On 2 July 2026, the European Court of Justice (ECJ) delivered its much-awaited judgment in the Google Android case (the Judgment). It dismissed Google's appeal against the General Court's judgment upholding the European Commission's (the Commission) 2018 decision finding that Google abused its dominant position in the context of the Android mobile operating system and imposing a record EUR 4.3 billion fine (the Decision).
What you need to know
- The ECJ confirmed that tying is regarded, in principle, as departing from competition on the merits, so that the Commission does not need additional specific evidence on that limb, though it must still show the capability to foreclose.
- The Commission enjoys a broad margin of appreciation to prove exclusionary abuses. It may take lawful conduct by the dominant firm into account as a relevant contextual factor amplifying the effects of potentially abusive conduct. It is also not required to run an as-efficient-competitor test (AECT), particularly in relation to non-pricing conduct, and is not mandated to carry out a counterfactual analysis to establish causation.
- The Judgment has immediate implications for digital companies, indicating that an AECT will generally be irrelevant in ecosystems with significant barriers to entry and network effects (at least for non-pricing conduct). Moreover, preinstallation conditions can suffice to show causality if the competition authority can prove a "status quo bias".
- Companies under investigation should proactively develop robust qualitative and quantitative evidence (including an AECT where appropriate) to demonstrate the absence of exclusionary effects, as this will oblige the competition authority to engage with and rebut such evidence.
Background
The Decision
In 2018, the Commission found that Google had abused its dominant position in the national markets for general search services and in the worldwide market for Android-compatible app stores by imposing allegedly anticompetitive restrictions on mobile phone manufacturers (OEMs) and mobile network operators (MNOs). It identified four distinct infringements in three sets of agreements:
- Mobile Application Distribution Agreements (MADAs): To obtain a Play Store licence, OEMs were required to pre-install Google Search and Google Chrome - both were classified by the Commission as separate instances of tying;
- Anti-fragmentation Agreements (AFAs): Google prevented OEMs wishing to pre-install Google apps from selling any device running alternative versions of Android not approved by Google, known as "Android forks"; and
- Revenue Share Agreements (RSAs): Google made payments to certain large OEMs and MNOs on condition that they exclusively pre-installed Google Search on their devices.
These categories of conduct were found to constitute a single and continuous infringement of Article 102 TFEU in relation to which the Commission imposed a record fine of EUR 4.3 billion.
Google appealed the Decision before the EU General Court (GC).
The GC judgment
The GC largely upheld the Decision, but annulled the Commission's findings on the RSAs and slightly reduced the fine to EUR 4.1 billion. The GC identified flaws in the assessment of the market coverage of the exclusivity payments in the RSAs and in the application of the as-efficient-competitor test (AECT) and concluded that those errors vitiated the finding that the RSAs constituted an abuse.
Google appealed the GC judgment to the ECJ. The Commission did not cross-appeal the GC findings on the RSAs – which are therefore final.
Key findings in the ECJ Judgment
Following the Opinion of Advocate General (AG) Kokott, the ECJ entirely dismissed Google's appeal. The Judgment consolidates previous case law on exclusionary abuses and provides some clarifications on the following points: (i) the test for assessing exclusionary abuses; (ii) tying; (iii) the relevance of as-efficient competitors and the AECT; (iv) the need to carry out a counterfactual analysis; and (v) the concept of a single and continuous infringement.
Requisite legal standard to establish exclusionary abuses
The ECJ consolidated the following general principles on the legal standard to establish exclusionary abuses:
- A competition authority must in principle show that the conduct (i) departs from "competition on the merits", i.e. conduct leading to lower prices, better quality and/or wider choice, and (ii) is capable of creating exclusionary effects (§§ 264-265).
- Exclusionary effects can be potential or actual, and foreclosure covers not only the exclusion of competitors but also hindering their growth (§§ 226, 270-271).
- To prove exclusionary effects, competition authorities can use different analytical templates but are always required to consider "all relevant factual circumstances", including the nature of the conduct, the characteristics of the market(s) concerned, and the broader context in which the conduct occurs (§§ 188-189 and 274). The ECJ clarified that lawful conduct under Article 102 TFEU (such as the RSAs) can be considered as part of the relevant context (§ 197).
On the allocation of the burden of proof, the Commission bears the burden of proving an infringement, while the undertaking must prove any defences. However, the provision of compelling factual evidence by one side may require the other to provide an explanation or risk the burden of proof being satisfied (§ 202).
