Microsoft leases 900 MW data center originally prepared for Oracle and OpenAI

Microsoft Rents 900MW Data Center Previously Reserved for Oracle and OpenAI

Microsoft will occupy a data center project that was previously reserved for Oracle and OpenAI. The project involves 900 megawatts of capacity — a rare-scale asset against the backdrop of acute scarcity in infrastructure for training and operating large AI systems.

What Happened

Microsoft has rented a data center project that was previously reserved for Oracle and OpenAI. Essentially, the company took over the facility after the two previous participants declined it. For the market, this is not simply a real estate or server transaction: in the era of generative AI, access to power supply and ready-made facilities becomes as strategically important as chips and the models themselves.

The key figure in this story is 900 megawatts. This volume shows that we are not talking about incremental expansion, but rather a major bet on future demand. Microsoft already remains one of the key players in AI infrastructure, and such projects help it scale computing power faster without the long cycle of building from scratch. If the facility is already prepared, the advantage amounts to months, and sometimes years.

Why It Matters

In the AI infrastructure market, value today is determined not only by GPUs, but also by the ability to physically house them at scale. Even if a company has money and access to accelerators, it still needs electricity, cooling, network connectivity, and operational permits. That's why the transition of such a facility from one set of tenants to another is an important signal about how priorities are being redistributed among the largest technology players.

• 900 MW is a scale designed for very large computational workloads
• A ready-made facility allows you to speed up the launch of new capacity compared to building from scratch
• Microsoft strengthens its control over scarce infrastructure for its own AI services
• The exit of Oracle and OpenAI shows that even the largest companies are reconsidering their facility plans

The fact itself that Oracle and OpenAI withdrew from the project while Microsoft quickly took their place is particularly important. This speaks to the high liquidity of such assets in an overheated market: if a facility truly suits large AI clusters, it won't sit idle. In such an environment, one company's decision to cut or postpone plans almost instantly creates a window of opportunity for another and allows for rapid redistribution of a scarce resource.

The Capacity Market Is Changing

The story of this facility illustrates what new competition around AI looks like. Not long ago, the main question was who had the stronger model and better product packaging. Now, on top of that layer, another has emerged: who is able to secure real physical resources faster. Without them, it's impossible to train new models, scale inference, and reliably serve corporate demand for AI services.

For Microsoft, this is especially sensitive because the company simultaneously develops cloud infrastructure and supports a large volume of AI workloads within its ecosystem. The more services, clients, and models tied to a single platform, the more expensive any capacity shortage becomes. That's why renting an already-prepared facility looks like a pragmatic move: not waiting for the market to free up, but taking a rare asset immediately when the opportunity appears.

What It Means

Competition for AI is increasingly moving away from the plane of just models into the plane of energy, land, and data centers. Microsoft's deal shows: those who win will be the ones who can not only create AI products but also pre-emptively ensure massive computational power for them. This makes infrastructure deals part of product strategy, and in turn, the value of pre-booked energy and ready-made facilities will grow.