OpenAI secures 10 GW computing capacity ahead of schedule

OpenAI has closed one of the most ambitious infrastructure goals in the AI industry ahead of schedule: the company has already secured 10 GW of computing capacity through contracts. Initially, this milestone was planned for 2029, but the last 3 GW was contracted in just 90 days.

The Plan Accelerated Sharply

For OpenAI, this is far more than an impressive number for reports and presentations. In the era of large language models, computing capacity has become a strategic resource on par with data, engineers, and capital. The sooner a company locks in future computing volume, the lower the risk that new model releases, agentic services, and enterprise products will hit a simple infrastructure bottleneck.

Essentially, OpenAI is pre-booking growth capacity while competitors continue fighting over the same data centers, GPUs, and power connections. The pace of recent months is particularly telling. What previously looked like a 10 GW target for the second half of the decade is now achieved well ahead of schedule.

Three additional gigawatts in 90 days is an extremely aggressive pace even by market standards, where such deals typically take months to close. This jump reflects not only OpenAI's ambitions but also the readiness of suppliers, investors, and infrastructure partners to quickly reserve large capacity volumes for it.

Contracts and Reality

It's important not to confuse signed contracts with the immediate deployment of all capacity. The fact of agreements alone doesn't mean OpenAI can simultaneously load all 10 GW with model training and inference today. Some capacity will roll out in phases: some sites need completion, some require equipment delivery, some await grid connections.

At this scale, delays and phased deployment are nearly inevitable, so formally "securing access" and "getting everything at once" are two different stories. But even so, the outcome changes the landscape. The company reduces the risk in advance that growing demand for ChatGPT, APIs, and future products will hit shortages of hardware, electricity, or available space.

For the entire industry, this is a signal that the AI race is increasingly shifting from demos and releases to energy, construction, and contracts.

• Reserve capacity for training larger models
• Ability to scale inference for ChatGPT and API
• Resources for new agentic and enterprise services
• Stronger negotiating position with suppliers and investors

Money and Allies

Another important nuance is who pays for and backs such deals. OpenAI often attracts investment while simultaneously relying on partners who then provide computing services or related infrastructure. The result is a tight coupling of capital and capacity: belief in the company's future growth directly translates into access to data centers and power.

The higher the confidence in OpenAI's business, the easier it is to reserve new volumes in advance and the lower the chance of running out of resources at critical moments. For partners, this model is also rational. They get a customer with nearly guaranteed multi-year demand and can expand infrastructure under already-confirmed load.

For investors, it's a bet not just on the next model but on a future share in the critically important part of the AI market. So the 10 GW news looks not like an accounting update but as an indicator that the battle between OpenAI, Google, Anthropic, and other players is increasingly waged over energy, space, racks, and the speed of data center construction.

What This Means

The main bottleneck in AI now isn't just talent and algorithms, but infrastructure. Whoever locks in energy and computing first will gain an advantage not for one model release but for years ahead.