European Union antitrust officials dropped a bombshell on the cloud computing industry on June 25, 2026, with a preliminary finding that Amazon Web Services and Microsoft Azure meet the threshold to be designated “gatekeepers” under the Digital Markets Act. The announcement, made in Brussels, signals a seismic shift in how the bloc regulates critical digital infrastructure, directly targeting the AI and cloud services that underpin modern enterprise technology. If finalized, the decision could force both hyperscalers to dismantle practices the EU sees as anticompetitive, particularly those that lock customers into proprietary AI ecosystems.

This preliminary determination marks the first time the DMA’s core platform services designation has been extended beyond the original list of search engines, social networks, and app stores. The move underscores mounting concerns in Brussels that the rapid consolidation of artificial intelligence capabilities inside a handful of cloud providers threatens innovation and consumer choice. For Microsoft, the ruling hits Azure at a delicate moment: the company has woven generative AI tools like Copilot deeply into its cloud fabric, often making them exclusive to Azure customers.

Regulators are not acting in a vacuum. The preliminary finding follows a months-long investigation sparked by complaints from European cloud providers and open-source AI firms. These smaller players argue that AWS and Azure use their dominance in infrastructure to steer customers toward proprietary AI services, making it economically and technically painful to switch providers. The DMA’s gatekeeper label would impose strict dos and don’ts, such as a ban on self-preferencing, mandatory interoperability with third-party services, and data portability obligations.

What the DMA Gatekeeper Status Means for Cloud

The Digital Markets Act, in force since 2023, was designed to curb the power of Big Tech platforms that act as gateways between businesses and consumers. Until now, designated gatekeepers included companies like Google, Apple, Meta, Amazon (for its marketplace), and Microsoft (for Windows and LinkedIn). But cloud infrastructure services were conspicuously absent from the initial list, a gap the European Commission has been examining as AI models increasingly run exclusively on the big three clouds—AWS, Azure, and Google Cloud.

Under a gatekeeper designation, AWS and Azure would have to meet a series of obligations aimed at ensuring fair competition. Key requirements would likely include:

  • Interoperability mandates: Cloud providers must allow third-party AI models and tools to run seamlessly on their infrastructure without artificial performance penalties or hidden costs.
  • Self-preferencing bans: A cloud provider cannot prioritize its own AI services over those of competitors when customers are setting up or managing workloads.
  • Data portability and switching: Businesses must be able to move their trained AI models and data to rival clouds without excessive egress fees or technical lock-in mechanisms.
  • Transparency in pricing and terms: Clear disclosure of how AI services are bundled with core compute and storage, preventing opaque practices that make it hard to compare alternatives.

The preliminary finding does not yet make the designation official; AWS and Microsoft now have the opportunity to respond and present counterarguments before a final decision, expected later this year. However, the EU’s move alone sends a chill through quarterly earnings calls, as compliance could reshape revenue models built on sticky AI ecosystems.

Why Cloud AI Lock-In Is the Heart of the Matter

The EU’s concern centers on what Commissioner for Competition Margrethe Vestager has repeatedly called “cloud AI lock-in.” As enterprises race to adopt generative AI, many are effectively tethered to a single cloud provider because their models, training pipelines, and data stores are built using platform-specific services. For example, Microsoft’s Azure OpenAI Service gives exclusive access to GPT models, but only within Azure’s walled garden. Amazon Bedrock offers a curated set of foundation models, again only on AWS. This vertical integration creates a classic vendor lock-in: once a company trains a model on one cloud’s toolchain, migrating to another is prohibitively complex and expensive.

Smaller European AI developers argue that they cannot compete on merit because the hyperscalers leverage their infrastructure dominance to preference their own AI offerings. They point to practices like: offering free credits for proprietary AI services, bundling AI with widely used enterprise software (think Microsoft 365 and Azure), and designing APIs that are deeply intertwined with the underlying cloud’s native services. This, they say, stifles innovation and forces European businesses into long-term dependency on non-European tech giants.

The DMA’s gatekeeper framework is tailor-made for such arguments. Its core premise is that designated companies occupy such a powerful position that normal competition rules are insufficient. By bringing cloud AI under the DMA umbrella, Brussels is signaling that it views control over AI infrastructure as a chokepoint akin to how app stores control mobile software distribution.

Microsoft Azure’s Dilemma: Windows, Copilot, and the Enterprise

For Microsoft, the preliminary finding lands at a time when the company is betting its future on an “AI everywhere” strategy. Since the launch of Copilot in 2023, the tech giant has embedded AI assistants into Windows, Office, Edge, and most critically, Azure. The cloud platform has become the central nervous system for enterprise AI adoption, with tools like Azure Machine Learning, Cognitive Services, and the exclusive access to OpenAI’s models.

If Azure is designated a gatekeeper, Microsoft could be forced to decouple some of these integrations. For instance, it might have to allow third-party AI models to be listed alongside Azure OpenAI in its console without any preferential placement. It may also need to offer equivalent performance guarantees for non-Microsoft AI services and make it easier to export trained models in industry-standard formats to other clouds.

