out in paragraph 886 of that judgment, an undertaking’s right to take advantage of the economic benefits linked to the services which it develops does not mean that it must be recognised as having the right to prevent any competitors from existing on the market.

199.Fourth and lastly, according to Google, the General Court misconstrued its arguments alleging that a trade mark strategy reserving the trade marks ‘Google’ and ‘Android’ to compatible devices, proposed by the Commission as a less restrictive option, would not have been a practicable alternative.

200.This argument is inadmissible because Google criticises the General Court’s factual findings in relation to a trade mark strategy and reiterates the arguments put forward at first instance in that regard, without establishing any distortion of evidence or error committed by the General Court in the legal characterisation of the evidence in paragraph 883 of the judgment under appeal.

201.All in all, it follows from the foregoing that the fourth ground of appeal must also be dismissed.

(c)Conclusion on the assessment of the AFAs

202.It must be concluded that Google has not succeeded in establishing that any errors of law were committed by the General Court when assessing Google’s arguments at first instance in relation to the abusive nature of the AFAs’ compatibility obligations. The third and fourth grounds of appeal must therefore be dismissed.

3.The single and continuous infringement (fifth ground of appeal)

203.By its fifth ground of appeal, Google contends that the General Court should have annulled the decision at issue in its entirety following its finding that the Commission had not established the abusive nature of the portfolio-based RSAs. Indeed, that decision was founded on the establishment of a single and continuous infringement, composed of four distinct infringements (the MADA-Google Search bundle, the MADA-Chrome bundle, the AFAs and the portfolio-based RSAs). After the disappearance of the portfolio-based RSAs, an essential and indivisible element of this single and continuous infringement, the rest of the decision at issue could no longer be maintained either.

204.According to the Commission, this ground of appeal is inadmissible because Google did not call into question at first instance the existence of a single and continuous infringement, nor did it allege that the decision at issue should be annulled in its entirety if one of its components was annulled.

205.That objection must be rejected. It is true that an appellant cannot put forward on appeal new reasons and arguments that have not been the subject of debate at first instance. However, it is permissible to lodge an appeal relying, before the Court of Justice, on pleas arising from the judgment under appeal itself which seek to criticise, in law, its merits. (81)

206.However, Google’s fifth ground of appeal fails on the merits.

207.If a series of acts or continuous conduct form part of an ‘overall plan’ because of their identical object of distorting competition, they can constitute together a single and continuous infringement. (82) Even if that term has so far primarily been used to demonstrate the unity and the continuity of infringements of Article 101 TFEU committed by a number of undertakings, it can also be relevant in the context of the application of Article 102 TFEU. (83)

208.In the present case, as the General Court set out in paragraphs 18, 64 to 72 and 1018 to 1029 of the judgment under appeal, because of their common objective and their combined effects, the Commission described the restrictions at issue not only as four separate infringements, but also as a single and continuous infringement. This was especially because those restrictions were part of an overall strategy by Google to cement its dominant position on the online general search market at a time when the importance of the mobile internet was growing significantly. The common objective of this strategy was to ensure that Google had the best possible access to general searches carried out by consumers on smart mobile devices.

209.As Google rightly states, partial annulment of an act of EU law is possible only if the elements which it is sought to have annulled can be severed from the remainder of the measure. That requirement is not satisfied where the partial annulment of a measure would cause the substance of that measure to be altered. (84)

210.However, in the present case, the partial annulment of the decision at issue as regards the fourth infringement in the form of the portfolio-based RSAs does not, as Google contends, cause the substance of that measure to be altered as regards the three other infringements as well as the single and continuous infringement constituted by them.

211.On the one hand, as explained, notably, in points 60 to 69 and 179 of this Opinion, the annulment of the findings relating to the portfolio-based RSAs does not affect the findings of the decision at issue with regard to the other three infringements. The mere fact that those other parts continue to exist precludes the annulment of that decision in its entirety. (85)

212.On the other hand, the Commission’s findings, confirmed by the General Court, in relation to the existence of a single and continuous infringement also continue to be valid despite the elimination of the portfolio-based RSAs as an independent infringement.

