The Digital Markets Act (DMA) may be a young regulation but, curiously, is already showing signs of being misaligned with actual market realities. As the Act was introduced, there was an immediate implementation agenda that followed on the heels of previous and current antitrust cases. Essentially, this agenda has been completed. New ideas for the DMA have been kicked around in the current DMA review: some are a consequence of steps already taken (e.g., an investigation on cloud); others are much less connected to the Act and its purpose. At the same time, there is a growing view in policy circles that the DMA is struggling to achieve its stated objectives because the market it was designed to regulate has changed.
Clearly, as technology evolves, markets change. The AI era is changing competitive dynamics, with similar shifts likely to follow from advances in quantum computing, edge AI, 6G, and next-generation cloud infrastructure. We are simply no longer in the platform market that we had in the 2010s and that remains the core market and regulatory concept of the DMA. Therefore, the DMA is anchored in assumptions about digital markets that may be becoming increasingly misaligned with the reality of the fast-evolving market, raising questions about whether it is addressing the right dimension of competition. Importantly, the DMA does not merely risk becoming outdated if it reinforces old assumptions; it may also intervene along the wrong dimensions altogether and cause results that are bad for competition, consumers, and European growth.
The programmatic, almost mechanical, concept of interoperability is a case in point. It is shifting the regulation from a focus of competitive conduct toward the redesign of digital products and systems. This risk was always rooted in the assumptions that underpin the DMA’s approach. The case for heightened competition scrutiny of online platforms in Brussels has traditionally rested on the view that these markets exhibit structural features such as network effects, economies of scale, switching costs, and data advantages. That was true and, as a result, EU competition officials spent years addressing several of the problems associated with such markets. Yet these characteristics are not static. New technology and service developments have already eroded some of these advantages, lowering barriers to entry in ways the DMA did not accommodate.
From Competition Enforcement to Market Design
The starting point for the DMA was that an ex-ante regulation was needed to reduce the market power of selected platforms. It was designed to regulate firm behaviour through a series of do’s and don’ts. However, the DMA does no longer simply regulate firm behaviour in the traditional antitrust sense; it actively seeks to reshape markets through prescriptive rules that govern how digital services are designed and delivered.
This ambition has been reinforced over time and is now changing the character of the DMA. But this pursuit risks fragmenting services along provider lines rather than rewarding quality and performance, marking a significant shift from competition enforcement toward regulatory market design. This transformation is also asymmetric, applying primarily to a limited set of designated firms. Programmatic interoperability demands on operating systems are a good example: it is increasingly mandating specific product designs and the direction of the Act now seems be to towards architecture design.
The orientation towards product design restrictions started already with the implementation process. Especially Apple and Google implemented many interoperability measures that had an impact on the actual product design and experience – pushing the former to delay new features in their European products and services like live translation and device mirroring because of regulatory risks. Remarkably, a European user of an iPhone now has access to fewer features and services than a user in other and similarly mature markets.
In the case of Google, there has been some quantitative evidence published on the results of the interoperability measures and they are a useful illustration of how regulatory obligations translate into system configuration changes. In January 2024, as part of the DMA implementation, the company redesigned its search results interface for EU users. For location-based queries, the previously interactive map is no longer clickable, and direct access to Google Maps has been removed. Users can now obtain directions, but can no longer zoom, explore nearby services, or interact with the map directly. These changes were introduced to address concerns that Google was steering users from search toward its own mapping service, thereby potentially limiting user choice.
Yet the outcome runs counter to the DMA’s stated objectives. Empirical evidence suggests that search queries for “maps” and “Google Maps” increased by more than 20 per cent across EU countries following these changes, reflecting a significant rise in user effort to access Google’s mapping service. Much of this additional search activity ultimately redirects users back to Google Maps rather than to competing services.
The evidence further suggests that, while there has been a modest increase in searches for Apple Maps, largely among Apple users, there has not been any corresponding shift toward other alternatives such as Bing Maps. Geo-localised queries, including those related to restaurants or retail, remain broadly stable, indicating that reliance on Google Search has not materially declined. In this sense, although the DMA has modified the user pathway, it has not resulted in improved contestability or diverted traffic to rival services. If anything, it appears to have reinforced users’ reliance on one ecosystem, highlighting the limitations of design-based regulatory interventions.
