On 25 February 2025, the Court of Justice of the EU (CJEU) delivered its judgment in Android Auto, which refined the so-called essential facilities doctrine. This doctrine is one of the most debated concepts in (EU) competition law because it challenges the freedom of dominant undertakings to conduct a business by imposing an obligation to deal with third parties. As Advocate General (AG) Jacobs put it in Oscar Bronner, the leading case on refusals to supply, “[t]he right to choose one’s trading partners and freely to dispose of one’s property are generally recognised principles in the laws of the Member States[…]. Incursions on those rights require careful justification” (see paragraph 56, emphasis added). The “careful justification” to which AG Jacobs refers requires a balancing exercise: weighing the dominant firm’s freedom to conduct a business (which is enshrined in Article 16 of the Charter of Fundamental Rights of the EU) against the harm to competition a refusal to supply may cause.
In Oscar Bronner, the CJEU introduced a set of conditions under which a refusal to supply is deemed to qualify as an abuse of dominance under Article 102 TFEU. These conditions are the following: (a) there must be a refusal to supply on behalf of the dominant firm; (b) the product to which access is sought must be indispensable to someone wishing to compete in a downstream market; (c) the refusal to supply must lead to the elimination of effective competition in the downstream market; (d) the refusal to supply must be likely to cause consumer harm; and (e) there is no objective justification for refusing to supply.
Over the years, the EU Courts have sought to clarify the rationale for and scope of application of the “Bronner test”. Despite such refinements, the Bronner test remained a strict test for establishing an abuse of dominance, especially in the case of digital platforms. This is because in digital markets alternatives often exist, rendering Bronner’s indispensability condition challenging to prove. This challenge was addressed in the Commission’s expert report on Competition Policy in the Digital Era which suggested revisiting the test. Arguably, this is precisely what the CJEU did in Android Auto, which this post analyses.
I. Background to and facts of the case
The factual background to the case concerns a dispute between Enel X (Enel), an electric car charging services provider, and Google. Enel had developed JuicePass, an app offering several functionalities relating to the charging of electric and hybrid cars. With the JuicePass app, drivers can search for and navigate to charging points, arrange bookings for charging sessions and perform payments for the use of a charging point.Enel aspired to create a streamlined version of JuicePass that would be compatible (i.e., interoperable) with Android Auto. Android Auto is a piece of software that Google developed for mobile devices running on Android OS to enable users to access apps on those devices directly on the screen of the infotainment system of a motor vehicle. Put simply, Android Auto is a phone projection tool that mirrors a phone’s screen on the car’s dashboard.
Upon request, Google informed Enel that it could not grant access to Android Auto for the reason that the technical template needed for apps like Juicepass had not been developed yet. At the time of the request, there were only templates for media and messaging apps, as well as for Google’s navigation apps (Maps and Waze). Enel requested that Google make its templates compatible with the Juicepass app, which Google refused to do raising (a) security concerns and (b) the justification that Google needs to allocate resources sensibly when developing new templates. Enel subsequently filed a complaint to the Italian competition authority (Autorità Garante della Concorrenza e del Mercato or AGCM), claiming that Google’s conduct constituted an abuse of dominance, which breaches Article 102 TFEU. AGCM’s is available (in Italian) here.
AGCM’S finding of abuse was based on the following considerations. Google controlled whether app developers would reach end users and could unilaterally decide the (types of) apps that could be made available on Android Auto. In addition, its own Google Maps and Waze apps were developed without the restrictions that were imposed on third-party apps. Against this backdrop, the AGCM concluded that Google’s refusal to provide interoperability harmed the competitive process and produced exclusionary effects by granting an unfair advantage to Google’s proprietary app.
Google put forward arguments that interoperability with Android Auto was not indispensable for JuicePass to compete. AGCM rejected Google’s arguments, noting inter alia that, within the Android ecosystem, Google is the sole source of tools for apps that can be used safely and easily when users drive. Moreover, AGCM considered that almost all smart mobile devices with an OS other than iOS use Android (in Italy, Android is used on about three quarters of smartphones and half of tablets). AGCM further found that, for the driving user, the modified and simplified user experience via Android Auto is not substitutable with the full and rich interaction experience of the smartphone, and Android Auto is not substitutable with other technological solutions for the use of apps via the car’s infotainment units.
AGCM also found that Google’s refusal to grant interoperability was likely to eliminate effective competition. AGCM noted that in the period of time for which Google has been able to keep the JuicePass app outside of Android Auto, Google has had the opportunity to build a user base for services related to electric charging via Google Maps on Android Auto whilst preventing Enel X from doing the same. AGCM further noted that the effects of Google’s conduct are exacerbated by the existence of network effects in the market, which is prone to a winner-takes-all scenario. AGCM also considered the specificities of the environments where the app is (or could be) offered. Specifically, AGCM mentioned that apps that provide services related to electric charging (including navigation apps) are intended to be used when the vehicle is on the move so as to allow a charging experience adapted to the needs of e-mobility and not disrupted by the so-called “e-mobility anxiety”. In the Android ecosystem, the context of choice for the use of the JuicePass app is the controlled and simplified environment of Android Auto and not the complex and interaction-rich context of the smartphone. Accordingly, the use of JuicePass outside of Android Auto is entirely marginal and, in any case, not sufficient to allow the developer to create a sustainable user base.
