A Landmark Judgment in the Refusal to Deal Doctrine: Android Auto (Case C‑233/23) and Bronner’s Future in Digital Platform Markets

Table of Contents: 1. Introduction. – 2. The Bronner criteria and their relevance in digital platform markets. – 3. Factual background to the dispute. – 4. The decision of the ICA. – 4.1. Market definition. – 4.2. The ICA’s assessment. – 5. Google’s Appeal to the Administrative Court. – 6. Google’s appeal to the Supreme Administrative Court. – 7. Analysis of the Court’s findings. – 7.1. The application of Bronner in cases involving infrastructures developed to enable third parties to use them. – 7.2. The role of actual effects in determining whether the conduct is capable of distorting competition. – 7.3. Objective justifications as a means to protect the dominant firm’s incentives to invest. – 7.4. Defining the relevant downstream market in order to establish anticompetitive effects. – 8. The judgment of the Consiglio di Stato. – 8.1. The Consiglio di Stato’s findings concerning the open nature of the platform. – 8.2. The Consiglio di Stato’s findings about potential anticompetitive effects. – 8.3. The Consiglio di Stato’s findings about objective justification. – 8.4. The Consiglio di Stato’s findings about market definition. – 9. Implications and unresolved issues going forward. – 9.1. Should the Court’s judgment be limited to digital platforms? – 9.2. Problems with defining a platform developed in order to enable third parties to use it. – 9.3. The assessment of effects going forward and insights from the Consiglio di Stato’s judgment. – 10. Conclusions.

Abstract: The Court of Justice in Android Auto established that certain dominant digital platforms may no longer rely on the strict Bronner requirements when outright refusing access to their infrastructure. The impact of this landmark judgment remains to be seen and issues such as its scope and relevant legal tests are still unclear. This paper explains the factual background to the dispute and provides a critical analysis of the Court’s key findings. It then assesses the judgment’s wider implications and argues that it should not be extended beyond digital platform markets. Furthermore, the uncertainty surrounding how to determine whether a platform was developed to enable third parties to use it is also discussed. Last, this paper assesses how the effects assessment may be carried out in future outright refusal cases involving open digital platforms.

Keywords: refusal to deal – digital platforms – Bronner – indispensability – anticompetitive effects – innovation.

1.   Introduction

On 25 February 2025, the Grand Chamber of the Court of Justice issued its judgment in the context of Case C-233/23 Android Auto.[1] The judgment is the latest guidance provided by the Court of Justice to the Italian Consiglio di Stato and, much like the previous recent judgments stemming from proceedings before the same court, it has the potential to greatly impact the future enforcement of Article 102 of the Treaty on the Functioning of the European Union (TFEU), in particular in the context of digital platform markets.[2] The judgment addresses not only important questions pertaining to the application of the Bronner criteria but also, to some extent, raises doubts as to what role it will play in the future of application of Article 102 TFEU. In judgments such as Telia Sonera and Slovak Telekom, the Court of Justice has avoided the application of Bronner to access scenarios whereby the dominant firm grants access to its infrastructure but on unfair terms.[3] The Court’s judgment in Android Auto continues this trend of dispensing with the indispensability criteria in Bronner but takes it further.[4] The judgment now challenges its application to outright refusals to deal, the quintessential refusal to deal, which considerably infringes the fundamental rights and innovation incentives of undertakings. Its implications going forward remain unclear and key questions such as whether it may be extended beyond digital platforms are still open.

2.   The Bronner criteria and their relevance in digital platform markets

The so-called Bronner criteria are a set of requirements which are used to determine whether competition authorities can impose a duty to deal on a dominant company. These requirements were outlined in the judgment and in the AG opinion in the Bronner case.[5] These criteria postulate that in order to impose a duty to deal on a company, the Commission or the relevant competition authority must find that (i) the infrastructure to which access is sought is indispensable to carrying out the business of the party seeking access; (ii) failure to provide access would likely eliminate all competition in the market on the part of the party seeking access; and (iii) the refusal to grant access is not objectively justified.

The Bronner criteria primarily seek to balance short-term competition with long term innovation incentives.[6] By requiring that the upstream infrastructure is indispensable and that the requesting undertaking be eliminated from the downstream market, the innovation incentives of dominant firms are to a greater extent protected as it is more difficult for rivals to gain access. At the same time, downstream competition is protected in the case that rivals are totally eliminated from the market and therefore prevented from developing their own products which may also be the result of their own innovation.[7] The strict Bronner criteria may be understood as generally favouring the promotion of long-term innovation over short-term competition.[8]

In the context of digital platform markets, the role of the Bronner criteria has significant relevance due to the fact that dominant digital platforms often control upstream digital infrastructures which downstream competitors require access to in order to compete.[9] Furthermore, digital platform markets are often characterised by strong network effects and economies of scale which has led, in certain circumstances, to only one dominant player controlling access to almost the entire upstream market.[10] In the Google Shopping case for example, downstream comparative shopping services, which were competing with Google’s own comparative shopping service, required access to Google Search on the upstream general search market in order to compete.[11] In Android Auto, similar to Google Shopping, downstream competitors also require access to an upstream digital infrastructure in order to offer their products.