The idea that regulation can influence system architecture is however not new. In EU competition law, the essential facilities doctrine (EFD) has long recognised that access to certain infrastructures may be mandated. However, such intervention is limited to exceptional circumstances, requiring that the input in question be strictly indispensable for effective competition. The doctrine has been narrowly applied and remains contested, reflecting broader concerns that mandatory access may undermine firms’ ability to design and control their systems. The DMA marks a significant departure from this framework. Rather than relying on a case-by-case assessment of indispensability, it establishes ex ante obligations that require access to be granted systematically by designated gatekeepers.
In doing so, it shifts the regulatory focus from whether access must be granted to how that access should be implemented in practice. This shift has important consequences. Once access is no longer conditioned on indispensability, regulators must determine the scope, quality, and modalities of access, thereby drawing them into specifying technical interfaces, system architectures, and even user-facing functionalities. In digital environments, where system behaviour is determined by technical architecture, such specification effectively becomes direct intervention in product design
Regulating Architecture: Specification Proceedings and System Design
This shift towards regulating how access must be implemented is primarily operationalised through specification proceedings under Article 8(2). While these proceedings are formally intended to clarify how designated firms can comply with their obligations, in practice they operate as instruments for defining the technical conditions of compliance. By initiating a specification proceeding, the Commission signals that a firm’s existing implementation is insufficient and begins to prescribe the concrete system-level changes required to meet DMA obligations.
This process moves beyond legal compliance and into the domain of system design. In specifying how interoperability, access, or data sharing must function, the Commission effectively determines the structure of interfaces, the organisation of system components, and the manner in which users and third parties interact with the platform. As a result, specification proceedings do not merely interpret legal obligations but actively shape the architecture of digital products, transforming regulatory enforcement into a form of design intervention.
The Commission can open proceedings pursuant to obligations laid down in Articles 5, 6 and 7 of the DMA. To date, enforcement has focused on Article 6. This is particularly evident in relation to Articles 6(7) and 6(11), both of which extend beyond traditional conduct remedies and reach directly into product design. These provisions operate as de facto constraints on the designated firms’ design choices. For instance, firms with an operating system must enable third-party interoperability on equal technical terms, while designated search engines must share aggregated data with rivals on FRAND terms. Thus, compliance with these provisions therefore necessitates architectural changes, rather than merely adjustments in commercial behaviour.
Several principal questions arise immediately as regulators move closer to product design features. As pointed out in other ECIPE work, the more prescriptive that mandated behaviour becomes, the more likely it is that compliance with one EU regulation will impact negatively on another EU regulation. More specifically, does the Commission really have the competence and experience to regulate product design? Can they do so with a clear understanding that the measure is actually going to impact competition and users positively?
Developments under Article 6(7) illustrate how specification proceedings extend into system implementation. It started in 2024 with the launch of a specification investigation for Apple mandating faster design interoperability in nine areas of connected devices, regardless their degree of and consequences for security. This development stems from the obligation imposed on gatekeepers under Article 6(7) which requires them to enable access to, and interoperability with, hardware and software features controlled via their operating systems. As a result, compliance may constrain the platform’s ability to maintain tightly integrated or modular system design. In the case of Google, the Commission recently initiated specification proceedings despite the platform’s existing interoperability framework. The first set of proceedings concerns Google’s obligation under Article 6(7) to provide third-party developers with free and effective interoperability with hardware and software features controlled by the Android operating system.
These proceedings are interesting because they suggest the Commission is going further into system designs in ways that can be imitated for other gatekeepers. There has been an evolution from the implementation mandates, over the first specifications, to these new proceedings. The development is also causing confusions through the full stack. All systems rely on an interplay between hardware and software that the OS have standards for but do not always control. Obviously, if it is an open-source model like Android it is even more challenging because implementers make their own decisions. A gatekeeper may be a gatekeeper, but they are not Gods. Google may be asked to do things that are also influenced by the implementer who use the source code to develop their own features and standards for technical interoperability.
