OpenAI claims to have surpassed its goal of 10GW of AI capacity ahead of schedule.

In a company blog post shared on Wednesday, OpenAI revealed that it has met 10GW ahead of its previous goal of by 2029, and in the past quarter alone had added 3GW to its footprint.

OpenAI launched its Stargate project in January 2025, with the aim of investing $500bn on AI infrastructure over the following four years and targeting 10GW of capacity.

With the company claiming that the 10GW goal has been achieved, it is unclear if it has also spent $500 billion on doing so. This number has been revised numerous times, with the latest projection suggesting $600bn over the next four years.

It also seems likely that the 10GW is contracted capacity, rather than live capacity, with OpenAI only claiming to have ended 2025 with 1.9GW of compute capacity, according to CFO Sarah Friar.

OpenAI wrote: "Our strategy has not changed because the reality has not changed: bringing the benefits of AI to more people requires more compute ... The financing models and partnership structures may evolve, but what matters is capacity coming online at scale, on time, and in a way that preserves flexibility as technology and demand evolve."

Certainly, the approach has changed over the past year and three months. Originally, OpenAI was focusing on building its own sites, but it has increasingly turned to cloud capacity, restructuring its compute and infrastructure teams to accelerate this in March 2026.

Cloud contracts signed by the AI lab include an agreement to use 2GW of Amazon Web Services' (AWS) Trainium chips, expanding its cloud contract by $100bn, in late February 2026, shortly after the company said it planned to spend around $600bn on compute over the next four years, seemingly a reduction from the $1.4 trillion planned over eight years. OpenAI is also a known customer of Microsoft, Oracle, and CoreWeave, among others.

Looking to the future, OpenAI has said that it, along with its partners, is "evaluating potential data center locations across the country (the US), and will significantly expand our compute capacity in the years ahead. These projects are complex, and they require the right combination of power, land, permitting, transmission, workforce, community support, and partner readiness.

"But the reason we are looking is simple: demand for AI is growing, and we are moving quickly to secure the infrastructure needed to meet it."