The implications of not applying the Bronner criteria to refusals to deal in digital platform markets are significant as they may positively impact downstream competition and innovation by allowing third parties to develop competing products which may in turn benefit consumers. However, it may also stifle innovation in markets for digital infrastructures and in turn also harm consumers since downstream competitors can easily gain access and benefit from the dominant firm’s investments. This is one of the key issues that the Android Auto judgment raises.

3.   Factual Background to the Dispute: The Decision of the ICA to Open Proceedings

The Case originated from a complaint brought forward by Italian major gas and electricity distributor Enel S.p.A. (ENEL). The Italian Competition Authority (ICA) opened proceedings on 8 May 2019.[12] In its opening decision (Opening Decision), the ICA outlined its preliminary findings as below. On 16 May 2019, the day on which the Opening Decision was notified to Google Italy S.r.l., the ICA carried out unannounced inspections.

According to the ICA in its opening decision, ENEL distributes an app called ENEL X Recharge (ENEL X)[13] which allows users to for example find electric vehicle charging stations provided by ENEL and by other operators, access information relating to said charging stations, and reserve and pay for them. ENEL X is distributed via the Google Play Store and the Apple App Store. ENEL X was developed to comply with the guidelines issued by Google on ‘Car app quality’.[14] However, when ENEL asked Google to integrate ENEL X into Android Auto, Google replied that this was not possible as ENEL X didn’t fall within the categories of ‘media’ or ‘messaging’, which are the only categories of applications allowed to be integrated with Android Auto. As an alternative Google offered to either (i) integrate Google Maps with the information provided on recharging stations by ENEL X or (ii) use Google’s own Software Development Kit to develop ad hoc apps for the different car manufacturers.[15]

ENEL rejected the proposal and insisted on integrating ENEL X with Android Auto. In doing so, ENEL pointed out that Android Auto allows for apps which are not related to ‘media’ or ‘messaging’ to be integrated into Android Auto. Specifically, Google Maps and Waze (both owned by Google) are integrated into Android Auto and don’t fall within the above categories. ENEL also stressed that ENEL X had been developed to ensure the safety of drivers by allowing interactions between user and app in the car only via voice commands. Google reiterated its denial, by pointing out that ENEL X didn’t fall within the categories of ‘media’, ‘messaging’, or ‘navigation’, which are the only ones allowed to be integrated with Android Auto. Furthermore, ‘for reasons related to the safety of users and for other technical reasons’ Google could not allow other types of apps or other ‘navigation’ apps on Android Auto.

4.   The Decision of the ICA

On 27 April 2021, the ICA issued a decision fining Google for over 100 million Euros[16] in relation to its denial to integrate ENEL X into Android Auto (ICA Decision).

4.1.  Market definition

The ICA defined the following markets: (i) the market for the licensing of smart mobile OSs; (ii) Smartphone Projection Apps; (iii) OSs for infotainment units; (iv) apps related to EV charging (including navigation and other functions related to EV charging). The ICA defined the market in which Android is active, by reference to the previous decision of the European Commission concerning Android[17] as the ‘the licensing of smart mobile OSs’,[18] but also that app stores for other licensable smart mobile OSs do not belong to the same product market as Android app stores,[19] in line with the Google Android decision.[20] In relation to Android Auto, the ICA considered it as a ‘smartphone projection app’ i.e., ‘an app that projects the content of the screen of the mobile device on the infotainment unit of the car and transmits the inputs from the infotainment unit (through touchscreen, buttons, etc.), to the mobile device, which responds to these inputs’.[21] While the ICA viewed Android Auto and Apple CarPlay as performing the same functions, it considered that they performed the same functions for separate ecosystems, meaning that they were not substitutable with one another.[22] The ICA considered also whether MirrorLink, an app developed by a consortium of car manufacturer and with somewhat similar functionalities could be considered as a competitor to Android Auto, ultimately concluding that that was not the case.

4.2.  The ICA’s assessment

In its defence, Google first considered that the reference to the market definition adopted in Google Android was incorrect as that conclusion was based on the fact that iOS was not licensable, whereas Apple Car Play is licensable.[23] The ICA responded that the fact that Apple Car Play was licensable didn’t change the conclusion as iOS and Android underpin different ecosystems.[24] Google argued that Google Maps and ENEL X were not competitors as the apps had different functionalities[25] and that it had no interest in the flux of data relating to the use of EV charging stations.[26] The ICA refused the argument by stating that both apps ‘offer services of search and navigation related to EV charging stations (actual competition) and, in addition, ENEL X offers new functionalities which could, in the future, be integrated into Google Maps (potential competition); in addition, ENEL X and Google Maps compete for users and for the data generated by said users’.[27] Google argued also that Android Auto was used only by a minority of drivers worldwide, thereby not making it a necessary gateway to access users.[28] The ICA considered that there were network effects and ‘winner takes all’ dynamics and that Google was a ‘gatekeeper for Android users as (…) Android Auto is the access point for final users for app distributors’.[29] Google submitted that the refusal to publish ENEL X was due to safety reasons. The ICA rejected this argument on the basis that ENEL X didn’t create additional security risks compared to Google Maps[30] and that, in any event, any risk could have been managed.[31]