In the new specification proceedings, this orientation towards system design includes issues relating to Google’s AI services, such as Gemini. A second set of proceedings concerns the obligation to grant third-party providers of online search engines access, on FRAND terms, to anonymised ranking, query, click, and view data held by Google Search. The Commission is expected to conclude these proceedings by the summer. While no formal finding of non-compliance has been made, these developments suggest that compliance may soon increasingly be shaped through regulatory interpretation of interoperability requirements, creating profound uncertainty as to the scope of future obligations.
However, it is important to note that specification decisions adopted by the Commission are binding, requiring gatekeepers to implement the prescribed measures. Unlike non-compliance decisions, these proceedings do not sanction infringements but instead define in precise terms how obligations under the DMA must be fulfilled. This creates a double-edged effect: on the one hand, gatekeepers are required to introduce potentially significant design and technical changes, which may be complex and resource-intensive; on the other hand, failure to comply with these specifications may ultimately expose them to fines, thereby increasing regulatory and financial pressure.
A similar specification dynamic can be observed in the case of Apple. Prior to the Commission’s specification decisions, Apple had already introduced a range of DMA-related changes, including an interoperability request process and, in the EU, the ability for browser applications to use alternative browser engines rather than WebKit. There had also been a lot of pressure on interoperability under the DMA which led to significant misuse by third-party developers, like pornographic apps who branded themselves of being “approved by Apple. The specification decisions on Apple require interoperability across a range of core system functionalities, including, inter alia, connectivity, notifications, and device pairing, thereby compelling Apple to implement significant technical adjustments to enable third-party devices to interoperate more fully with iOS. These measures shift regulatory obligations from a model of controlled, discretionary interoperability to one of mandatory, system-level interoperability, extending into areas closely tied to system design and implementation. Apple is challenging these decisions before the EU General Court, arguing that the Commission has exceeded its mandate by imposing overly detailed technical obligations.
Integration, Differentiation, and User Choice
A deeper issue is that different platforms and systems have certain designs for a reason. The same underlying service can perform differently depending on the provider – not arbitrarily, but because of deliberate integration choices. Consider Android device manufacturers such as Samsung and Xiaomi. They both rely on the same core operating system, yet differentiate through their own app stores, software layers, and optimisation strategies, while still allowing users to choose browsers and applications.
All firms invest heavily in hardware–software co-design to reduce latency, enhance performance, and deliver distinct user experiences. If regulation requires that such improvements be immediately and fully shared with every actor in a certain design – even when they raise core concerns over security, privacy, and technical control – the incentive to develop them in the first place is diminished. The likely result is not more effective competition, but fewer performance-enhancing innovations overall.
Regulatory approaches often overlook an important dimension of consumer behaviour: users do not passively accept system constraints but actively navigate them, even where this involves trade-offs. For example, Android-based systems already allow users to access multiple app stores and install applications from external sources, including open-source repositories. While such choices may involve security risks, they reflect deliberate user preferences for flexibility and access. At the same time, other users choose more tightly integrated ecosystems, such as Apple’s iOS, where restrictions on app distribution and system access are more pronounced but are often associated with greater security, consistency, and ease of use. These contrasting choices suggest that users are not merely constrained by system design, but actively select between different models based on their preferences.
The core issue lies in the wording of Article 6(7) itself, which requires “(….) access for the purposes of interoperability to the same operating system, hardware or software features, regardless of whether those features are part of the operating system, as are available to, or used by, that gatekeeper when providing such services.” This formulation goes beyond access obligations. This does not simply require gatekeepers to open existing interfaces but pushes towards a structural reconfiguration of system architecture and internal functionalities, amounting to a design mandate.
This suggests that consumer welfare cannot be reduced to the availability of formally “contestable” options as defined by the DMA. In practice, users often value the ability to exercise discretion within integrated systems, rather than being steered toward predefined pathways. To the extent that the DMA emphasises a programmatic version of interoperability, it risks underestimating user agency and the role of system design in enabling differentiated experiences.
