At a time when geopolitical crises dominate the headlines, the UK is preparing to make a less visible but highly consequential choice about its digital infrastructure. The Competition and Markets Authority (CMA) board will soon decide whether to intervene in cloud services under its digital competition regime.

The cloud services sector is the backbone of Britain’s digital economy. A year ago, its investigation concluded that competition was not working as well as it could, pointing to barriers that make it harder for alternative providers to compete with Amazon and Microsoft, the UK’s two largest cloud firms. The case team recommended a possible designation under the DMCC regime as a first step before intervention with potential requirements. The CMA’s board should resist.

The authority has no shortage of tools. It can accept commitments, enforce traditional competition law after the fact, or impose regulatory measures under its digital ex-ante competition regime before the fact (DMCCA). The latter is the most intrusive option, designed for markets where dominance is entrenched and competition is structurally impaired. The CMA provisionally argued that this was the case: limited customer choice dominated by Amazon and Microsoft, high commercial and technical switching costs to move to alternative cloud providers, Microsoft's licensing restrictions on its software for US cloud providers, and steep investment requirements for new entrants.

These concerns are not without merit. But they risk being overtaken by events.

Start with choice. UK cloud customers are not confined to two providers. Google, Oracle, IBM and Alibaba all compete actively in the market. Dominance is not, in itself, a market failure, especially when it reflects sustained investment and innovation.

And investment is surging. Globally, Amazon, Microsoft, Google, and Meta are expected to spend nearly USD$700 billion on AI infrastructure in 2026, a 60% increase from 2025. Amazon alone plans to invest close to USD$200 billion this year. Google, the third-largest provider in the UK, is expected to spend between USD$175 billion and USD$185 billion in 2026, nearly doubling its 2025 outlay. In the UK alone, Microsoft, Amazon, and Google have announced multi-year investments of USD$30 billion, USD$54 billion and USD$5 billion, respectively, to support Britain’s AI infrastructure.

To put this in perspective, each of these global figures represents a substantial share of the UK Department of Health’s £205 billion budget in 2024-2025 (around USD$275 billion), while UK-specific figures are comparable in scale to the Ministry of Defence’s £60 billion budget (around USD$80 billion). This is not a stagnant market driven solely by large cloud providers. It is one of the most capital-intensive and strategically contested sectors in the global economy.

New entrants are arriving just as quickly. AI labs such as OpenAI, alongside “neocloud” providers like CoreWeave and Nebius, which rent chips, are investing at scale. In February 2026, OpenAI alone announced a USD$110 billion investment programme to build its own AI infrastructure. Far from entrenching incumbents, this wave of investment signals an intensifying AI infrastructure race.

Innovation is also keeping pace. Cloud providers are rolling out new tools, from Amazon Bedrock and Microsoft Copilot Studio to Google Vertex AI, to meet booming demand for developing and deploying AI services quickly at scale.

Innovation is not limited to services; it extends to hardware as well. Major cloud providers, including Amazon, Microsoft, and Google, are offering their own chips, reducing reliance on Nvidia and increasing supply in a constrained market marked by chip shortages. This strengthens not only competition but also resilience.

Regulation abroad is also reshaping the market. In the European Union, the EU’s Data Act is lowering switching costs by removing fees for moving data to alternative cloud providers and by easing technical barriers. Major providers, including Amazon, Microsoft, and Google, have already applied these commercial changes globally. Microsoft has adjusted some aspects of its licensing practices for a selection of European cloud providers in response to European scrutiny. These shifts directly address several of the CMA’s original concerns.

None of this suggests that the CMA should ignore competition concerns. But it should match its tools to the evidence. Heavy-handed regulatory intervention risks overshooting a market that is evolving rapidly and already addressing many of the issues identified.

A more proportionate approach would be to engage through structured dialogue with prospective designated firms and relevant third parties, resolving concerns through targeted voluntary commitments before launching a formal investigation. This would enable concerns to be addressed in a timely manner rather than through a resource-intensive and time-consuming process.

The test for contemplating intervention under the UK digital competition regime should be simple: is competition truly failing, or is it working? If the evidence does not show market failure, the CMA board should acquit.