5.   Google’s appeal to the Administrative Court

Google appealed the decision to the Administrative Regional Court in Rome (TAR), which rejected the appeal on 6 April 2022 by ruling in favour of the ICA. The first argument brought forward by Google concerned market definition on the ‘upstream’ market. The TAR rejected it by concluding that the ICA’s reference to Google Android was appropriate.[32] The second and third arguments brought forward by Google concerned competition between ENEL X and Google Maps. The TAR considered that the ICA’s assessment was correct both in finding that Google Maps’ and ENEL X’s functionalities overlapped in so far as they concerned the navigation to EV charging stations as well as to consider that there was competition for the acquisition of data generated by the users.[33] The fourth argument raised by Google concerned the requirement of ‘indispensability’. The TAR rejected this argument considering the evidence put forward by the ICA, such as the responses to car manufacturers, and considering the dynamic nature of digital markets.[34] The TAR considered as relevant that car manufacturers have not developed similar systems as consumers want to access infotainment functionalities via their smartphones.[35] Finally, the TAR rejected Google’s objective justifications related to the additional safety concerns relating to ENEL X compared to Google Maps or Waze.[36]

6.   Google’s appeal to the Supreme Administrative Court

Google appealed the judgment of the TAR to the Italian Supreme Administrative Court (Consiglio di Stato). Appeals to the Consiglio di Stato can be made only on points of law and the Court is considered to be the highest level of review, meaning it is obliged to refer a case to the Court of Justice if the relevant conditions are met. The Consiglio di Stato decided, on 7 April 2023, to refer the five questions below to the Court of Justice.

The first question concerned the meaning of indispensability i.e., whether indispensability included also access to an infrastructure which would make it ‘more convenient’ to compete in a neighbouring market. The second question concerned the extent to which actual effects were required to establish an abuse. The third question concerned objective justifications. The fourth question concerned the need for the dominant company to modify its products or to develop new ones to allow interoperability. The fifth question concerned the need for a competition authority to define specific markets in order to find that a conduct is abusive.

7.   Analysis of the Court’s findings 

7.1.  The application of Bronner in cases involving infrastructures developed to enable third parties to use them

7.1.1.   Findings of the Court

With regards to the first question concerning the application of Bronner to digital platforms, the Court followed the Advocate General’s opinion. It sought to clarify, like the Court has done in previous cases, that Bronner is limited to very specific circumstances where the dominant firm had developed the infrastructure for the needs of its own business and owns that infrastructure.[37] In those settings, the imposition of a duty to supply has the consequence of forcing an undertaking to contract with a rival.[38] Hence, this has significant consequences for the freedom of contract and right to property of an undertaking as well as a negative impact on its long term incentives to invest and innovate which in turn is detrimental to consumers.[39]

The Court however agreed with the Advocate General that when that infrastructure is developed with the purpose of allowing third parties to use it, the Bronner criteria are not applicable because neither the freedom of contract, right to property nor the incentives to invest in high quality products of the undertaking are harmed to the same extent.[40] Furthermore, since the costs assumed in developing the infrastructure were not solely for the undertaking’s own needs but also for third parties to use the infrastructure, requiring it to grant access does not fundamentally alter its economic model.[41]

In order to determine whether this is the case, the Court provides that it is necessary to assess whether the infrastructure 1. was developed solely for the needs of its own business and 2. it is owned by that undertaking or whether on the contrary, it was developed in order to enable third parties to use that infrastructure which may be supported by the fact that access had been granted to those third parties.[42] In the context of the case, the Court also found that Google had made its platform compatible for specific third party apps which those third parties had developed.[43] In addition, the fact that it was a digital platform intended to enable use of apps developed by third parties further suggested that Bronner was not applicable.[44] Last, the Court also held that in the case that Bronner is not applicable, it is necessary to establish whether the conduct produces actual or potential effects of excluding, delaying or obstructing the development of a market which is at least potentially in competition with the product or service of the dominant firm and which is outside the scope of competition on the merits.[45]

7.1.2.   Analysis

As a preliminary point, it is important to note that the Court reinterpreted the first question from the Consiglio di Stato which actually concerned the interpretation of indispensability and whether the fact that the infrastructure was more convenient was relevant in this assessment.[46] Instead, the Court chose to interpret the question in a way that sought to determine whether indispensability was applicable at all.[47] Presumably, the Court was aware that a response to the question in light of Bronner might have likely resulted in Google’s infrastructure not being considered indispensable. An alternative that is less convenient would most probably fall outside the scope of Bronner which interprets indispensability as requiring there to be no alternatives, even if they are less advantageous.[48] Furthermore, in platform markets, it is uncommon that a platform will be indispensable since upstream alternatives will often still exist, even if they are less favourable.[49] Hence, if the Court were to respond by saying that convenience was not a relevant consideration and therefore indirectly that Android Auto was not indispensable, this would result in indispensability still being applicable to platforms outright refusing access. The Court may therefore have chosen to reinterpret the question in order to address the concern that platforms have a quasi-regulatory or gatekeeper function, even if not indispensable, and therefore specific rules for platforms are necessary.[50]