213.As the General Court stated, in particular, in paragraphs 1021 to 1023 and 1025 of the judgment under appeal, the factual circumstances of the established infringements show that the first and second restrictions at issue, that is, the MADA-Google Search bundle, the MADA-Chrome bundle and the AFAs, which have been classified as three distinct infringements, were part of an overall strategy. This strategy consisted in attaching special conditions to the use of the Android OS, on the one hand, and of certain apps and services, on the other. It was aimed at anticipating the development of the mobile internet, while preserving Google’s own business model, which is based on the revenues which it derives essentially from the use of its general search service. In the context of that overall strategy pursued by Google, the preservation of the dominant position which it held, throughout the entire period of the infringement, on the national markets for general search services was therefore of decisive importance, to which the first and second restrictions at issue contributed.

214.Those findings, the factual bases of which Google does not call into question, do not disclose an error of law in relation to the conclusion that, even after the elimination of the portfolio-based RSAs as an independent infringement, the MADAs and the AFAs continued to represent an intertwined system that served the implementation of an abusive overall strategy to reach an identical goal.

215.Moreover, it follows from the statements in points 60 to 69 and 179 of this Opinion that the relevant circumstances of the portfolio-based RSAs, even if not characterised as an independent infringement, were an indivisible component of Google’s overall strategy. Accordingly, the General Court did not err in law by taking those elements into account for the assessment of the infringement. Contrary to Google’s submission, this holds true in the context of the finding of separate infringements as well as in the context of the finding of a single and continuous infringement. (86)

216.It follows from the above that the fifth ground of appeal must also be dismissed.

4.Conclusion on the establishment of the infringement

217.As none of the first five grounds of appeal is well founded, the General Court’s findings in relation to the existence and the characteristics of the infringement must be upheld.

B.The fine (sixth ground of appeal)

218.By its sixth and last ground of appeal, Google contends that the General Court committed four errors of law when exercising its unlimited jurisdiction in relation to the fine. In so doing, as it did not correctly take into account the elimination of the portfolio-based RSAs as an independent infringement when recalculating the fine, it has de facto raised the fine for the other parts of the infringement.

219.As the General Court set out in paragraph 19 of the judgment under appeal, in the decision at issue, the Commission imposed a fine of EUR 4 342 865 000 on Google LLC, of which EUR 1 921 666 000 jointly and severally with Alphabet Inc. In determining that amount, the Commission took into consideration the value of relevant sales within the EEA, in relation to the single and continuous infringement, achieved by Google during the last year of participation in the infringement (2017), and applied a gravity coefficient (11%). The Commission then multiplied the amount obtained by the number of years of participation in the infringement (approximately 7.52) and added an additional amount (equivalent to 11% of the value of sales in 2017) in order to deter similar undertakings from engaging in the same practices. The Commission also considered that it was not appropriate to find that there were mitigating or aggravating circumstances, or to take particular account of Google’s significant financial capacity in order to decrease or increase the amount of the fine.

220.In paragraphs 1016 and 1029 of the judgment under appeal, the General Court explained that the annulment of the decision at issue in relation to the abusive nature of the portfolio-based RSAs entailed a need also to vary that decision in so far as it established Google’s participation in a single and continuous infringement of Article 102 TFEU which constituted the fourth infringement. However, it was also necessary to take into account the fact that the establishment of a single and continuous infringement as part of an overall strategy, supported by the first and second aspects of that infringement, that is, the MADAs and the AFAs, was not unlawful.

221.In paragraphs 1036 to 1112 of the judgment under appeal, the General Court then set out the aspects that it took into account when recalculating the fine. In so doing, it explained that it took into consideration the same value of sales as the Commission (paragraph 1072), but that the application of the fixed gravity coefficient of 11% of that value of sales determined by the Commission did not sufficiently reflect the implementation and the intensity of the infringement over the relevant period (paragraph 1081). In relation to the intensity of the infringement, the General Court distinguished three periods, that is from 1 January 2011 until 1 August 2012, from 1 August 2012 until 31 March 2014, and from 31 March 2014 until the adoption of the decision at issue on 18 July 2018. In the General Court’s view, the intensity of the infringement was particularly high during the second period (paragraphs 1082 to 1094).

