Stargate Data Center Set to Host 400,000 AI Chips from Nvidia

The first data center complex for the Stargate infrastructure project is set to accommodate up to 400,000 powerful AI chips from Nvidia. This information comes from a report by Bloomberg, citing a statement from Crusoe.

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

Stargate is a joint venture between OpenAI, Softbank, and Oracle. In January, Donald Trump announced its launch and revealed a plan for a $500 billion investment in AI infrastructure over four years, starting with an initial injection of $100 billion.

Bloomberg reports that Oracle has agreed to dedicate all its resources in Abilene, Texas, to the Stargate project. Currently, OpenAI aims to utilize about 1 GW of power at the site.

Presently, approximately 2,000 workers are involved in the project’s construction, with plans to expand the workforce to 5,000. Ultimately, eight buildings will be constructed, each designed to house 50,000 Nvidia GB200 chips.

OpenAI has previously stated that Stargate will expand to a total of 10 facilities nationwide.

The race for powerful data centers to support artificial intelligence is intensifying. In September 2024, the AI startup xAI launched what was then the largest AI training cluster, Colossus, utilizing 100,000 Nvidia H100 GPUs. Musk announced plans at the time to double the chip count to 200,000.

In December, reports emerged indicating that he aimed to deploy at least one million processors.

Earlier, Meta expressed its ambition to secure computing power from 600,000 Nvidia H100 chips by the end of 2024. The AI-focused cloud service provider, CoreWeave, currently operates over 250,000 Nvidia GPUs across 32 data centers.

It’s worth noting that in January 2025, experts questioned Stargate’s ability to invest $500 billion into AI.