With regards to Bronner not being applicable to refusals to grant access to an infrastructure which was developed for third parties to use, this is a significant shift in the case law on refusals to deal.[51] Some would argue that this shift in the case law is an inevitable outcome in the context of fast evolving dynamic industries which are different from the traditional industries in which Bronner took place.[52] Additionally, it has been argued that this shift was not so significant when considering the Court’s findings in Slovak Telekom and Google Shopping.[53]

The significance in the shift should not be underestimated. In previous cases where the Court deviated from the application of Bronner, they did not involve outright refusals to deal. In both Slovak Telekom and Google Shopping for example, which involved unfair access terms, Bronner was held to be inapplicable because access had already been granted to the infrastructure which was considered less detrimental to the freedom of contract of the dominant firm.[54] In those cases, access had already been granted to a specific competitor and therefore it is more justifiable to argue that a duty to deal is less detrimental for the dominant firm’s right to property and innovation incentives. In outright refusal cases, the dominant firm has never granted access to the requesting competitor and therefore even if it has granted access to other competitors, obliging it to grant access is far more detrimental to its property rights and innovation incentives. The extension of the reasoning in cases like Slovak Telekom and Google Shopping to outright refusals should therefore be carried out with caution.[55]

By extending the Court’s reasoning in Slovak Telekom and Google Shopping to outright refusals, this could lead to even greater uncertainty surrounding the possibility for firms to recoup their initial investment in the infrastructure if rivals, which were never given access in the first place, no longer have to prove indispensability.[56] Furthermore, forcing digital platforms that are often successful due to considerable innovation and risky investments to share the benefits with competitors in these circumstances is likely to significantly diminish investment incentives, regardless of whether the infrastructure was designed to be open or not.[57]

The Court duly sought to take into consideration the interests that Bronner aims to protect when justifying a deviation from its application. Unlike the Advocate General who mostly considered the investment incentives when justifying the deviation from Bronner, the Court welcomely referred also to the freedom of contract and right of property of the dominant firm.[58] As explained, in Slovak Telekom and Google Shopping, the freedom of contract was referred to in order to justify a deviation. In Baltic Rail, Advocate General Rantos in his opinion was very clear that the economic incentives to invest in an essential facility would not be affected if a dominant firm destroyed its own property.[59] He also considered the legal perspective and found that the dominant firm did not invest in the infrastructure and was not in fact the owner of the property therefore there was no plausible harm to its right to property and freedom of contract.[60] The importance of taking both the legal and economic considerations into account is fundamental for justifying any deviation from the Bronner requirement as they constitute its underlying rationale. It is worth noting that the Court does not clarify why the freedom of contract and right to property are not infringed. Nevertheless, similar to how the long-term investment incentives would be affected to a lesser extent, it would arguably be less detrimental to the freedom of contract and right to property since the dominant firm created the infrastructure with the intention of contracting with third parties.

The argument that the platform was built for the purpose of allowing third parties to gain access is similar to the Google Shopping judgment where it was considered that Google’s favouring of its own comparative shopping service was not consistent with its general search service which was to offer the most relevant search result at the top of the page.[61] In that case, the argument was more linked to competition on the merits and the fact that Google’s conduct was not normal in the context of the market.[62] This is equivalent to a no economic sense test which essentially asks whether the conduct of the dominant firm made economic sense in the context of the market.[63] Similar arguments were in fact used by the General Court in Microsoft which rejected Microsoft’s justifications for its refusal to deal by finding that Windows became more valuable and attractive to customers by being interoperable with several work group server operating systems.[64]

With regards to the legal test for assessing whether the denial of access to an infrastructure is abusive, the Court lays out some form of structure. The first step requires determining if Bronner is applicable by assessing whether the infrastructure was developed solely for the needs of the undertaking’s business and whether it is owned by the undertaking. The second step requires assessing whether the refusal produces actual or potential effects of excluding, delaying or obstructing the development of a market which is at least potentially in competition with the product or service of the dominant firm and which is outside the scope of competition on the merits.

