On June 25, 2026, the European Commission told Amazon and Microsoft that it provisionally considers Amazon Web Services and Microsoft Azure to be "gatekeepers" under the Digital Markets Act — not because either company cleared the law's numeric thresholds, but despite the fact that neither did. The finding was reached jointly with the Netherlands Authority for Consumers and Markets (ACM), which the Commission's DMA portal credits with combining "knowledge and expertise" through one of the DMA's joint investigation teams. It is the first time Brussels has pointed the DMA at cloud infrastructure, and it does so through a route — Article 3(8) — built for exceptions, not the rule.
What the Commission actually found
The ACM's own account of the joint investigation is unambiguous about the legal posture: national authorities like the ACM cannot enforce the DMA themselves — that power sits exclusively with the Commission — but they can investigate domestically and support Brussels' casework, which is what happened here. The substantive finding is that AWS, the largest cloud provider operating in the EU, and Azure, the second-largest, function as "an important gateway between businesses and their customers," and that both benefit from entrenched user bases, high switching costs, and expansive AI-linked ecosystems. Under the DMA's default mechanism (Article 3(2)), a service qualifies as a gatekeeper by clearing quantitative thresholds — roughly €7.5 billion in EU turnover, tens of millions of monthly users, thousands of business users. Article 3(8) exists as a qualitative backstop for services that miss those numbers but still show a durable, entrenched market position. Amazon and Microsoft now have a formal opportunity to contest the finding before any final designation; if it stands, both companies would have six months to comply with the DMA's switching, interoperability, and data-portability obligations for their cloud businesses.
The case for scrutiny
The steelman here is real. Cloud infrastructure genuinely is a chokepoint for the modern economy — enterprise software, AI training, and increasingly government services all run through a small number of hyperscalers. Analysts at Tech Policy Press note that Amazon, Microsoft, and Google together control over 65% of the EU cloud market, a level of concentration that has only deepened as European providers' collective share fell from 26% in 2017 to roughly 10% a few years later. Egress fees, proprietary APIs, and bundled AI tooling can make switching between providers costly enough that "competition" becomes theoretical for a locked-in enterprise customer. Max von Thun of the Open Markets Institute, quoted by Tech Policy Press, frames the designation as overdue recognition of "documented use of unfair and anti-competitive tactics to lock out rival providers." If cloud dependency is as structural as electricity or telecoms once were, treating it as a gateway market is not an obviously unreasonable stretch — it is a judgment call about where market power actually sits today.
Why the qualitative route should worry regulators who want the DMA to last
But the way the Commission got here matters as much as where it landed. Truth on the Market's analysis makes the sharper case: the DMA's gateway concept was built for two-sided platforms where a business reaches consumers through an intermediary — an app store, a search engine, a social network. Cloud computing is an input layer, not an intermediation layer; a company hosting on AWS is not reaching its customers via Amazon the way a developer reaches iPhone users via the App Store. Treating infrastructure-as-a-service like a consumer gatekeeper risks category error dressed up as consistency.
More importantly, the entrenchment finding sits awkwardly against the market's own recent behavior. The same analysis points to AWS's EU market share declining even as Azure and Google Cloud gained ground, to basic cloud compute prices that have fallen more than 80% over the past decade, and to roughly 70% of enterprise cloud customers now multi-homing across providers — the opposite of the lock-in the DMA's qualitative test is supposed to detect. A market where the leader is losing share and most customers already split their business across vendors looks like ordinary, functioning competition, not a durable chokepoint.
There's also a comparative signal worth weighing: the UK's Competition and Markets Authority examined much the same cloud market and, according to Truth on the Market, chose supervised commitments over formal designation — a lighter-touch response to essentially the same facts. That divergence should prompt Brussels to ask whether it is applying a proportionate remedy or defaulting to its heaviest tool because the tool exists. Dr. Christophe Carugati of Digital Competition put the risk starkly to Tech Policy Press: stretching the DMA this way threatens to turn it "from a legal tool addressing identified concerns to a political tool supporting European industry development.**
The DMA's thresholds were meant to provide certainty precisely so regulation wouldn't hinge on discretionary judgment calls about who counts as powerful enough. If "gateway" can be read broadly enough to reach infrastructure providers whose market shares are moving the wrong way for an entrenchment story, the threshold test stops doing the work it was designed to do — and every large tech company operating in Europe, cloud or otherwise, should take note of how easily the qualitative door can open.