222.On the basis of those considerations, the General Court fixed the fine at EUR 4 125 000 000 instead of EUR 4 342 865 000, of which EUR 1 520 605 895 jointly and severally with Alphabet Inc. (paragraphs 1099 and 1100).

223.Before addressing Google’s criticism in that regard, it must be recalled that the unlimited jurisdiction conferred on the EU judicature by Article 31 of Regulation (EC) No 1/2003 (87) in accordance with Article 261 TFEU empowers the competent Court, in addition to carrying out a mere review of legality with regard to the penalty, to substitute its own appraisal for the Commission’s and, consequently, to cancel, reduce or increase the fine or penalty payment imposed. (88)

224.It is true that the exercise of that jurisdiction does not amount to a review of the Court’s own motion. (89) However, in order to satisfy the requirements of Article 47 of the Charter of Fundamental Rights of the European Union when conducting a review in the exercise of its unlimited jurisdiction with regard to the fine, the EU judicature is bound, in the exercise of the powers conferred by Articles 261 and 263 TFEU, to examine all complaints based on issues of fact and law which seek to show that the amount of the fine is not commensurate with the gravity or the duration of the infringement. (90)

225.By its first complaint in that regard, Google claims that the General Court infringed its rights of defence because it had not had the opportunity to comment on the elements taken into account for the recalculation, especially the allegedly higher intensity of the infringement during the second period. Moreover, the General Court had distorted evidence when finding that the period between 2012 and 2014 had been decisive for the development of Google’s competitors.

226.So far as the claim of distortion of evidence is concerned, this claim is not sufficiently substantiated. Indeed, Google does not mention any allegedly distorted elements, but merely argues that the Commission had found in the decision at issue that the transition to mobile internet had taken place from 2007; accordingly, there was no indication that the period between 2012 and 2014 was particularly important for Google’s competitors, as held by the General Court. However, on the basis of this general affirmation, it is not possible to find a distortion of evidence.

227.The claim that the General Court infringed Google’s rights of defence is equally unfounded.

228.From the moment a claimant asks the General Court to exercise its unlimited jurisdiction with regard to a penalty, it must be aware that the General Court can assess that penalty in the light of all of the circumstances of the case and take those circumstances into account when recalculating it. Respect for the rights of defence requires in that context that the parties may put their case in relation to all relevant circumstances, but not that the General Court ask them to comment on the recalculation actually envisaged. (92) In the present case, in the proceedings at first instance, Google specifically asked the General Court to take into account the intensity of the infringement, and it does not claim to have been unable to make submissions concerning the factual elements considered by the General Court for that assessment.

229.Thereafter, by its second complaint, Google claims that the General Court has infringed the presumption of innocence because it relied, when assessing the intensity and duration of the infringement, on contextual elements such as the RSAs, Google’s politics as regards its APIs or Google’s agreement with Apple, which had not been characterised as abusive in themselves.

230.However, I have already explained in points 60 to 69 and 179 of this Opinion that the contextual elements of abusive conduct, especially when they stem from other conduct of the dominating undertaking itself, can be taken into account in the assessment of the effects of an infringement.

231.In the context of its third complaint, Google contends that the General Court infringed the principle ne ultra petita because it applied the gravity coefficient in a way that none of the parties had requested.

232.This objection must also be rejected.

233.As I have already set out elsewhere, the unlimited jurisdiction enjoyed by the Court does in principle also afford it the option of increasing the fine. (93) There is no prohibition of reformatio in peius either in the proceedings at first instance or in the appeal proceedings. In addition, with regard to the amount of the fine, the Courts of the European Union are not bound by the forms of order sought by the parties, provided that they adhere to the subject matter of the proceedings as determined in the action for annulment and in the appeal. (94) It is not disputed that that is the case here.

234.Lastly, by its fourth complaint, Google argues that the fine set by the General Court is disproportionate and not properly reasoned.

235.That argument must also be dismissed.

236.According to the Court of Justice’s case-law, the Commission fulfils its obligation to state reasons when it sets out the factors which enabled it to determine the gravity of the infringement and its duration. Although it is not required to provide all of the figures relating to each of the intermediate steps relating to the method of calculation, it is nevertheless incumbent on it to explain the weighting and assessment of the factors taken into account. The Commission’s obligation to provide sufficient reasons for the relevance and weighting of the factors which it has taken into account in determining the method it has favoured does not mean that it is required to provide figures relating to the method of calculating the fine or that it is required to explain in detail the internal calculations which it has carried out. (95)

237.The same must apply to the General Court.

