OpenAI and Oracle Launch Stargate Project With Billion-Dollar Nvidia Chips

OpenAI and Oracle Corp. have teamed up to build the world’s largest data center as part of the $100 billion Stargate project. Tens of thousands of powerful Nvidia GB200 AI chips will begin installing at the Texas site in the coming months, the first step toward realizing the ambitious plan.

Tech Mar 11, 2025 0 36 Add to Reading List

OpenAI and Oracle Launch Stargate Project With Billion-Dollar Nvidia Chips

The Abilene facility is scheduled to have 64,000 GB200 chips installed by the end of 2026, with an initial phase of 16,000 units scheduled to be completed by the summer of 2025. This number of chips will provide the massive computing power needed to train and deploy generative AI models.

The Stargate project, unveiled by OpenAI, SoftBank Group Corp. and Oracle in January 2025, envisions up to 10 such data centers, underscoring the scale of the companies’ investment and ambitions in advancing AI technology.

Nvidia's GB200 chips are a key part of Stargate's infrastructure. While the exact cost of each chip has not been disclosed, Nvidia CEO Jensen Huang has previously said that the less powerful B200 chip costs between $30,000 and $40,000. So the chip costs could run into the billions of dollars in the first phase of the project alone.

Stargate is joining the tech giants’ race for computing power. Elon Musk’s xAI has invested $5 billion in servers for a Memphis supercomputer. Cloud provider CoreWeave Inc. already has more than 250,000 Nvidia GPUs in its data centers.

The Stargate project not only strengthens OpenAI and Oracle's position in the AI market, but also highlights the industry's growing dependence on powerful computing resources.