Regarding the first step of the assessment, the Court appears to suggest that if the dominant firm has developed the infrastructure for third parties to use, it does not own the infrastructure. This could be interpreted as suggesting that there is some form of co-ownership with digital platforms and its users if that platform is made for users to build for example apps on top of it. The Court seems however to later state that, on the contrary, a dominant undertaking does own the platform infrastructure which it refuses to grant access to.[65] Nevertheless, this requirement of ownership of the infrastructure appeared in Baltic Rail where the Court held that Bronner is applicable if the dominant firm owns the infrastructure.[66] The ownership point was brought up by the Court in Baltic Rail to support not applying Bronner because the dominant firm was not the owner of the infrastructure but rather managed it.[67] Hence, a firm that only manages the infrastructure does not benefit from an exclusive property right which rewards investment and innovation in the same sense as Bronner.[68]

As regards the second step, the Court does not provide much clarity on the extent of the effects assessment required. It is clear that elimination of all competition as established in IMS Health is not necessary however to what extent competitors must be excluded remains unclear.[69] The effects assessment does not seem to focus so much on whether the conduct hampers the ability of rivals to compete and grow their market share but rather on whether the conduct prevents the development of a competing product.[70] There is also no mention of whether the fact that the infrastructure is indispensable might influence the assessment of effects.[71] Furthermore, it might be the case that even though the firm is dominant, there are sufficient alternatives which means that rivals requesting access are still able to apply competitive pressure. In this regard, it is not clear to what extent a rival with a competing product has to be able to apply competitive pressure for anticompetitive effects to exist.[72]

It is also worth adding that the requirement of the conduct not fulfilling competition on the merits in the second step arguably adds little value. As explained above, a platform that refuses to grant access to the infrastructure which it has developed for third parties is behaving in a way that goes against normal competition. Hence, the platform is developed for third parties to use. It suggests that the conduct is already outside the scope of competition on the merits. However, it might be the case that if the platform altered several times between being closed and then open, this might suggest that closing the platform was normal in the context of the case.

7.2.  The role of actual effects in determining whether the conduct is capable of distorting competition 

7.2.1.   Findings of the Court 

When addressing the second question, the Court also followed the Advocate General opinion and held that actual exclusionary effects do not need to be demonstrated.[73] Therefore, even though competitors may grow their market share after the conduct is implemented, their growth in market share could be caused by factors other than the conduct of the dominant firm.[74] In this regard, the Court found that the ability of competitors to mitigate the effects is not relevant for determining an abuse and furthermore, absent the conduct in question, competition may have grown even more.[75] However, the absence of actual effects may still constitute evidence for establishing that the conduct was not capable of producing anticompetitive effects.[76]

7.2.2.   Analysis

The Court’s findings are consistent with its judgment in Servizio Elettrico Nazionale.[77] They are unsurprising and do not provide a great deal of room for discussion. With regards to the Court stating that the assessment of effects cannot depend on the ability of competitors to mitigate the effects of the conduct, this provides further clarity as to how the assessment of effects may be carried out. As already discussed, the focus of the Court appeared to be primarily concerned with whether the refusal would lead to the delay or obstruct the development of competing products. There was little clarification as to how the refusal might affect a rival’s ability to compete and their ability to offset the harm. By referring to the ability to mitigate the effects of the refusal, the assessment of effects in these types of cases involving digital platforms requires a more thorough analysis of the impact on competition.

In refusal to deal cases, the ability of a competitor to offset the harm caused by the dominant firm should be a relevant consideration. Furthermore, the fact that the upstream infrastructure is not indispensable suggests that the effects could be offset by using competing upstream platforms or even replicating the platforms upstream infrastructure.[78] This suggests that the ability to offset the effects in the context of refusal cases should have a more significant role.[79] It may even be argued that contrary to the Court’s findings, the fact that there were alternative upstream infrastructures allowing rivals to compete could be a determinative factor.

7.3.  Objective justifications as a means to protect the dominant firm’s incentives to invest 

7.3.1.   Findings of the Court

In response to the third and fourth questions concerning possible objective justifications, the Court held like the Advocate General that if access was to compromise the integrity or security of the platform or if there were other technical reasons making it impossible to grant interoperability, this could justify a refusal.[80] The absence of a specific template for granting access or difficulties in developing that template are not sufficient justifications.[81] When assessing whether a refusal is justified, factors that are relevant may be technical difficulties in developing a template for certain apps and constraints pertaining to the absence of human resources to develop the template within a short period of time.[82] If this assessment leads to the conclusion that there are no objective justifications, the dominant firm may charge a fair and proportionate fee for providing interoperability.[83] Last, the burden of proof regarding objective justifications initially falls on the dominant firm but is then passed on to the competition authority which must demonstrate that those justifications cannot be accepted.[84]

7.3.2.   Analysis

Although not specified by the Court, the Advocate General considered that the assessment of the proportionality of objective justifications relating to for example time or resource constraints required both the balancing of the need for open markets with innovation incentives.[85] Similarly, the Advocate General saw the ability of the dominant firm to request a financial contribution as a way to protect innovation incentives.[86] The importance of objective justifications in the context of refusals to deal involving open digital platforms becomes far more important now that the protection granted to dominant firms under Bronner is not applicable.[87] Objective justifications in the context of refusal to deal cases are often raised but have rarely been successful in exempting the dominant firm from the application of Article 102 TFEU. With this judgment, there will be a far greater focus on objective justifications. In fact, dominant firms will be able to rely on their innovation incentives in this part of the assessment even if indispensability is not applicable.