238.Accordingly, it is certainly permissible to wonder why, in the present case, the General Court did not just include in its reasoning the calculation it had carried out in order to obtain the numerical amount of the fine in the light of the extensive account of the matters taken into consideration. It is also permissible to regret that approach and to consider that setting out those calculations would have clarified the General Court’s reasoning and facilitated the work of the Court of Justice. However, this does not mean that the approach of the General Court must lead to the judgment under appeal being set aside.

239.According to established case-law, it is not for the Court of Justice, when ruling on points of law in the context of an appeal, to substitute, on grounds of fairness, its own assessment for that of the General Court exercising its unlimited jurisdiction to rule on the amount of fines imposed on undertakings for infringements of EU law. Only where the Court of Justice considers that the level of the penalty is not merely inappropriate, but also excessive to the point of being disproportionate, does it have to find that the General Court erred in law, on account of the inappropriateness of the amount of a fine. (96)

240.In the present case, first, the General Court has extensively set out, in around 80 paragraphs, the aspects that it took into account for the recalculation of the fine, particularly as regards the gravity and duration of the contested infringement. Second, the new fine, reduced by slightly more than EUR 200 000 000 as compared to the original one (from initially EUR 4 342 865 000 to now EUR 4 125 000 000), does not appear, even in view of the elimination of the portfolio-based RSAs as independent abuse, to be excessive to the point of being disproportionate. Indeed, as the General Court pointed out in particular in paragraphs 1015 to 1029 of the judgment under appeal, the other aspects of the infringement as well as its single and continuous overall strategy, which amplified its impact, persist notwithstanding this elimination.

241.It follows that Google’s sixth ground of appeal cannot succeed either.

C.Conclusion on the assessment of the appeal

242.It follows from the foregoing considerations that all six grounds of appeal must be dismissed, as, therefore, must the appeal in its entirety.

V.Costs

243.Pursuant to Article 184(2) of its Rules of Procedure of the Court of Justice, where the appeal is unfounded, the Court is to make a decision as to the costs.

244.First, under Article 138(1) and (2) of the Rules of Procedure, which applies to appeal proceedings by virtue of Article 184(1) thereof, the unsuccessful party is to be ordered to pay the costs if they have been applied for in the successful party’s pleadings; where there is more than one unsuccessful party, the Court is to decide how the costs are to be shared. Since the Commission has applied for costs and the appellants have been unsuccessful in their grounds of appeal, they must be ordered to bear their own costs and to pay those incurred by the Commission. Since they brought the appeal jointly, they must bear the costs jointly and severally.

245.Next, under Article 184(4) of its Rules of Procedure, the Court may decide that an intervener at first instance who has participated in the written or oral part of the proceedings before the Court is to bear its own costs. Since ADA, CCIA, Gigaset, HMD and Opera, on the one hand, and BDZV, BEUC, FairSearch, Qwant, Seznam and VDZ, on the other, participated in the written and/or oral parts of the present appeal proceedings, they should be ordered to bear their own costs.

VI.Conclusion

246.On the basis of the above considerations, I propose that the Court should:

–Dismiss the appeal;

–Order Google LLC and Alphabet Inc. to bear their own costs and to pay those incurred by the European Commission;

–Order Application Developers Alliance, BDZV – Bundesverband Digitalpublisher und Zeitungsverleger eV, Bureau européen des unions de consommateurs (BEUC), Computer & Communications Industry Association, FairSearch AISBL, Gigaset Communications GmbH, HMD global Oy, Opera Norway AS, Qwant, Seznam.cz, a.s. and Verband Deutscher Zeitschriftenverleger eV each to bear their own costs.

—

1Original language: English.

2Judgment of 14 September 2022 (T‑604/18, ‘the judgment under appeal’, EU:T:2022:541, paragraph 54).

3Commission Decision of 18 July 2018 relating to a proceeding under Article 102 [TFEU] and Article 54 of the EEA Agreement (Case AT.40099 – Google Android) (summary found under OJ 2019 C 402, p. 19) (‘the decision at issue’).

