A designation built on interpretation, not arithmetic
On June 25, 2026, the European Commission announced its preliminary position that Amazon Web Services and Microsoft Azure should be designated as 'gatekeepers' under the Digital Markets Act (DMA) — the first time cloud infrastructure has been pulled into a regime built for consumer-facing platforms like search engines and app stores. The twist is procedural as much as substantive: neither company's cloud unit meets the DMA's quantitative thresholds — €7.5 billion in EU turnover, €75 billion market capitalization, or 45 million monthly users, as set out in Regulation (EU) 2022/1925 (EUR-Lex). The Commission instead invoked Article 3(8), which lets it designate a company through a qualitative 'market investigation' if it has 'an entrenched and durable position' as 'an important gateway,' even without hitting the numerical bar.
The Commission's own announcement frames AWS and Azure as the largest and second-largest cloud providers in the EU, with 'vast and entrenched user bases,' 'lock-in effects and high switching costs,' and — crucially — a 'portfolio of Artificial Intelligence (AI) tools and partnerships' that has become 'a decisive factor in cloud procurement' (Digital Markets Act portal, European Commission). In plain terms: AI bundling, not revenue or user counts, is doing the regulatory work. Amazon has already objected that the findings 'disregard the breadth of cloud services available to European customers and risk deterring European investment' (Data Center Dynamics). Both companies now have roughly 30 days to respond before a final decision, expected by the end of 2026, which would trigger a six-month compliance clock.
The case for intervening early
The Commission's underlying worry is not fanciful. Cloud infrastructure is no longer a back-office utility — it is the substrate on which AI training, deployment and distribution now run, and AWS and Azure are selling compute, model access and enterprise AI tooling as an integrated stack. If a handful of providers can use infrastructure lock-in to steer which AI models and applications enterprises adopt, that is a plausible channel for durable market tipping that ex-post antitrust cases, which take years to resolve, may arrive too late to reverse. That is the same logic — market tipping outruns reactive enforcement — that justified ex-ante regimes in the first place, and it is a real cost of waiting.
But the DMA's thresholds exist precisely to constrain how far that logic can be stretched. They were the price of legal certainty: a company can plan around a €7.5 billion turnover line in a way it cannot plan around a discretionary finding that its ecosystem is 'entrenched.' Reaching for Article 3(8) to bring in firms that clearly fall short of the bright line — rather than amending the DMA to formally lower or redefine the cloud threshold — signals that the numbers were never really binding, only a starting point for negotiation. That is a hard environment for any company deciding whether to build its next data center, or its next AI partnership, inside the EU.
Why this should matter in Delhi
India has already had this exact argument — and reached the opposite conclusion, for now. The Committee on Digital Competition Law's March 2024 report proposed an ex-ante Digital Competition Bill modeled partly on the DMA, designating 'Systemically Significant Digital Enterprises' using dual quantitative tests on turnover, market capitalization and user base, alongside 'qualitative criteria' on enterprise resources and data volume (PRS Legislative Research). Industry and MSME groups objected that low or elastic thresholds would sweep in fast-growing Indian platforms alongside Big Tech, and the Bill has stalled since — parliamentary review sent it back for reassessment rather than tabling it. The concern was identical to what AWS is now raising in Brussels: discretionary, qualitative designation criteria erode the predictability that ex-ante regimes are supposed to provide.
Having set that framework aside, India has defaulted to ex-post enforcement through courts and the Competition Commission of India — and the results illustrate the other failure mode. A Delhi High Court single judge ruled in May 2026 that Google's keyword-advertising practice constitutes trademark infringement; Google's July 7 appeal — heard July 10 before a Division Bench of Justices V. Kameswar Rao and Manmeet Pritam Singh Arora — argues the ruling makes India 'the sole outlier' among global jurisdictions on ad-tech law (MediaNama). Whatever the merits of that specific case, 'outlier by litigation' is not a substitute for a codified competition framework — it just relocates the unpredictability from a statutory threshold fight to a courtroom.
The actual lesson
India is trying to attract exactly the hyperscaler and AI infrastructure investment the EU's move threatens to chill — AWS, Azure and Google are all expanding Indian data center footprints. The right takeaway from Brussels is not 'revive the ex-ante bill' or 'trust the courts instead.' It is that whichever path India picks, the test for designation has to be countable in advance, not argued after the fact. A regime that can reach past its own numbers whenever regulators find the qualitative story compelling — whether that regime sits in Brussels or in a courtroom in Delhi — gives investors nothing to plan around, which is the one thing both ex-ante and ex-post models are supposed to provide.