Requisite legal standard to establish abusive tying
It follows from the GC judgment in Microsoft (T-201/04) that conduct can constitute abusive tying if the following conditions are met:
- the tying and tied products are two separate products;
- the undertaking concerned is dominant in the market for the tying product;
- customers are coerced to obtain the tied product with the tying product;
- the practice is capable of generating foreclosure effects; and
- the dominant company has not established that the practice was objectively justified.
The Judgment clarifies that, in principle, tying departs from competition on the merits (§ 269), and the Commission does not have to adduce specific additional evidence to establish that condition (e.g. that the conduct is not replicable by an as efficient competitor).
Interestingly, the ECJ referred back to the Commission's finding that the conduct was not only capable of foreclosure but also that it had actual exclusionary effects (§ 275). The Judgment does not, however, confirm explicitly that this is an evidentiary requirement.
Relevance of as efficient competitors and the AECT
The ECJ addressed both the concept of as-efficient competitor as a standard by which to measure exclusionary effects and also the AECT (i.e. a price-cost test determining whether a hypothetical as efficient competitor could replicate the conduct). These serve different functions and, while the Judgment sometimes conflates the two, it does generally treat them as being distinct.
The Judgment confirms the established case law that the Commission is not required to carry out an AECT (see Post Danmark II, Unilever). However, it goes further in holding that the AECT is not appropriate for non-pricing conduct such as tying (building on a more cautious approach in Unilever).
Turning to the as-efficient competitor standard, this can be used to prove both the existence of conduct that departs from competition on the merits and the exclusionary effects derived from that conduct (§§ 268 and 270). However, the ECJ confirmed its previous case law that the Commission is not systematically required to show that conduct is capable of excluding an as-efficient competitor for an abuse to be found under Article 102 TFEU (§ 272).
Building on AG Kokott's Opinion (§§ 132-133), the ECJ clarified that there are situations where it is "not possible, nor does it make sense" to base the analysis of exclusionary effects on whether an as-efficient competitor could replicate the conduct. This is so in particular when "the structure of the market, involving, for example, an ecosystem characterised by significant barriers and network effects" and "the conduct at issue" make "the entry, maintenance or even emergence of an as-efficient competitor practically impossible" (§ 273). The ECJ observed that the relevant markets in the present case are digital markets where variables such as innovation, access to data, multi-sidedness, user behaviour and network effects are indicative of high barriers to entry and complex interactions that render the emergence of an as-efficient competitor practically impossible (§ 278).
The ECJ stressed that it is all the more so "where the conduct at issue does not lend itself to a quantitative, price-based analysis" (§§ 273 and 278).
Causality and counterfactual analysis
The ECJ confirmed that the Commission is required to establish a causal link between the contested behaviour and its anticompetitive effects. It then clarified that the Commission can rely on "a range of evidence, without being required systematically to use any single methodology, in particular a counterfactual analysis, to prove the existence of such a causal link" (§ 225). In other words, a counterfactual analysis is not essential to show causality.
Therefore, the Commission was not required in the present case (i) to examine the new opportunities created for competitors by the free licensing model developed for Android, which Google claimed was only viable due to the preinstallation conditions in the MADAs; or (ii) to identify a realistic counterfactual scenario which would have resulted in equal or greater distribution opportunities for rivals (see notably §§ 227-229).
Google also questioned the existence of a causal link between the conduct and the alleged anticompetitive effects by arguing that users had the possibility to download competing apps, but refrained from doing so due to the superior quality of Google's products. The ECJ clarified that "there are situations in which it is not possible to analyse whether users' behaviour is attributable to the qualitative characteristics of the dominant undertaking's products". This is so where the conduct (in this case, the preinstallation conditions) creates a positive bias in favour of the dominant company (§ 203).
As a result, it was in principle enough for the Commission to establish that the preinstallation conditions created a 'status quo bias', namely that the conduct distorted the users' choice. Referring to the rules on the allocation of the burden of proof, the ECJ considered that it was for Google to demonstrate the superior quality of its products and that the users' behaviour was attributable to that quality (§ 211).
Single and continuous infringement
Finally, the ECJ found that the partial annulment of the Decision by the GC could not affect the findings of the Commission on the existence of a single and continuous infringement. The ECJ recalled that a single and continuous infringement occurs when different conduct forms part of an 'overall plan', i.e. when the elements have the identical object of distorting competition (§ 381). Moreover, it is possible for the GC to partially annul a decision insofar as such a partial annulment does not cause the substance of that measure to be altered (§ 382).