4Judgment of 12 May 2022 (C‑377/20, EU:C:2022:379).

5Judgment of 19 January 2023 (C‑680/20, EU:C:2023:33).

6Judgment of 21 December 2023 (C‑333/21, EU:C:2023:1011).

7Judgment of 10 September 2024 (C‑48/22 P, EU:C:2024:726).

8Judgment of 24 October 2024 (C‑240/22 P, EU:C:2024:915).

9Paragraph 120 of the judgment under appeal, and recitals 217 and 402 of the decision at issue.

10Paragraph 121 of the judgment under appeal, and recital 439 of the decision at issue.

11Paragraphs 274 and 300 of the judgment under appeal, and recitals 752, 753, 1009 and 1010 of the decision at issue.

12Paragraphs 274, 300 and 314 to 316 of the judgment under appeal.

13Paragraphs 803 and 805 of the judgment under appeal, and recitals 1015 and 1016 of the decision at issue.

14Paragraphs 621 and 622 of the judgment under appeal, and recital 1192 of the decision at issue.

15Paragraph 38 of the judgment under appeal.

16Order of the President of the Court of Justice of 19 January 2023, Google and Alphabet v Commission (C‑738/22 P, EU:C:2023:44).

17Order of the President of the Court of Justice of 18 April 2023, Google and Alphabet v Commission (C‑738/22 P, EU:C:2023:326).

18See judgment of 10 September 2024, Google and Alphabet v Commission (Google Shopping) (C‑48/22 P, EU:C:2024:726, paragraphs 223 to 230).

19See, concerning that criterion, judgment of 24 October 2024, Commission v Intel Corporation (C‑240/22 P, EU:C:2024:915, paragraph 90).

20See paragraphs 623 and 631 of the judgment under appeal.

21Recital 1195 et seq. of the decision at issue, and paragraph 637 of the judgment under appeal.

22See recitals 1225 to 1271 of the decision at issue, and paragraphs 700 to 711, 733 and 734 of the judgment under appeal.

23See paragraphs 679 to 698 and 733 to 802 of the judgment under appeal.

24See, in particular, recitals 197 and 1292 of the decision at issue, and paragraphs 452, 632, 633 and 1090 of the judgment under appeal.

25See paragraphs 419 to 538 of the judgment under appeal, in particular paragraphs 426, 428, 434, 443, 446 to 452, 457, 480, 481, 494, 495 and 537.

26Paragraphs 451, 452, 1018 and 1019 of the judgment under appeal, with reference to recital 822 of, and footnote 908 to, the decision at issue.

27Judgments of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 72); of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraphs 40, 44, 52 and 62); of 21 December 2023, European Superleague Company (C‑333/21, EU:C:2023:1011, paragraph 130); of 10 September 2024, Google and Alphabet v Commission (Google Shopping) (C‑48/22 P, EU:C:2024:726, paragraphs 166 and 168); and of 24 October 2024, Commission v Intel Corporation (C‑240/22 P, EU:C:2024:915, paragraphs 179 and 331).

28Judgments of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 70), and of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraph 43).

29Judgment of 6 December 2012 (C‑457/10 P, EU:C:2012:770, paragraph 140).

30Judgment of 16 June 2022 (C‑698/19 P, EU:C:2022:480, paragraphs 64 and 80).

31Judgment of 10 September 2024 (C‑48/22 P, EU:C:2024:726, paragraphs 170 to 172 and 240 to 247).

32Paragraphs 451 and 452 of the judgment under appeal; see also paragraphs 495 and 537 of that judgment.

See points 112 to 124 of this Opinion.

—

34See paragraphs 336 to 338 and 565 of the judgment under appeal.

—

35See judgment of 10 September 2024, Google and Alphabet v Commission (Google Shopping) (C‑48/22 P, EU:C:2024:726, paragraphs 240 to 247), as well as my Opinion in Google and Alphabet v Commission (C‑48/22 P, EU:C:2024:14, points 177 to 182).

—

36Paragraph 428 of the judgment under appeal.

—

37See, in that context, paragraphs 332 and 333 of the judgment under appeal.

—

38See paragraph 330 of the judgment under appeal.