Windows enthusiasts watching this space will recognize the echoes of past antitrust battles. In the late 1990s, Microsoft fought a landmark case over bundling Internet Explorer with Windows. The DMA now threatens to impose remedies that could similarly unbundle Azure’s AI services from the core cloud. However, Microsoft has publicly committed to proactive compliance since the DMA’s inception, having previously adjusted Windows and LinkedIn practices. The company may argue that Azure is fundamentally different—it is a pay-as-you-go platform with many competitors like Google Cloud and IBM Cloud—but EU officials seem unconvinced given the lopsided market shares. AWS and Azure together control over half of global cloud infrastructure spending, with Azure’s share growing fastest thanks to AI demand.

AWS Under the Microscope

Amazon, too, faces major upheaval. AWS pioneered the modern cloud, and its dominance in IaaS is unparalleled. Its AI strategy, anchored by Bedrock and the SageMaker family, has been remarkably successful in tying enterprises to the AWS ecosystem. Amazon’s custom machine learning chips, Trainium and Inferentia, further deepen lock-in by offering superior price-performance only within AWS.

If designated a gatekeeper, Amazon would have to treat its own AI hardware on par with alternatives, meaning customers could request support for third-party accelerators without penalty. It might also be required to open its data centers to rival AI platforms on technical parity, a requirement that could upend its entire service model. Amazon’s legal team is expected to push back strongly, pointing to the highly competitive nature of cloud computing and the lack of switching barriers when compared to, say, social networks.

The Broader Impact on Windows Users and Developers

Though the DMA ruling primarily targets cloud platforms, its ripple effects will undoubtedly reach the Windows desktop. Many enterprise Windows users rely on Azure for identity, management, and security through services like Entra ID and Intune. These services are deeply integrated with Windows 11 and the upcoming Windows 12, providing seamless single sign-on and device management. If Azure must open up, third-party identity and management providers could demand equal integration with Windows’ native experiences, potentially complicating Microsoft’s lock on enterprise PC management.

Moreover, developers who build Windows applications using Azure AI services may find themselves with more choices. A world where Azure is forced to be more interoperable could lower costs and encourage multi-cloud AI strategies. This could, in turn, accelerate innovation in native Windows AI apps beyond the Copilot paradigm. However, it might also fragment the currently smooth developer experience that Microsoft has cultivated.

Reaction from EU Member States and Industry

The preliminary finding has already drawn mixed reactions. France and Germany, home to vocal advocates of digital sovereignty, have largely welcomed the move. French cloud provider OVHcloud and German data processor Ionos have long lobbied for such action, arguing that the big three clouds suffocate local champions. Conversely, the tech industry’s lobbying arms warn that heavy-handed regulation could slow AI adoption in Europe and give an edge to Chinese and American competitors who face no such constraints.

Political analysts note that the timing is delicate. The EU is simultaneously trying to foster homegrown AI models and infrastructure through programs like Gaia-X and the AI Innovation Package. Slapping AWS and Azure with gatekeeper rules could be seen as contradictory if it hampers the very platforms European startups rely on. Yet regulators insist the DMA’s design allows gatekeepers to thrive while ensuring markets remain contestable.

What Comes Next: A Timeline to the Final Decision

The preliminary finding initiates a formal consultation period running through September 2026. Both Microsoft and Amazon will submit detailed responses, and the Commission will hold hearings with competitors, customers, and independent experts. A final decision is expected by early 2027, with obligations taking effect six months after designation. During this period, gatekeepers must also appoint independent compliance officers and submit to regular auditing, per DMA rules.

For Windows and cloud professionals, the next eight months will be critical. Microsoft may proactively adjust some practices to avoid the most stringent remedies, much as Apple did by opening the App Store slightly. Industry watchers will scrutinize every update to Azure’s terms of service, pricing, and API documentation for signs of pre-compliance. Already, Microsoft has begun highlighting its use of open-source models and partnerships with Hugging Face and Meta as evidence that Azure is not a walled garden.

The Long View: Cloud AI Regulation Goes Global

The EU’s step is likely to reverberate far beyond Brussels. The United Kingdom’s Competition and Markets Authority has been conducting a parallel investigation into cloud services, and its interim report closely mirrors the EU’s concerns about AI lock-in. In the United States, the Federal Trade Commission under the current administration has expressed interest in non-discrimination rules for cloud platforms, though legislative action remains stalled. Thus, the DMA decision may become a template for global regulation.

For the Windows community, the message is clear: the AI tools that increasingly ship with every update—Copilot, AI-powered search, generative art—are not just features but components of a strategic ecosystem whose legal foundations are about to be tested. Whether the DMA can truly open up cloud AI remains an open question, but the preliminary finding ensures that the debate will dominate tech policy for years to come.

The EU’s move is a high-stakes gamble that could democratize AI infrastructure or inadvertently slow the very innovation Europe desperately wants to cultivate. For now, all eyes turn to Microsoft’s response and the mounting evidence that cloud AI lock-in is more than a theoretical worry—it is a market reality that regulators are no longer willing to tolerate.