The irony is that the DMA may not impact much on contestability but, in the name of the consumer, usher in product designs that do not conform to the varieties of user preferences. The DMA seems to assume that greater contestability will automatically translate into better competitive outcomes. However, in digital platform markets characterised by strong network effects and scale economies, this relationship is not straightforward. Efforts to engineer contestability through regulatory intervention do not necessarily reproduce the outcomes associated with competitive markets. In fact, they rarely do. Contestability rather comes from encouraging product and design variety, and allowing for new innovation to be diffused rapidly.
As illustrated by the Google Maps example, design-based interventions aimed at improving contestability may alter user pathways without meaningfully changing underlying market dynamics. More importantly, such interventions risk substituting competitive processes with regulatory design choices. In fast-evolving digital markets, where firms compete through system integration, performance, and user experience, competition cannot be reduced to the availability of formally contestable options. Instead, attempts to impose contestability through prescriptive design requirements may limit differentiation and weaken the very mechanisms through which competition operates.
Expanding Intervention and Regulatory Tensions
The DMA review is a good opportunity for the Commission to better align the purpose, design, implementation, and enforcement of the Act with each other and develop better approaches than programmatic interoperability demands. The shift toward design-based intervention has broader institutional implications. The core issue is not simply the level of regulatory ambition, but the Commission’s growing reliance on ex ante rules aimed at large technology firms. While dominant platforms have at times engaged in problematic conduct, the European competition framework already provided tools to address such behaviour through antitrust enforcement. The DMA added an additional regulatory layer, which could work efficiently as well as supporting competition and consumer welfare if implementation and enforcement are not programmatic. Now, however, there are rising concerns about overlap and inconsistency. This is particularly evident in areas such as potential conflicts with national enforcement, platform-specific competition rules, and sector-specific legislation.
A broad prohibition of practices such as self-preferencing may further overlook the efficiencies associated with vertical integration, including improvements in product design and user experience. Traditionally, competition law intervention is justified where a coherent theory of harm is established, yet the treatment of self-preferencing has not always met that standard. In cases such as Android, it has been argued that remedies focused less on restoring competitive conditions and more on altering the terms under which integrated services, such as app distribution through the Play Store, are provided. In this respect, the DMA appears to go beyond addressing enforcement gaps and instead intervenes directly in the design and organisation of digital ecosystems.
Therefore, the DMA appears to rest on the implicit assumption that markets with fewer firms are inherently less competitive and that increasing entry will lead to better outcomes. While this may hold in traditional monopoly settings, it does not necessarily apply to digital markets shaped by strong network effects and scale economies. The platform market structures do not, by themselves, indicate harm; nor do they ensure that new entrants can compete sustainably or improve welfare. Where intervention simply reallocates market shares without altering the underlying economics, the benefits for consumers are limited.
By redesigning interfaces and access conditions in the name of contestability, the DMA appears to be fragmenting products without necessarily generating meaningful competitive rivalry. None of this denies that digital markets can become concentrated, or that firms may design ecosystems in ways that favour their own services. But integration and incompatibility are not inherently anticompetitive; they often reflect product design choices, technical constraints, or efforts to improve performance.
The issue therefore is not simply whether large platforms should be subject to regulation, but how such regulation is implemented. By relying on prescriptive interoperability requirements and specification proceedings, the DMA increasingly operates through the redesign of digital systems rather than the correction of anticompetitive conduct. In doing so, it shifts the focus of competition policy away from outcomes and toward redefining competition from a process driven by performance to one shaped by regulatory design and provider identity. This shift carries risks: it places regulators in the position of determining product architecture, reduces firms’ ability to differentiate through design, and may ultimately undermine the innovation and user benefits that competition policy seeks to promote. In fast-evolving digital markets, the ultimate challenge is to avoid substituting competitive processes with regulatory design choices.
3 responses to “The DMA Is Not Fixing Competition, It’s Redesigning It”