OpenAI Partners with Oracle and SoftBank to Launch Five New Stargate Data Centers

OpenAI has announced plans to construct five new data centers in the United States as part of its Stargate initiative, in collaboration with Oracle and SoftBank. This expansion aims to increase Stargate’s total capacity to nearly 7 gigawatts, which is comparable to the output of seven large nuclear reactors.

During a press briefing in Abilene, Texas, OpenAI CEO Sam Altman emphasized the unique infrastructure demands of AI, stating that the United States must maintain its competitive edge in the sector. He praised Texas’s innovative spirit, suggesting it serves as a model for scaling infrastructure in a cost-effective manner.

The new data centers will be situated in Shackelford County, Texas; Doña Ana County, New Mexico; and a yet undisclosed location in the Midwest, developed in tandem with Oracle. This follows an earlier agreement between the two companies aimed at developing up to 4.5 gigawatts of U.S. data center capacity, building upon the existing facility in Abilene. OpenAI estimates that these projects, alongside a planned expansion of the Abilene site, will generate over 25,000 onsite jobs, despite the fact that fewer workers are typically needed once the centers are operational.

Two other sites in Lordstown, Ohio, and Milam County, Texas, will be developed directly by OpenAI and SB Energy, a SoftBank subsidiary. The Stargate project is significant as one of several major technology infrastructure initiatives launched since President Donald Trump took office, reflecting a commitment to bolstering American leadership in AI and creating substantial job opportunities.

Trump has actively promoted the Stargate initiative, suggesting it could accelerate advancements in AI and enhance the U.S.’s standing against China. His administration’s AI action plan has focused on expediting infrastructure development and minimizing regulatory barriers.

Originally, OpenAI described Stargate as a new company led by SoftBank’s CEO Masayoshi Son. However, it has since been clarified that "Stargate" is an umbrella term for all of OpenAI’s data center projects, excluding those in partnership with Microsoft. The flagship facility in Abilene, primarily owned by Oracle, is on track for completion by mid-2026.

Oracle is currently establishing eight data center halls in Abilene, each capable of supporting around 100 megawatts of power. The finalized facility is expected to accommodate over 400,000 GPUs, sustaining more than 1.4 gigawatts of power.

Following the initial announcement of Stargate, OpenAI expressed a strong interest from various firms wishing to participate, having reviewed around 700 potential sites across the U.S., with over 300 proposals for the new projects submitted by partners from roughly 30 states.

OpenAI’s recent policy white paper stresses the importance of AI infrastructure for U.S. competitiveness, comparing the rapid electricity resource development in China to America’s slower pace and highlighting the potential risks of falling behind in AI and industrial capacities.

Additionally, OpenAI has formed a strategic partnership with Nvidia, aiming to set up 10 gigawatts of data center capacity, significantly surpassing its engagement with Oracle. Nvidia is poised to invest up to $100 billion in OpenAI, with plans for the first Nvidia-powered system to be operational in the latter half of 2026.

Despite these ambitious plans, experts like UC Berkeley professor Jonathan Koomey caution that mere scale may not guarantee success over competitors that utilize less computing power more efficiently. This concern is echoed by anxieties regarding the environmental impact of expanding data center capacity.

Local support for the Abilene site is robust, with politicians like Mayor Weldon Hurt and Senator Ted Cruz advocating for a regulatory approach that prioritizes rapid development to ensure U.S. leadership in AI.

As OpenAI expands its Stargate initiative internationally, having established Stargate UK and Stargate UAE, the potential for delays due to trade uncertainty and high hardware costs remains a concern, with some analysts doubting the viability of such expansive and complex projects.

Total
0
Shares
Leave a Reply

Your email address will not be published. Required fields are marked *

Previous Article

Bridging the Divide: Insights from Riverbed's Survey on AI Readiness Gap

Next Article

‘SIM Farms’: The Spam Plague Threatening US Infrastructure – Insights from the New York Incident

Related Posts