—

39Judgment of 9 November 2023, Global Silicones Council and Others v ECHA (C‑559/21 P, EU:C:2023:842, paragraph 50 and the case-law cited).

—

40Judgment of 19 December 2013, Siemens and Others v Commission (C‑239/11 P, C‑489/11 P and C‑498/11 P, EU:C:2013:866, paragraphs 38 to 46 and the case-law cited).

—

41See, to that effect, ibid.

—

42See order of 29 September 2010, EREF v Commission (C‑74/10 P and C‑75/10 P, EU:C:2010:557, paragraph 41 and the case-law cited).

—

43See, to that effect, judgment of 10 September 2024, Google and Alphabet v Commission (Google Shopping) (C‑48/22 P, EU:C:2024:726, paragraphs 223 to 229).

—

44See, to that effect, judgment of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraphs 98 to 102).

—

45See, to that effect, judgment of 27 June 2024, Commission v Servier and Others (C‑176/19 P, EU:C:2024:549, paragraph 353).

—

46See, on this point, paragraphs 115 and 116 of the judgment under appeal.

—

47See, for example, judgment of 24 October 2024, Commission v Intel Corporation (C‑240/22 P, EU:C:2024:915, paragraph 26).

—

48See judgments of 14 November 1996, Tetra Pak v Commission (C‑333/94 P, EU:C:1996:436, paragraph 37); of 12 December 1991, Hilti v Commission (T‑30/89, EU:T:1991:70, paragraphs 100 and 101); and of 6 October 1994, Tetra Pak v Commission (T‑83/91, EU:T:1994:246, paragraphs 136 and 137).

—

49See, to that effect, judgment of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraphs 78 and 91 et seq.).

—

50Judgment of 17 September 2007 (T‑201/04, EU:T:2007:289, paragraphs 857, 867 to 869, 977 and 1034 to 1058).

—

51Paragraphs 336 and 493 of the judgment under appeal.

—

52Judgment of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 61), and Opinion of Advocate General Rantos in Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2021:998, points 42 to 46).

—

53Opinion of Advocate General Rantos in Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2021:998, points 47 to 50).

—

54Judgments of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 78), and of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraphs 39 and 57). See also judgments of 12 January 2023, Lietuvos geležinkeliai v Commission (C‑42/21 P, EU:C:2023:12, paragraph 91), and of 21 December 2023, European Superleague Company (C‑333/21, EU:C:2023:1011, paragraphs 134 to 138, 147, 148 and 152). See, further, judgment (subsequently set aside by the Court of Justice) of 12 June 2014, Intel v Commission (T‑286/09, EU:T:2014:547, paragraphs 198 to 210); the question now arises again in pending case T‑1129/23.

—

55The analysis relates to hypothetical as-efficient competitors and not to the actual competitors of the dominant undertaking; see judgments of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraphs 78, 79, 82, 91 and 101); of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraphs 56 and 59); and of 24 October 2024, Commission v Intel Corporation (C‑240/22 P, EU:C:2024:915, paragraphs 311 and 343).

—

56See, to that effect, judgments of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 82), and of 24 October 2024, Commission v Intel Corporation (C‑240/22 P, EU:C:2024:915, paragraph 181).

—

57See judgments of 17 February 2011, TeliaSonera Sverige (C‑52/09, EU:C:2011:83, paragraphs 40 to 43); of 27 March 2012, Post Danmark (C‑209/10, EU:C:2012:172, paragraph 38); of 6 October 2015, Post Danmark (C‑23/14, EU:C:2015:651, paragraph 57 et seq.); of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraphs 56 and 57); and of 24 October 2024, Commission v Intel Corporation (C‑240/22 P, EU:C:2024:915, paragraph 181). See also Opinion of Advocate General Rantos in Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2021:998, point 70).

—

58See judgments of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraphs 79 to 83), and of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraph 59); see also Opinion of Advocate General Rantos in Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2021:998, points 68 to 71). See, however, my Opinion in Google and Alphabet v Commission (C‑48/22 P, EU:C:2024:14, point 192 et seq.).

—

59Judgments of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632, paragraphs 138 and 139), and of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraphs 52 to 62).

