Stargate Data Center to House 400,000 AI Chips

The inaugural data center complex of the Stargate infrastructure project will accommodate up to 400,000 powerful AI chips from Nvidia, according to Bloomberg, citing a statement from Crusoe.

Construction of the facility is set to be completed by mid-2026, with a capacity of 1.2 GW.

Stargate is a joint venture between OpenAI, Softbank, and Oracle. In January, its launch was announced by Donald Trump. The project plans to invest $500 billion in AI infrastructure over four years, with an initial investment of $100 billion.

Oracle has agreed to dedicate all its capacity in Abilene, Texas, to Stargate, Bloomberg reports, citing sources. OpenAI currently intends to utilize about one GW of capacity at the site.

Currently, around 2,000 people are involved in the project’s construction, with plans to increase the workforce to 5,000. Ultimately, eight buildings will be constructed, each designed to house 50,000 Nvidia GB200 chips.

OpenAI previously stated that Stargate will expand to 10 sites nationwide.

The race for powerful data centers fueling artificial intelligence is accelerating. In September 2024, the AI startup xAI team launched the largest AI training cluster at the time, Colossus, using 100,000 Nvidia H100 GPUs. Musk then announced plans to double the number of chips to 200,000.

In December, media reported his intention to deploy at least one million processors.

Previously, Meta declared its ambition to secure computing power from 600,000 Nvidia H100s by the end of 2024. AI-focused cloud computing provider CoreWeave boasts over 250,000 Nvidia GPUs across 32 data centers.

Earlier in January 2025, experts questioned Stargate’s ability to invest $500 billion in AI.