Essential facilities doctrine and digital ecosystems: case C-233/23 alphabet (android auto)
Year: 2025
Type: article
Abstract: Judgment of 25 February 2025, Alphabet and Others (Android Auto), C-233/23, EU:C:2025:110. The Court of Justice denied the applicability of the Bronner criteria in cases of refusal to grant interoperability with digital platforms generally open to third-party complementors and provided guidance on objective justification in these cases. Android Auto is yet another judgment dealing with practices by Google. Similar to Google Shopping and Google Android, it is set to have long-lasting effects on European competition law. As in previous cases involving refusals of access to resources, it raised the question whether the refusal had to be assessed under the conditions of the essential facilities doctrine (EFD) as formulated in Magill (Case C-241/91 P, EU:C:1995:98), Bronner (Case C-7/97, EU:C:1998:569), IMS Health (Case C-418/01, EU:C:2004:257) and Microsoft (Case T-201/04, EU:T:2007:289). According to them, a refusal constitutes an abuse where the resource is indispensable to the exercise of an activity on a downstream market, the refusal excludes any competition on that market, and the defendant fails to provide an objective justification. Where the product is moreover protected by IP rights, the CJEU, in addition, requires the refusal to prevent the appearance of a new product for which there is potential demand. The strict conditions are justified by two considerations: First, while access might be beneficial for competition in the short term, it harms competition in the long run by lowering incentives to invest in resources. Second, forcing companies to deal with competitors and transferring resources significantly interferes with their freedom of contract and right to property. The Courts restricted the doctrine’s applicability in recent years. Already in Van den Bergh Foods (Case C-552/03 P, EU:C:2006:607, para 137), the ECJ held that where a dominant undertaking had not retained a resource for its own business and, thus, was not forced to enter into agreements or transfer assets for the first time by the Commission’s decision, the Bronner criteria would not apply. In the decisions Telia Sonera (Case C-52/09, EU:C:2011:83, paras 55ff), Slovak Telekom (Case C-165/19 P, EU:C:2021:239, paras 43ff) and Google Shopping (Case C-48/22 P, EU:C:2024:726, paras 89ff), it evolved into the general principle that cases in which access had been granted but under unfair or discriminatory conditions could not be dealt with under the EFD. Further exemptions complemented the principle: Assuming that legislators had already balanced effects, the Bronner criteria were not applicable where regulation ordered access (Case C-165/19 P, Slovak Telekom, EU:C:2021:239, paras 54ff, Case C-42/21 P, Lithuanian Railways, EU:C:2023:12, para 88). As investment incentives could not be hurt, the criteria were equally inapplicable where resources had been developed with public funds or where the resource had just been destroyed and, thus, sacrificed (Case C-42/21 P, Lithuanian Railways, EU:C:2023:12, paras 81ff, 86f). Android Auto joins the ranks of these decisions, further restricting the Bronner criteria’s applicability. Android Auto is an extension to Google’s Android OS that allows users to access apps from automobile infotainment systems. Google provided templates as interoperability solutions for whole categories of apps. At the time, Google had not yet developed templates for map services. Only Google’s own services Google Maps and Waze could be used via Android Auto. Enel X operated charging stations for electric vehicles in Italy. Its app allowed users to search for and operate charging stations as well as navigate to them via Google Maps. Enel X wanted its users to conveniently use the app in their cars. It therefore asked Google to provide interoperability with Android Auto. Google denied the request, citing security concerns and the template’s resource-intensive development. After a second refusal, Enel X launched a complaint with the Italian competition authority (AGCM). The AGCM found that Google’s refusal violated Art. 102 TFEU and ordered it to develop the necessary template. In its referral to the CJEU, the Consiglio di Stato asserted that Google occupied a dominant position due to its control of Android OS and Google Play and submitted five questions for a preliminary ruling. The Consiglio di Stato first asked whether it is ‘sufficient that access be indispensable for a more convenient use of the product’. The Court reinterpreted the question. It asserted that the referring court was uncertain whether, in light of the ‘specific characteristics of the functioning of the digital markets’ and ‘the function of a digital platform’ (both para 35), the Bronner criteria were applicable. The Court held that the Bronner criteria had been developed in a specific case in which competitors had requested access to an infrastructure a company had developed for its own needs and exclusive use. The strict criteria served to strike a ‘delicate balance’ (para 32 of AG Medina’s opinion). As incentives to invest in such infrastructure would diminish since competitors could just request access to the dominant firm’s infrastructure while the dominant company would lose its competitive advantage, a short-term increase in competition is met with a long-run decrease. The duty to give access to proprietary infrastructure and transact with competitors, moreover, constitutes a serious restriction of undertakings’ fundamental