—

60See, to that effect, judgments of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632, paragraphs 141 to 144), and of 10 September 2024, Google and Alphabet v Commission (Google Shopping) (C‑48/22 P, EU:C:2024:726, paragraph 265).

—

61See judgments of 6 October 2015, Post Danmark (C‑23/14, EU:C:2015:651, paragraphs 57 and 58); of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 81); of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraphs 57 and 58); and of 10 September 2024, Google and Alphabet v Commission (Google Shopping) (C‑48/22 P, EU:C:2024:726, paragraph 264).

—

62See judgments of 6 October 2015, Post Danmark (C‑23/14, EU:C:2015:651, paragraphs 57 to 61); of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 82) (‘when [that test] can be carried out’); of 19 January 2023, Unilever Italia Mkt. Operations (C‑680/20, EU:C:2023:33, paragraph 57); and of 10 September 2024, Google and Alphabet v Commission (Google Shopping) (C‑48/22 P, EU:C:2024:726, paragraphs 266 to 269). See also my Opinions in Post Danmark (C‑23/14, EU:C:2015:343, points 71 to 74) and in Google and Alphabet v Commission (C‑48/22 P, EU:C:2024:14, point 192 et seq.).

—

63See paragraphs 115 and 116 of the judgment under appeal.

—

64See, for example, paragraph 294 of the judgment under appeal.

—

65See point 18 of this Opinion.

—

66Judgments of 21 December 2023, European Superleague Company (C‑333/21, EU:C:2023:1011, paragraphs 129 and 131), and of 10 September 2024, Google and Alphabet v Commission (Google Shopping)

(C‑48/22 P, EU:C:2024:726, paragraph 167).

67See, in that regard, my Opinion in Google and Alphabet v Commission (C‑48/22 P, EU:C:2024:14, points 191 to 198). See also point 73 of the Commission’s 2024 draft Guidelines on the application of Article 102 [TFEU] to abusive exclusionary conduct by dominant undertakings, as well as the accompanying Staff Policy Brief of March 2023 with the title ‘A dynamic and workable effects-based approach to abuse of dominance’, p. 5 et seq., both available under Application of Article 102 TFEU – European Commission.

68See, to that effect, judgment of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 99); see also judgment of 17 September 2007, Microsoft v Commission (T‑201/04, EU:T:2007:289, paragraph 1069).

69Judgment of 12 May 2022, Servizio Elettrico Nazionale and Others (C‑377/20, EU:C:2022:379, paragraph 102).

70Ibid., paragraphs 78, 91 and 101.

71See, to that effect, the case-law set out in point 106 of this Opinion. See, to that effect also, especially as regards digital markets, judgment of 10 September 2024, Google and Alphabet v Commission (Google Shopping) (C‑48/22 P, EU:C:2024:726, paragraph 265), in which the Court only partially refers to the judgment of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632, paragraphs 138 and 139), without specifically mentioning the criterion of market coverage that was dispensable in that case because of Google’s dominant market position.

72As set out in point 13 of this Opinion, a fork is new software established from the source code of existing software.

73Paragraph 808 of the judgment under appeal.

74An API is a particular set of rules and specifications that a software program follows in order to access and make use of the services and resources provided by another software program or hardware that also implements that API. In essence, APIs allow software programs and hardware, or different software programs, to communicate with each other (recitals 89 and 90 of the decision at issue). According to Google, APIs enable Android app developers notably to take advantage of Google’s proprietary services, like Google Maps.

75Judgment of 17 September 2007 (T‑201/04, EU:T:2007:289).

76See, in that regard, point 107 of this Opinion.

77See, in that regard, judgment of 6 October 2021, World Duty Free Group and Spain v Commission (C‑51/19 P and C‑64/19 P, EU:C:2021:793, paragraphs 70 to 73).

78Judgments of 14 February 1978, United Brands and United Brands Continentaal v Commission (27/76, EU:C:1978:22, paragraph 184); of 27 March 2012, Post Danmark (C‑209/10, EU:C:2012:172, paragraphs 40 to 42); and of 30 January 2020, Generics (UK) and Others (C‑307/18, EU:C:2020:52, paragraphs 165 to 167).

79See, in that regard, points 79 and 80 of this Opinion and the case-law cited.

