The European Commission has turned to the Digital Markets Act (DMA) to address contestability in cloud computing. But it already has a purpose-built instrument for exactly this problem, and it has not yet fully deployed it.
This post makes two points. First, the Data Act, with its dedicated switching obligations for cloud providers, is a more tailored instrument for addressing contestability in cloud than the DMA. Second, the Commission should demonstrate why this purpose-built regime, many parts of which remain unimplemented, is insufficient before committing scarce enforcement resources to DMA proceedings.
A few months ago, the Commission opened an investigation under the DMA into cloud computing services. The purpose of this investigation is to determine whether several firms (Microsoft and Amazon, in particular) meet the definition of ‘gatekeeper’ and, more fundamentally, whether cloud services fit the DMA’s gateway logic.
The investigation appears to lean heavily on contestability as its orienting aim, as opposed to fairness. The Commission’s press release cautions against the risk that some players occupy ‘very strong positions’ that may be ‘further reinforced’ if action is not taken. A recent parliamentary inquiry reaffirms the goal of preventing ‘new forms of market lock-in’. These statements imbue the investigation with a contestability logic.
Contestability describes a market situation in which the absence of entry barriers enables potential entrants to maintain pressure on incumbents, ensuring that even structural monopolists cannot behave in a monopolistic manner. Conversely, an incontestable market is one in which, due to the ‘existence of very high barriers to entry’ (DMA Recital 3), entrants’ ability to ‘challenge’ incumbents is limited (DMA Recital 32).
Contestability depends on many factors. But the specific factor that the cloud investigation singles out is lock-in: the inability of customers to switch from their current cloud provider. This focus is well-placed. The premise that the threat of entry keeps incumbents on their toes is conditional on customers switching to the entrant when the incumbent tries to extract monopoly rents. Therefore, improving contestability depends, in some part, on reducing customers’ costs of switching to alternative suppliers.
From this viewpoint, using the DMA to address contestability issues in cloud markets appears odd. For there is another Act in the EU regulatory toolbox that is readily available and much more tailored to facilitating switching in cloud markets. It is the Data Act, one of the main purposes of which is to ‘facilitate switching between data processing services’, understood broadly to encompass essentially every cloud computing service on the market today (Article 2 (8)).
The Data Act dedicates an entire chapter to bolstering ‘the ability of customers of data processing services to switch from one data processing service to another’. To that end, it introduces commercial, technical, and organizational obligations on cloud providers to facilitate effective switching. Key obligations include a duty to refrain from undermining the extraction and export of data from one service to another; to preserve and effectively transfer a customer’s digital assets (applications, metadata, visualization technologies, and the like) to another provider; and to provide support in good faith to facilitate customers’ switching processes.
Importantly, the Data Act views switching as ‘a key condition for a more competitive market with low entry barriers for new providers of data processing services’ (Recital 78). In other words, the Data Act specifically targets switching costs in cloud computing with a view to ensuring contestable cloud markets. This context raises questions about the Commission’s decision to address cloud services under the DMA.
The Commission, by its own admission, is understaffed. Its resources are stretched thin across traditional antitrust and merger enforcement, the DMA, the Foreign Subsidies Regulation, and a host of other measures. As previous enforcement episodes have shown, the machinery of the DMA can be slow, expensive, and time-consuming. It may, therefore, be imprudent for the Commission to resort to it before fully exploring the tools available under the more tailored Data Act.
Consider, for example, that the Data Act foresees the introduction by the Commission of an ‘EU Cloud Rulebook’. This instrument is meant to facilitate switching between data processing services. Customers and cloud providers are openly invited to consult it for that purpose (Recital 96). But after three years since the Act’s entry into force, and more than half a year since it became applicable, this Rulebook has not been drawn up. Would it not be more appropriate to explore this avenue first (among others) before committing scarce resources to DMA proceedings?
The DMA cloud investigation also raises legal certainty concerns. The DMA applies alongside other regulatory tools that comprise the EU’s enforcement framework in the digital economy. However, the DMA’s interaction with other regulations has been criticized for insufficiently addressing potential overlaps and clashes. For its part, the Data Act establishes some ground rules for interacting with the DMA. To illustrate, undertakings designated as gatekeepers under the DMA may not be the beneficiaries of data portability obligations under the Data Act. This was designed to limit gatekeepers’ access to data and to prevent their positions from becoming more entrenched. Given this precedent of textual communication between the Acts, and given the links between cloud switching and contestability, the Commission should explain more fully why additional action under the DMA is necessary to address contestability in cloud markets.
One might object that the DMA offers something the Data Act does not: the gatekeeper designation mechanism and the obligations that come with it. The Data Act applies horizontally to all cloud providers, while the DMA can impose targeted obligations on a few players whose structural position may warrant more intensive intervention.
But the point is not that the DMA and the Data Act are substitutes. They are not. Rather, in the specific context of cloud services, the DMA investigation appears to target market frictions that the Data Act’s switching rules were specifically designed to address. If the Commission’s rationale is indeed facilitating contestability, then it should consider demonstrating more fully why the Data Act should not be given a chance to work—especially since it has been applicable for only six months. If, on the other hand, the Commission’s purpose is to ensure contestability through means other than switching (or pursue a more fairness-oriented approach), this would also benefit from clarification. For example, what other DMA obligations, aside from switching, would ensure a more contestable cloud market?
Almost two years ago, Mario Draghi warned against ‘overlaps and inconsistencies across the whole legislative chain’ of the EU. The latest enforcement action under the DMA toward cloud computing risks adding another link to that chain. A more prudent approach would first explore regulatory tools that are more tailored to maintaining contestable cloud markets.
Selcukhan Unekbas is a Postdoctoral Researcher at the European University Institute and Research Fellow at UC Berkeley Haas School of Business.
Share:
With the support of