The Court, which may have wisely avoided specifically referring to innovation considerations when assessing these objective justifications, will nevertheless likely have to take them into account in future cases. For example, a dominant firm can raise arguments to suggest that the technical difficulties that exist are valid justifications as otherwise it would have no incentive to innovate. Competition agencies and courts might therefore struggle in finding the correct or optimal balance. They may end up for example placing more weight on innovation incentives in order to compensate for not applying Bronner. It will also be interesting to see how the balancing of innovation incentives and open markets will be carried out when dealing with super-dominant platforms that benefit from strong network effects.[88] Technical difficulties may have less weight in markets with these characteristics.

Even though Bronner was held to be inapplicable, the dominant firm may still be able to raise arguments regarding innovation incentives more generally and even potentially its right to property. For example, in Microsoft, innovation incentives were raised as an objective justification for refusing to provide interoperability.[89] In Flughafen Frankfurt, the dominant firm argued as an objective justification that the obligation to deal limited its ability to exercise its right to property.[90] In the actual case, obliging a platform to grant access to its operating system concerns the central purpose of the property right.[91]

Although it may be argued that innovation incentives and the right to property have already been considered when assessing the application of indispensability, these considerations may still be relevant under objective justifications. The application or non-application of indispensability is a way to balance innovation incentives and the right to property with undistorted competition. In the context of objective justifications, even if indispensability is not applicable, innovation incentives and the right to property in the specific circumstances of the case may still be relevant. For example, if a certain platform refused access to all of its rivals in a certain downstream market, even if the platform was not indispensable and anticompetitive effects were produced, it might be the case that obliging the platform to grant access to all of its rivals would significantly undermine its incentives to invest when considering the specific circumstances of the case.

7.4.  Defining the relevant downstream market in order to establish anticompetitive effects

7.4.1.   Findings of the Court

With regards to the last question on the need to define the downstream relevant market in the context of a refusal to deal, the Court again followed the Advocate General’s opinion. It held that it is necessary to distinguish two separate markets but defining the relevant market in light of the Commission’s market definition guidelines is not necessary.[92] Furthermore, the Court considers, in line with IMS Health, that it is sufficient to identify a potential or hypothetical market, in particular in markets which are subject to rapid change.[93]

7.4.2.   Analysis

The Court’s response to this question comes as little surprise. Since IMS Health, it has been clear that two separate (potential or hypothetical) markets must be distinguished in order for there to be a refusal to deal. The clarification in this case was really about the extent to which this downstream market had to be defined with the Court finding that a more thorough relevant market definition was unnecessary. The question was raised by the Consiglio di Stato because with a relevant market definition, Google’s own map service and the JuicePass map service may not have actually been in competition with each other and hence the conduct could not have produced anticompetitive effects. This however will be a welcome finding for competition authorities who will have to spend less resources on providing thorough analyses of downstream markets.

Although the relevant market definition was considered unnecessary, the underlying logic for its application as provided in the case law may help the assessment of effects.[94] For example, if one were to apply the assessment of effects provided by the General Court in Microsoft which required the elimination of all effective downstream competition, one way to determine whether all effective competition is removed may be by asking whether the dominant firm could act independently of its rivals in the downstream market.[95] This could involve asking whether Google’s downstream competitors would be able to apply effective competitive pressure and obtain a greater market share if Google were to decrease quality or increase prices of its own applications.

8.   The judgment of the Consiglio di Stato

8.1.  The Consiglio di Stato’s finding concerning the open nature of the platform

The judgment of the Consiglio di Stato follows the Court of Justice and also confirms the decision of the ICA.[96] Specifically, the Consiglio di Stato finds that Google has allowed access to third parties to Android Auto and that, as such, the Bronner case law is not applicable.[97] Interestingly, it appears from the judgment that Google tried to distinguish between the Android operating system (which is undoubtedly an open-source system) and the Android Auto App which, while based on the Android operating system, should be considered as a closed platform. The Consiglio di Stato, however, considered that Google was still under an obligation to grant access and that the fact that other apps (namely, Kakao) had been granted access to Android Auto, undermined Google’s argument.[98]

8.2.  The Consiglio di Stato’s findings about potential anticompetitive effects

The Consiglio di Stato considers that the exclusion of JuicePass from Android Auto was ‘concretely suitable’ (concretamente idonea), to limit the possibility of JuicePass to create its own user base.[99] As such, it is irrelevant that other companies active in the same market have reinforced their position, as the ‘lack of effects’ may be due to other changes in the market.[100] This seems to go beyond paragraph 59 of the Judgement of the Court which considered that the continued presence on the market of JuicePass could not, in itself, mean that Google’s conduct was incapable of having anticompetitive effects.