AGCM further found that, by hindering and delaying the entry of an app that would have made it possible not only to search for, identify and reserve a socket for recharging but also to book and pay for recharging, Google harmed consumers who were thereby unable to access the different and broader range of services offered by JuicePass. Moreover, according to AGCM, since the services in question are necessary for electric mobility, Google’s conduct might have been detrimental to a more rapid uptake of electric vehicles (and, by extension, to the transition towards a more environmentally sustainable mobility). Google’s conduct was also found to affect Enel’s business model because it deprived Enel of valuable user-generated data capable of improving its commercial and technical operations. On that account, Google’s conduct to obstruct the emergence of a new app for which there is demand was found to hinder innovation and technical advancement.
Based on the above, the AGCM imposed on Google a fine of over 100 million EUR. Arguably more importantly, Google was required to (a) refrain from engaging in practices similar to the conduct that was found to infringe Article 102 TFEU; (b) promptly take action to ensure the interoperability of JuicePass with Android Auto, and (c) submit a proposal for the appointment of a Trustee and provide the latter with all the necessary information to monitor the implementation of the obligations imposed by the AGCM.
It has been argued that, compared to the CJEU’s judgment in Android Auto (which is analysed further below), the assessment conducted by AGCM provides a preferable approach to cases of refusals to deal in digital markets.
Google first appealed the decision of the AGCM before the Regional Administrative Court of Lazio, which dismissed Google’s action in its entirety. The case reached the Italian Council of State, which decided to stay the proceedings and refer questions to the CJEU regarding the interpretation of the Bronner test. The questions referred to by the Council of State enable the Court to clarify the applicability of Bronner (a) in general and (b) to digital platforms specifically.
Before discussing the CJEU’s judgment in Android Auto, in order to grasp the implications of Android Auto for future cases, it is important to briefly revisit the evolution of the essential facilities case law.
II. The evolution of the Bronner case law up to Android Auto
As mentioned above, under Oscar Bronner, a refusal to supply is abusive if the following conditions are fulfilled: (a) there must be a refusal to supply on behalf of the dominant firm; (b) the product to which access is sought must be indispensable to someone wishing to compete in a downstream market; (c) the refusal to supply must lead to the elimination of effective competition in the downstream market; (d) the refusal to supply must be likely to cause consumer harm; and (e) there is no objective justification for refusing to supply.
In the years that followed this judgment, the Court has expanded Bronner’s scope to the area IP-protected facilities. Notably, IMS Health and Microsoft clarified the possibility of breaking the exclusivity of access that would otherwise be protected by IP rights (this was first discussed in Magill). To eliminate this additional layer of legal protection, which would normally be afforded to the (dominant) undertaking holding the IP rights concerned, the Court added the ‘new product/new development’ criterion in the test for abuse. This criterion presumably protects the incentive of the dominant undertaking to innovate.
In Slovak Telekom, it was indicated that Oscar Bronner would not apply where the restriction of access does not result from an outright refusal but rather from a so-called constructive refusal. The latter should rather be treated as an abusive unfair trading condition (which is also caught by Article 102 TFEU) that renders access to the respective facility non-viable economically. This is sensible, given that the choice of dominant undertakings to deal (or not) with third parties is protected by the Charter of Fundamental Rights of the EU (whereas the ability to impose unfair trading conditions is not).
Therefore, in cases such as Slovak Telekom, balancing competition against fundamental rights and the incentive to innovate is not necessary. Furthermore, when access to the facility is covered by secondary or national legislation, the balancing act in Bronner is presumed to have been part of the legislative process. In these circumstances, there would be no room for repeating the process in the context of a competition law case (for more details see here). This point was also confirmed in Lithuanian Railways, which added that the application of Oscar Bronner may also depend on whether the respective facility resulted from an entrepreneurial risk. This is because, in addition to protecting fundamental rights, the Oscar Bronner criteria have been developed to ensure that undertakings are incentivized to invest in creating such facilities. In situations where the respective facility came into being through means other than entrepreneurial activity, such as public funding, this aspect of entrepreneurial risk and the incentive to invest and innovate would not exist. Finally, the Court also ruled that destroying one’s facility is not equivalent to a refusal to supply within the meaning of Oscar Bronner which intends to do quite the opposite (i.e., capture conduct which seeks to ensure that the facility is reserved for the exclusive use of the dominant undertaking) – for an analysis of this judgment see here.