rights. According to the Court, the case at hand was different. Google had not developed Android Auto for its own needs but had planned and, in fact, already granted access to third-party firms. Expected third-party use had seemingly not eliminated its investment incentives. As the access obligation did not fundamentally alter Google’s business model, it constituted a limited interference with its rights. Since circumstances differed, the Bronner criteria were not applicable. Secondly, the referring court asked whether the fact that the undertaking requesting access and its competitors were and continued to be active in the market and even grew their presence, prevented the finding of abuse. Reiterating recent case law, the ECJ held that, while Art. 102 TFEU required conduct to be capable of restricting competition, it did not require evidence of actual exclusionary effects. Art. 102 TFEU served to prohibit exclusionary conduct irrespective of its success, which might depend on external factors. However, the capability had to be demonstrated ‘based on tangible evidence which establishes [the capability] beyond mere hypothesis’ (para 57). Enel X’s continued growth did not in itself exclude the conduct’s abusiveness but might nevertheless be evidence that the conduct was incapable of having anti-competitive effects. Questions three and four dealt with a potential objective justification. The Consigli di Stato asked whether the template’s inexistence at the time of the request provided an objective justification or whether the dominant undertaking can be forced to modify or develop a template and, if so, based on which criteria. The Court held that the template’s mere inexistence did not provide an objective justification. Instead, it was for the dominant undertaking to modify or develop the template. A refusal was only justified where interoperability ‘would […] compromise the integrity or security of the platform concerned, or where it would be impossible for other technical reasons’ (para 73). Difficulties involved in the development could only justify delaying access for a reasonable time. The reasonableness had to be determined with a view to the needs of the requesting undertaking and the difficulties involved in the development (e.g. technical difficulty), resource constraints (e.g. human resources), and external constraints (e.g. regulation). The ECJ clarified, however, that the dominant undertaking could demand a fair and proportionate financial contribution, inasmuch as it had to allow the dominant undertaking ‘to derive an appropriate benefit from it’ (para 76). In line with existing case law, the ECJ maintained that the burden to present and prove an objective justification was on the dominant undertaking. However, once it had done so, the authority had to show that arguments and evidence did not suffice. In that process, the initial refusal played a significant role. If the dominant undertaking did not provide a reason, it might be indicative of a lack of objective justification. Otherwise, the authority should take particular account of the reasons given in the refusal. With its last question, the referring court enquired whether the authority had to define and identify the downstream market on which the conduct was capable of restricting competition. The ECJ differentiated two markets: the upstream platform market, on which the undertaking refusing access occupied a dominant position, and the downstream market, where the refusal led to restrictions of competition. While the upstream market had to be defined to ascertain the dominant position, the downstream market merely had to be identified. It may be sufficient to identify a potential or hypothetical downstream market, in particular where it is still developing or rapidly evolving. The authority then had to show the capability of anti-competitive effects with reference to the market’s potential scope and characteristics. The decision entails three significant aspects for Art. 102 TFEU case law. First, the Court further limited the Bronner criteria’s applicability. The circumstances in Bronner differed from the case before it. In Bronner, the infrastructure was developed to serve the undertakings’ own needs to provide it with a competitive advantage. In this case, the platform was developed with the explicit aim of encouraging complementors’ participation. The fact that Google had made Android Auto available to third parties served as a presumption that expected third-party access had not lowered investment incentives. Without a change to the business model, the interference with Google’s fundamental rights was limited. Similar to previous judgments, the Court based the inapplicability on the fact that access had already been granted. AG Medina went a step further. She held that platforms and ecosystems ‘create[d] value by ensuring that [their] users have access to a wide variety of interoperable and complementary products and services’ and they were ‘conceived not only to allow but to encourage third-party developers to create versions of their own apps that are compatible with it’. At the ‘heart of an ‘ecosystem", Android Auto was to be shared with external developers to service its users (all three quotes para 37 of her opinion). AG Medina, thus, based her argument on the changed economic rationale of access in ecosystems. The Bronner criteria envisioned resources as exclusive assets to be used in competition with other firms. In ecosystems, however, products and services always depended on interaction with closely related, complementary