8.3.  The Consiglio di Stato’s findings about objective justification

The Consiglio di Stato briefly considers the possible existence of objective justification to Google’s refusal to grant access to JuicePass on Android Auto. This possibility, however, is rejected as, in the Consiglio di Stato’s view, Google has not raised convincing justifications, and an internal document suggests that the denial of access to JuicePass is a policy choice, rather than a choice made on the basis of security concerns.[101]

8.4.  The Consiglio di Stato’s findings about market definition

The Consiglio di Stato also considers Google’s objections concerning the Authority’s failure to define a downstream market. In fact, while the ICA has defined, in line with the Commission’s precedent, a market for operating systems and a market for app stores, it has not defined the precise borders of a market, but rather considered a ‘competitive space’ in which Google was dominant on one end and competed with JuicePass on the other.[102] The Consiglio di Stato, aligning with the Court of Justice finds that the Authority was not under an obligation to define a precise downstream market.

9.   Implications and unresolved issues going forward

9.1.  Should the Court’s judgment be limited to digital platforms? 

A key question going forward is whether the Court’s judgment should be limited to digital platforms. The reason why extending this judgment beyond digital platforms is problematic is that it may have more significant implications for innovation incentives in other markets. If one were to argue in the Magill case that the copyright had actually been created for the purpose of allowing third parties to access it since it was licensed for free to other third parties such as newspapers, this would result in indispensability and possibly the new product rule not being applied to refusals involving intellectual property rights.[103] Innovation incentives are particularly important to protect in intellectual property cases because they often involve substantial investments.[104]

The Court seems to be quite clear that it intends the assessment of whether Bronner is applicable to be extended to non-digital platform infrastructures.[105] In paragraphs 44 to 47 of the judgment, it refers to a dominant undertaking but does not specifically mention a digital platform. If a non-digital platform infrastructure was developed for the purpose of enabling third parties to use it, then Bronner will also not be applicable. With regards to the specific findings on objective justifications, it appears that the judgment is aimed at digital platforms since in paragraph 81, the Court specifically refers to digital platforms. One reason for why it chose to refer specifically to platforms in the objective justifications part might be that it made more sense to create specific rules for interoperability which are often more straightforward to provide than in non-digital markets. Stretching these objective justifications to other markets might therefore have been unreasonable and place excessive demands on undertakings that own infrastructure in non-digital markets.

As previously mentioned, it has been argued by scholars that platform specific rules under Article 102 TFEU may be an appropriate solution with Bronner not being applicable because of a platform’s quasi regulatory status.[106] This quasi regulatory status might also be linked to a heightened special responsibility of digital platforms.[107] This can further be supported by the high barriers to entry that often exist in platform markets due to for example strong network effects.[108] Additionally, platforms may be considered gatekeepers performing an intermediary role connecting supply and demand and thus controlling how companies reach end users.[109]

It may also be even more prudent to limit the Court’s findings to super-dominant platforms which make downstream rivals so dependent on their platform infrastructure that even if there are alternative upstream platforms, they are not a viable option for the purpose of competing in the downstream market.[110] Hence, applying Bronner would make little sense in this context as although the platform is not indispensable according to the meaning of Bronner, in reality the platform is since a competitor would be unable to apply effective competitive pressure if it were to use alternative upstream infrastructures.[111]

It would be unwise to stretch this line of case law beyond platform markets due to the risks that it would have for innovation incentives in a variety of industries. This is particularly important because unlike Slovak Telekom and Google Shopping, this judgment would mean that Bronner is not applicable to an outright refusal to deal rather than a constructive refusal to deal. Due to the specific open nature of digital platforms and the gatekeeper role that they may hold, the non-application of Bronner makes more sense. It is more logical to place more emphasis on open markets rather than upstream innovation since the follow-on innovation that may result in the downstream market can greatly outweigh the innovation impact on the upstream market.[112] It is less clear in other non-digital platform markets whether placing greater emphasis on open markets will have a positive impact on overall innovation. This is especially true in cases involving intellectual property rights such as in Magill.

9.2.  Problems with defining a platform developed in order to enable third parties to use it

The Court’s reinterpretation poses significant issues from a practical perspective for digital platforms as it essentially lowers the standard for an abuse of dominance for platforms which were ‘developed in order to enable third-party undertakings to use it’. The Court does not provide much clarity on how to determine whether this is fulfilled other than stating that evidence such as the dominant firm having already granted access may support this finding.[113] It refers to the fact that Google had granted access to its Android Auto infrastructure and that Google had made specific templates to allow apps to be compatible as further evidence.[114]

The fact is, there are examples of platforms which were not developed to enable third parties to use the infrastructure. For example, the first iPhone, released in 2007, did not allow third party apps on iOS.[115] Steve Jobs famously envisioned third party app developers to develop apps exclusively within the iPhone.[116] The Amazon platform was not originally developed for third party users and marketplace was launched only in 2000, years after the creation of Amazon.[117] Facebook famously was originally a closed social network with limited integration for third-party apps. The policy rationale for applying a less strict approach for closed platform presents conceptual challenges. Furthermore, the dual standard for platform developed in order to enable third parties to use the infrastructure, appears fictitious. Nowadays, there is no plausible explanation for developing any sort of platform without relying on any third parties as third parties activate network effects and bring audiences to any platform. The effect, therefore, would be to make Bronner practically inapplicable to any digital platform, thereby defining platforms as inherently open facilities.[118]

