Oracle and OpenAI Use Powerful Nvidia AI Chips for Stargate Project

OpenAI and Oracle are planning to install tens of thousands of AI chips from Nvidia into a new Texas data center in the near future. This is one of the initial steps in getting the first facility that is part of their $100 billion Stargate infrastructure project. Data Center Knowledge reports that the data center is anticipated to have 64,000 of Nividia’s GB200 semiconductors by the end of 2026.

The 64,000 chips alone showcase the large amounts of power needed just for the beginning phases of one data center built for one customer. Nvidia’s chips are primarily used to train and deploy generative AI models. Stargate is just one venture among major tech companies, such as Meta, who are investing billions to have high-powered computing capabilities to accommodate AI.

Although Nvidia has not announced the cost of the GB200, considering 2024’s less powerful B200 chip costs between $30,000 and $40,000, the math easily shows that this Texas facility will cost billions for just the chips.

Other potential Stargate data center facilities may be built in Pennsylvania, Wisconsin, Oregon, and Salt Lake City.