Against this background, the ECJ found that annulling the Decision’s RSA findings could not undermine the existence of an overall strategy to impose conditions on the Android OS and certain apps and services in anticipation of mobile internet development while preserving Google’s business model (§§ 385-386). That strategy was implemented through the three remaining infringements, which the Commission had correctly proved (§ 382). Accordingly, the RSA portion of the Decision was severable, and the GC was entitled to partially annul it while confirming the existence of a single and continuous infringement (§ 388).
Comments
While the Judgment largely endorses and consolidates the existing case law, it does contain a number of noteworthy developments and clarifications. Disappointingly, it fails to engage properly with a number of arguments made by Google and leaves a number of grey areas that will need to be resolved by future case law.
Tying does not require proof of departure from competition on the merits
As discussed above, it follows from the Judgment that the Microsoft framework continues to apply to the assessment of tying. The ECJ clarified that tying departs 'in principle' from competition on the merits. Accordingly, the Commission does not need to adduce additional specific evidence to show that conduct departs from competition on the merits. This clarification is arguably new and noteworthy, but its practical impact should not be overstated: competition on the merits has never really operated as a self-standing criterion in the assessment of tying, and competition authorities must still demonstrate the conduct's capability to foreclose rivals.
Capability to foreclose: required but with broad evidentiary flexibility
The Judgment confirms that the Commission needs to prove the capability of conduct (including tying) to have exclusionary effects, taking all the relevant circumstances into account. In that sense, the ECJ does not depart from the effects-based approach and does not introduce a presumption that tying has exclusionary effects.
However, the Commission seems to enjoy considerable flexibility when it comes to the type of facts and evidence it can rely on to establish potential or actual exclusionary effects:
- It can take lawful conduct into account as a relevant contextual factor amplifying the effects of other conduct.
- It is not required to run an AECT for tying and the test might, in fact, be inappropriate for such conduct. This echoes the ruling in Unilever (C-680/20) where the ECJ indicated that competition authorities cannot be under a legal obligation to conduct the AECT and that the test may be inappropriate for certain non-pricing conduct such as refusal to supply.
- While it needs to show causality between the conduct and the alleged effects, the Commission can use various tools to do so and does not need to carry out a counterfactual analysis.
A lower evidentiary threshold to show abuses in digital markets?
The Judgment has the most direct implications for digital companies, as it seems to lower the regulators' evidentiary threshold in these markets, as suggested by the following findings:
- Google Shopping (C-48/22) upheld the as-efficient competitor principle, while confirming that an AECT is not a mandatory element of proof; Google Android goes further and finds that it "does not make sense" to analyse exclusionary effects based on an as-efficient competitor in the context of digital markets that are characterised by ecosystems with significant barriers to entry and network effects. That applies "all the more" where "the conduct at issue [such as tying] does not lend itself to a quantitative, price-based analysis." That said, the analysis carried out by the Commission and reviewed by the GC in relation to the exclusivity payments under the RSAs suggests that the as-efficient competitor benchmark, and specifically the AECT, remains an available defence for digital companies at least for pricing conduct. The Guidelines appear to confirm this, addressing the irrelevance of the as-efficient competitor benchmark in digital markets only in their section on non-pricing conduct.
- The ECJ summarily dismissed all of Google’s arguments that its free licensing model should be considered in the assessment. This is a notable confirmation of the GC's findings in this case, being at odds with the obligation to consider all relevant circumstances and contrasting with the approach in, for example, Qualcomm (T-235/18), where the GC criticised the Commission for failing to consider that the customer had no viable alternative to Qualcomm’s chipsets.
- Preinstallation conditions or by-default settings, which are often found in the digital economy, can lower the evidentiary bar to show causality. For such conduct, it is in principle sufficient for competition authorities to establish that the conduct created in favour of the dominant company a status quo bias distorting user choice, in order to show causality. This effectively reverses the burden of proof to the dominant company, which can demonstrate the superior quality of its products and that user behaviour is attributable to that quality.