A final point with regards to the uncertainty surrounding what is meant by an infrastructure being developed to enable third parties to use it is that it leads to legal uncertainty which in turn may further hamper innovation incentives. A platform that is unable to precisely determine whether it merits the protection of Bronner when refusing to grant interoperability may decide to grant access to third parties with which it competes. This would of course occur in the case that there are no objective justifications. The issue is that the platform might lose its innovation incentives by having to grant access to competitors even though it may actually have merited the protection of Bronner. Hence, the legal uncertainty resulting from this judgment may lead to unnecessary and unwelcome consequences for innovation incentives.[119]

9.3   The assessment of effects going forward and insights from the Consiglio di Stato’s judgment

It is important to note that although digital platforms may be concerned about the implications of this judgment, anticompetitive effects still need to be established, even if Bronner is not applicable. As discussed above, the level of effects is not entirely clear, but the Court does not seem to require a level that would result in the elimination of all effective downstream competition. Nevertheless, in order to foreclose downstream rivals, the conduct must be capable of hampering their ability to effectively compete by preventing them from either growing on the downstream market or discouraging them from entering altogether.[120] In the context of vertically integrated platforms, economic literature also provides that foreclosure is less likely when the upstream infrastructure is not indispensable.[121] Hence, establishing to the requisite standard anticompetitive effects may be a challenge if there are viable alternatives to the dominant platform which allow rivals to offset the harm caused by the platform refusing access or if the downstream structure of the market would allow competitors to challenge the allegedly dominant behaviour on the upstream neighbouring market.

In its judgment, as explained above, the Consiglio di Stato did not assess whether Android Auto was indispensable following the Court of Justice’s ruling. Nevertheless, in the effects assessment, it considered that Android Auto was in fact indispensable since alternative apps outside of Android Auto could only be used when the vehicle was stationary and therefore there were no viable alternatives for ENEL X to leverage network effects and gain sufficient scale.[122] With regards to these findings, first, the Consiglio di Stato essentially answered the initial question which it referred to the Court of Justice by establishing that convenience was relevant in the assessment of indispensability, at least for the purpose of assessing effects. The use of mobile apps was not a convenient and therefore viable alternative and would not allow ENEL X to offset the harm. Second, the Consiglio di Stato demonstrated its awareness that in order to establish effects in vertical foreclosure scenarios, the infrastructure of the dominant firm must either be indispensable or close to indispensable in order for foreclosure to arise. This is consistent with the economic literature discussed above. Furthermore, it ensures that the investment incentives of digital platforms are still to some extent protected. Had Google’s Android operating system been considered a viable alternative to Android Auto, the refusal was unlikely to be capable of foreclosing ENEL X from the market.

Although the Consiglio di Stato took indispensability into account when assessing effects, its findings may be disputed. The fact that ENEL X had continued to grow after Google’s conduct was implemented strongly suggests that this convenience was not of significant importance for being able to compete on the downstream market. The Consligio di Stato only briefly states that these actual effects are not relevant and does not reason why they are not an indication that Android Auto is not indispensable.[123] Furthermore, from the facts of the case, there still appears to be other competitors that also remain on the market and therefore exert competitive pressure on Google maps.[124] In some instances, these competitors also have a considerable advantage over Google and other competitors in that they own the underlying infrastructure (i.e., the EV charging stations). For example, ENEL (unrelated to ENI, but also largely owned by the Italian Government)[125] has installed over 21,000 thousand charging stations at 31 December 2024,[126] which corresponds to roughly a third of all the EV charging stations installed in Italy at the same time,[127] while also developing its own app to find charging stations.[128] The Consiglio di Stato does not take these considerations into account in its assessment of effects.

10. Conclusions

The Bronner legal test was designed to protect the fundamental rights of firms and their innovation incentives. Stretching from Telia Sonera up until Android Auto, Bronner is now becoming the exception rather than the rule. The Court in Android Auto has now taken this approach even further by providing deviations from the application of Bronner in cases involving outright refusals to deal. Digital platforms that have been developed to enable third parties to use them will no longer be able to rely on their fundamental rights and innovation incentives and only have objective justifications as a means to refuse downstream competitors from gaining access. The implications of the judgment are significant and may have either a positive impact on follow-on innovation in downstream markets or an excessively detrimental impact on upstream platform innovation. However, extending this judgment beyond platform markets would be unwise as the negative impact on upstream innovation may significantly outweigh the follow-on downstream innovation in non-platform markets. Furthermore, the uncertainty regarding how to determine whether a platform is designed to be open is particularly problematic as certain platforms may have initially been developed as closed platforms but later on been made open. It also remains to be seen how future cases involving refusals to deal by open digital platforms will assess effects. Although Bronner is not applicable, anticompetitive effects must still be established. As demonstrated by the Consiglio di Stato, the assessment of indispensability should remain important for any effects assessment in this context and can ensure that investment incentives are still to some extent protected.