More recently, in Bulgarian Energy, the General Court of the EU (GCEU) reiterated the relevance of secondary or national regulatory frameworks when determining the applicability of Oscar Bronner, as well as the distinction between different types of refusal (outright vs. constructive). The GCEU also looked into the relevance of the legal status the dominant undertaking has with respect to the respective facility. The GCEU ruled that ownership is not required; an exclusive right to make use of the facility is sufficient. In this respect, the fact that the dominant undertaking paid a fixed fee to obtain this exclusive right constituted an investment and, by extension, an entrepreneurial risk in the sense of Oscar Bronner. Accordingly, the balancing of interests under Oscar Bronner may not be limited to those who developed the facility but also to those who obtained exclusive control over it through other commercial routes.
III. The CJEU’s judgment in Android Auto: Re-shaping Oscar Bronner
The questions referred to by the Italian Council of State in Android Auto allowed the CJEU to tackle multiple aspects of the Oscar Bronner line of case law.
A. Indispensability and access by design (paras. 33-52)
The condition of indispensability is the most difficult to fulfil. In practice, it makes it difficult for claimants and competition authorities to establish the existence of an abusive refusal, even when market conditions are suboptimal. This is particularly true in the case of digital platforms where indispensability in the sense of Oscar Bronner is unlikely to be met because there will often be alternatives to the service or product to which access is sought. According to Oscar Bronner, indispensability requires that no such alternatives exist, or that such alternatives are unlikely to exist. In the case of platforms, this criterion can lead to an arbitrary outcome whereby access to platforms with a market share of 90% or more is not considered indispensable.
To address this weakness, the CJEU indicated that the relevance of the indispensability criterion depends on the nature of the facility to which access is denied. According to the CJEU, Android Auto is a platform designed to (inter)operate with third-party service providers. Accordingly, it was intended to be open to third parties by design. The CJEU ruled respectively that “[a] digital platform intended to enable the use […] of apps developed in particular by third parties and downloaded on users’ mobile devices cannot be regarded as having been created solely for the needs of that undertaking in a dominant position” (see paragraph 48).
With this new approach, the CJEU eliminates the previously insurmountable barrier of the Bronner test. This is because all platforms are inherently designed to be open to third parties. While this approach may resolve the friction between the Oscar Bronner line of case law and digital platforms, it must be qualified. A more nuanced approach to this matter would be to apply the logic of the CJEU’s ruling in Android Auto to the specific aspects of the business model and intended use of the platform under consideration.
In the case of Android Auto this would mean that Google would not be able to rely on Oscar Bronner when it comes to apps that are meant to be distributed within a car’s infotainment system but could do so with respect to apps that have little or nothing to do with the intended use of such a system (e.g., a stock exchange app). Until this matter is resolved, the Android Auto judgment is a warning sign for dominant platforms that the Bronner test may no longer shield them against third-party access requests.
B. Restriction or elimination of competition in light of the success of the third-party app (paras. 53-61)
In Android Auto, the CJEU was asked to indicate what happens in a situation where the undertaking that is being refused access to the facility manages to compete effectively and even grow in popularity. According to the CJEU, such circumstances do not, in principle, preclude the finding that the refusal to supply (or other practice) is abusive. According to the CJEU, Article 102 TFEU is concerned with the ability of the practices to produce exclusionary effects, which need not materialize. This is a sensible approach because, as the CJEU notes, there can be numerous reasons why anti-competitive strategies may ultimately fail. That alone should not make them any less abusive. Furthermore, it cannot be excluded that, in the absence of an unsuccessful anti-competitive strategy, competition could have been more effective.
Moreover, according to the CJEU, when establishing the exclusionary potential of the refusal under consideration, it is not necessary to accurately define the relevant market where such exclusionary effects can occur. Again, this is a sensible approach, given that in the case of refusals to supply, the undertaking seeking access to the facility may wish to enter a market that does not yet exist.
C. Objective justification – technical impossibility and temporary nature (paras. 69- 81)
In Android Auto, the CJEU ruled that the integrity and security of the platform could constitute an objective justification for the refusal. The same applies to cases where it is technically impossible to facilitate interoperability.
However, the dominant undertaking cannot rely on technical reasons to prevent interoperability ad infinitum. While this process may take some time (e.g., in the case of Android Auto, a specific template was required), this grace period is not unlimited. The permissible length of such a period will differ from case to case and must strike a balance between the needs of the undertaking requesting interoperability and the technical challenges facing the dominant undertaking. In other words, the absence of built-in interoperability serves only as a temporary basis for permissible refusal. Importantly, the CJEU ruled that facilitating interoperability with the platform may involve development costs. In such cases, the dominant undertaking may be allowed to recover the costs it incurred by imposing FRAND fees once the development has been completed.
IV. Conclusion and outlook
Android Auto is arguably the most impactful judgment since the extension of Oscar Bronner to IP-protected facilities. Through its findings, the CJEU future-proofs the essential facility for the digital economy in ways that were not considered possible without dedicated regulatory rules, such as those introduced by the DMA. Contrary to the DMA, however, Android Auto applies to all dominant platforms (i.e., the scope of the judgment is not restricted to undertakings providing the “core platform services” within the meaning of the DMA). Overall, the CJEU significantly restricts the possibility of dominant platforms to leverage their power across markets.
Co-authored with Daniel Mandrescu.