- OpenAI and Oracle (ORCL) are set to install 64,000 Nvidia GB200 AI chips in a new Abilene, Texas data center by the end of 2026, with 16,000 chips due by summer 2025, as part of the $100 billion Stargate venture, per Bloomberg.
- The Stargate project, a collaboration with SoftBank unveiled in January, aims to expand to up to ten sites, with potential locations scouted in Pennsylvania, Wisconsin, Oregon, and Salt Lake City, reflecting its massive scale and multi-billion-dollar investment in Nvidia’s (NVDA) costly chips.
- This initiative competes with efforts like xAI’s $5 billion Memphis supercomputer deal, Meta’s (META) goal of 600,000 H100-equivalent chips by 2024, and CoreWeave’s (CRWV) 250,000 Nvidia GPUs, underscoring the race for AI computing power, as reported by Bloomberg.
A colossal new data center in Abilene, Texas, is set to become a cornerstone of the ambitious $100 billion Stargate infrastructure venture, with OpenAI and Oracle Corp. (ORCL) planning to equip it with 64,000 of Nvidia’s cutting-edge GB200 AI chips by the end of 2026, as reported by Bloomberg. This initiative, a collaboration unveiled in January at a White House event alongside SoftBank Group Corp., reflects the escalating demand for advanced computing power to fuel generative AI models, positioning Stargate as a key player in a fiercely competitive tech landscape. The phased rollout will see 16,000 chips installed by this summer, a significant step toward operationalizing the facility, which is just one of up to ten planned sites, highlighting the venture’s expansive vision.
The sheer scale of the Abilene project underscores the pivotal role of Nvidia’s semiconductors, with the 64,000 GB200 chips representing a multi-billion-dollar investment—Bloomberg notes that Nvidia’s less powerful B200 chips are priced between $30,000 and $40,000 each, suggesting a staggering cost for the GB200s alone. OpenAI, driving the design alongside Oracle, relies on the latter to procure and manage the supercomputer, a task Oracle has taken on without public comment, while Nvidia (NVDA) also declined to elaborate. This effort mirrors broader industry trends, with rivals like Elon Musk’s xAI securing a $5 billion deal for AI servers in Memphis, Meta Platforms Inc. (META) targeting capacity equivalent to 600,000 H100 chips by the end of 2024, and CoreWeave Inc. (CRWV) boasting 250,000 Nvidia GPUs across 32 data centers, according to its recent IPO paperwork.
Beyond Texas, the Stargate venture is eyeing further expansion, with OpenAI and SoftBank scouting potential sites in Pennsylvania, Wisconsin, Oregon, and Salt Lake City, where Oracle already operates cloud-computing infrastructure, per Bloomberg’s source speaking anonymously due to the internal nature of the plans. The Abilene facility’s initial 16,000-chip deployment this summer will mark a critical milestone, enabling early testing and deployment of AI capabilities that could redefine computational boundaries. As generative AI continues to reshape industries, the Stargate collaboration – backed by OpenAI’s innovation, Oracle’s operational expertise, and SoftBank’s strategic investment – positions itself at the forefront of a race to harness Nvidia’s latest technology, with the Texas site serving as a launchpad for a network that could transform the global AI infrastructure landscape.
WallStreetPit does not provide investment advice. All rights reserved.
Leave a Reply