The ECJ Judgment and the Guidelines on exclusionary abuses
The Commission was clearly waiting for the ECJ's Android judgment before finalising its first Guidelines on exclusionary abuses (Guidelines). The Judgment generally endorses the approach set out in the draft guidelines, including the following points: the presumption that certain conduct, including tying, departs from competition on the merits; and the absence of a requirement for the Commission to carry out a price-cost test except in relation to specific pricing conduct (i.e. predatory pricing, margin squeeze, and certain conditional rebates).
Following the Judgment, the Commission rapidly finalised and published its Guidelines on 3 September 2026. While the Guidelines have evolved following the consultation, the Judgment itself had a limited impact on the final text. Building on the Judgment, the Guidelines incorporate expanded guidance on causality (see Section 3.3.3), highlight the specificities of digital markets and their impact on the assessment of exclusionary practices in these markets (see e.g. § 94), and confirm that lawful conduct can be part of the relevant context (§ 105).
Open questions
A number of open questions remain on the assessment of exclusionary abuses:
- Whether different evidentiary thresholds to show exclusionary effects of tying may apply depending on the circumstances. The ECJ refrained from expressly confirming that the Commission is required to consider actual market conditions in certain situations (e.g. where the tied product is available for free and/or alternatives can be easily found), as the GC had suggested - in line with Microsoft.
- Whether the Commission is obliged to consider certain criteria for conduct other than exclusivity rebates. In Intel (C-413/14), the ECJ confirmed that abuse can be presumed for exclusivity rebates (noting that in that case the Commission applied an AECT and the decision's findings on the exclusivity rebates were ultimately annulled because of flaws in the application of that test). However, it clarified that if the dominant company submits – during the administrative procedure – evidence that its conduct was not capable of restricting competition, the Commission is required to examine that evidence and prove capability to foreclose in light of the extent of the dominant position, the market coverage, the rebates' conditions, their duration and the potential existence of an exclusionary strategy. Moreover, the findings of the GC in the Google Android case on the exclusivity payments also show the importance of certain quantitative criteria, such as market coverage, to prove the existence of foreclosure effects. The GC found that the Commission was wrong to consider a market coverage of maximum 10-20% as "significant" and did not establish in the Decision that the market segment was so strategic that Google's behaviour was likely to foreclose competition despite the small market coverage.
- Whether the Commission can find an abuse where the counterfactual analysis clearly demonstrates that absent the conduct, competitors would not have been in a more favourable position or there would be no potential competitors.
Further guidance on some of these issues is expected from the upcoming ECJ judgments in Google AdSense (C-826/24, appeal in case T-334/19) and BEH (C-14/24, appeal in case T-136/19). In Google AdSense, the Commission is appealing the GC's annulment of its decision arguing that it wrongly assessed the capability of exclusivity clauses to restrict competition. The GC had found that the Commission failed to properly account for the duration of Google's contracts and their market coverage. In BEH, the Commission is challenging the GC's annulment of a refusal-to-supply decision on the ground, inter alia, that the GC wrongly imposed additional evidentiary requirements to assess the causal link between the refusal and its anticompetitive effects. Notably, the GC found that even if an access refusal occurred, it was incapable of restricting competition because competitors could not access the market due to inherent features of the gas infrastructure that were not imputable to the dominant undertaking (§§ 951 and 1097-1098). It is worth noting that the European Commission has refrained from providing any guidance in its Guidelines on the points that remain under appeal in these cases.
Practical implications for companies under investigation
The key practical takeaway for companies facing an investigation is the importance of proactively developing robust evidence to demonstrate the absence of exclusionary effects, regardless of the type of conduct at issue. This includes qualitative evidence (such as short duration of the conduct, or genuine possibilities for early termination), quantitative analysis (such as low market coverage or an AECT where appropriate), and counterfactual analysis showing that competitors would not have been better off absent the conduct.
In line with recent ECJ judgments on exclusionary practices, notably Intel and Unilever, presenting such evidence will oblige the competent competition authority to consider and rebut the arguments and economic analysis submitted, or to provide alternative evidence of foreclosure effects and causation – even where the authority was not initially required to conduct such analysis or could otherwise rely on presumptions of foreclosure. This applies to all conduct except conduct specifically designed to exclude competitors (so-called "naked restrictions" – a term which refers to conduct with no economic interest for the dominant undertaking, other than that of restricting competition, e.g. payments made to customers in return for not selling certain competitors' products).
Authors: Donald Slater, Partner; Jessica Bracker, Senior Associate; and Maria Eugenia Finocchio, Associate.
Other author: Massimiliano Gelmi, Legal intern.