80See points 97 and 186 of this Opinion.

81Judgment of 10 April 2014, Commission v Siemens Österreich and Others and Siemens Transmission & Distribution and Others v Commission (C‑231/11 P to C‑233/11 P, EU:C:2014:256, paragraph 102).

82Judgments of 8 July 1999, Commission v Anic Partecipazioni (C‑49/92 P, EU:C:1999:356, paragraph 81); 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 258); and of 6 December 2012, Commission v Verhuizingen Coppens (C‑441/11 P, EU:C:2012:778, paragraph 41).

83See judgments of 18 April 2024, Heureka Group (Online price comparison services) (C‑605/21, EU:C:2024:324, paragraphs 62 and 79), and of 1 July 2010, AstraZeneca v Commission (T‑321/05, EU:T:2010:266, paragraphs 891 to 895); see also my Opinion in Heureka Group (Online price comparison services) (C‑605/21, EU:C:2023:695, points 90 and 91).

84Judgment of 6 December 2012, Commission v Verhuizingen Coppens (C‑441/11 P, EU:C:2012:778, paragraphs 35 to 39).

85See, to that effect, ibid., paragraphs 51 and 52.

86See, to that effect, judgment of 16 June 2022, Sony Optiarc and Sony Optiarc America v Commission (C‑698/19 P, EU:C:2022:480, paragraphs 64 and 80).

87Council Regulation of 16 December 2002 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty (OJ 2003 L 1, p. 1).

88Judgments of 15 October 2002, Limburgse Vinyl Maatschappij and Others v Commission (C‑238/99 P, C‑244/99 P, C‑245/99 P, C‑247/99 P, C‑250/99 P to C‑252/99 P and C‑254/99 P, EU:C:2002:582, paragraph 692); of 8 December 2011, Chalkor v Commission (C‑386/10 P, EU:C:2011:815, paragraph 63); and of 26 September 2018, Infineon Technologies v Commission (C‑99/17 P, EU:C:2018:773, paragraph 193).

89Judgments of 8 December 2011, Chalkor v Commission (C‑386/10 P, EU:C:2011:815, paragraph 64); of 10 July 2014, Telefónica and Telefónica de España v Commission (C‑295/12 P, EU:C:2014:2062, paragraph 213); and of 26 September 2018, Infineon Technologies v Commission (C‑99/17 P, EU:C:2018:773, paragraph 194).

90Judgments of 10 July 2014, Telefónica and Telefónica de España v Commission (C‑295/12 P, EU:C:2014:2062, paragraph 200); of 26 January 2017, Villeroy & Boch Austria v Commission (C‑626/13 P, EU:C:2017:54, paragraph 82); and of 26 September 2018, Infineon Technologies v Commission (C‑99/17 P, EU:C:2018:773, paragraph 195).

91Judgments of 17 December 1998, Baustahlgewebe v Commission (C‑185/95 P, EU:C:1998:608, paragraph 128); of 28 June 2005, Dansk Rørindustri and Others v Commission (C‑189/02 P, C‑202/02 P, C‑205/02 P to C‑208/02 P and C‑213/02 P, EU:C:2005:408, paragraph 244); and of 26 September 2018, Infineon Technologies v Commission (C‑99/17 P, EU:C:2018:773, paragraph 192).

92See, to that effect, judgment of 24 September 2009, Erste Group Bank and Others v Commission (C‑125/07 P, C‑133/07 P, C‑135/07 P and C‑137/07 P, EU:C:2009:576, paragraphs 158, 159 and 180 to 183).

93Judgment of 8 February 2007, Groupe Danone v Commission (C‑3/06 P, EU:C:2007:88, paragraph 61).

94See my Opinion in Joined Cases Fresh Del Monte Produce v Commission and Commission v Fresh Del Monte Produce (C‑293/13 P and C‑294/13 P, EU:C:2014:2439, paragraph 273).

95Judgment of 10 July 2019, Commission v Icap and Others (C‑39/18 P, EU:C:2019:584, paragraphs 31 and 38 and the case-law cited).

96See judgment of 25 March 2021, Lundbeck v Commission (C‑591/16 P, EU:C:2021:243, paragraphs 197 and 198 and the case-law cited).