third-party products. Platforms in ecosystems were infrastructures whose value to users increased with every complementary contribution by other firms. Investment incentives were, therefore, expected to increase with the availability of compatible complements. Leaner and closer to previous case law, the Court’s criterion provides legal certainty and an easy rule to be applied by courts. In the end, however, AG Medina’s reasoning builds the basis for the decision and explains why the ‘delicate balance’ underlying Bronner is different from the one in this case. Where complementary products add to the value of a platform, the presumption underlying Bronner that the refusal is generally pro-competitive can no longer hold. Secondly, regarding the objective justification, the Court adopted a notably regulatory approach to enable collaboration in ecosystems. On the one hand, it acknowledged that expected negative effects on the platform’s integrity and security could justify a refusal. The prevention of negative externalities was, thus, a legitimate reason to refuse to open the platform to third parties. On the other hand, the mere fact that interoperability required an active effort by the dominant undertaking did not justify the refusal. It implicitly recognized that in digital ecosystems, where complements intricately interlock, access always requires a coordinated effort to adapt products to each other and create compatible interfaces. Both statements acknowledge the coordinative nature of value creation in digital ecosystems. The exceptions to the duty to provide access—integrity and security of the platform – are notably borrowed from the DMA. Both are exceptions to the DMA’s interoperability provisions in Art. 6(4), 6(7) and 7. They are relatively narrow. It is questionable whether they account for all cases in which a refusal to give access might be pro-competitive. While the exceptions might be stretched to a degree, they certainly limit the platform operator’s ability to ensure coherence and compatibility across the platform’s product offering. They might, for example, not allow for differentiating the product offering from those of competing ecosystems or granting complementors exclusivity to induce them to invest in their presence on the platform. Unlike the Court, AG Medina held that ‘where the granting of access […] could affect, from a technical perspective, the performance of the platform or where access could run counter to its economic model or purpose’ a refusal could be justified (para 64). These grounds seem to go beyond the ‘integrity and security of the platform’ and seem better equipped to cater to the variety of legitimate grounds for refusal. It might be reasonable to limit exceptions from interoperability to integrity and security where the addressees are gatekeepers and, thus, might be considered akin to public utilities. This is, however, not the case for all dominant undertakings. The third—and easily overlooked—change pertains to the burden of proof. It is established case law that it is for the dominant undertaking to present and prove an objective justification. However, AG Medina and the Court made statements that point in the direction of a proceduralized burden-shifting framework in refusal to deal cases. They held that the dominant undertaking had to provide, with its initial refusal, reasons that build the basis for the evaluation of the objective justification. ‘[An] examination of the concrete terms of the refusal emerges as an important element in determining whether the justification put forward by the dominant undertaking did in fact form the basis for its conduct’ (para 71 of AG Medina’s opinion). The burden then shifted to the authority which had to verify the rationale: It had ‘to demonstrate that the arguments and evidence […] cannot prevail and, consequently, that the justification […] cannot be accepted’ (para 79 of the judgment). Burden shifting is another step toward a more regulatory framework apt to navigate the complexities of digital ecosystems. It first compels operators to give with their initial refusal a reason that provides the Commission from the start with a concrete thesis it can verify or refute. Secondly, it prevents the Commission from simply referring to the defendant’s burden of proof and forces it to seriously engage with the points brought forward. It forces the parties to discuss the grounds for the refusal and how potential access can be facilitated. These changes might prevent ineffective and unsatisfactory remedial decisions, as arguably epitomized by the aftermath of the Commission’s Google Shopping decision. It is an insight shared with the DMA: Expedited and productive ‘regulatory dialogues’ are essential for the effective resolution of competition concerns in digital markets. Android Auto, hence, holds three broader lessons: First, not only is the DMA influenced by Art. 102 TFEU cases. The DMA has started to influence Art. 102 TFEU case law. Second, the Court—although not explicitly—recognized the economics of digital ecosystems and started to adapt the law to it. Third, the realities of competition in digital ecosystems—the dependence of complementors on interoperability and the active coordinative effort needed to facilitate interoperability—require Art. 102 TFEU to become more ‘regulatory’ to remain effective.
Author: Philipp Hornung
Institution: Humboldt-Universität zu Berlin
Language: English
Subfield: Strategy and Management
Domain: Social Sciences
Open Access status: hybrid
Estimated APC paid: $4,850
OpenAlex ID: W4411729780