OpenAI and Oracle Expand AI Partnership to 5-Gigawatt Capacity, Exceeding 10-Gigawatt Target

OpenAI and Oracle Corp. have announced a significant expansion of their collaboration, aiming to deploy 4.5 gigawatts of new AI computing power across the U.S. This initiative will further bolster OpenAI’s growing infrastructure, with the combined capacity expected to surpass 5 gigawatts when including the company’s existing projects in Abilene, Texas. The partnership underscores a strategic push to meet surging demand for large-scale artificial intelligence training and deployment.

The new data centers will rely on over 2 million chips to power AI workloads, with Oracle handling the infrastructure and OpenAI focusing on algorithmic development. Specific locations for the expansion are still under evaluation, with Texas, Michigan, Wisconsin, and Wyoming identified as potential sites. This aligns with a broader $500 billion commitment from Oracle and SoftBank Group Corp., initially announced in January, to build 10 gigawatts of AI infrastructure in the U.S. over four years. OpenAI now anticipates exceeding that target, though Oracle has not yet responded to requests for comment on the revised timeline.

A key component of the expansion is OpenAI’s internal Stargate project, which has already seen progress in Abilene. Parts of the facility are operational, with Oracle delivering its first batch of Nvidia GB200 racks and AI training activities underway. The success of this “beta test” has accelerated plans for scaling similar facilities nationwide. However, SoftBank’s role has diminished, as it is not involved in the current phase of the project. Earlier reports highlighted SoftBank’s financial challenges linked to global economic uncertainties, which may explain its reduced participation.

The collaboration includes a unique operational model: OpenAI will rent the expanded computing capacity from Oracle rather than managing the infrastructure directly. This arrangement allows the AI firm to concentrate on refining its models while leveraging Oracle’s expertise in data center operations. The project is also projected to create over 100,000 jobs, a significant figure for an industry typically associated with low labor intensity post-construction. The employment surge reflects a short-term focus on planning, construction, and power systems, rather than ongoing staffing needs for fully operational facilities.

Simultaneously, OpenAI is deepening its engagement with U.S. policymakers. The company will establish a permanent office in Washington, D.C., in early 2024, transitioning from a temporary co-working space. This move underscores OpenAI’s intent to strengthen ties with federal agencies and lawmakers, particularly after securing a $200 million contract with the U.S. Department of Defense and launching the OpenAI for Government initiative. The D.C. office will also house a new lab, The Workshop, designed to facilitate collaboration with educators, nonprofit leaders, and policymakers through hands-on tech demonstrations and training sessions.

With this multi-pronged strategy, OpenAI aims to solidify its leadership in the AI landscape while navigating complex infrastructure and political challenges. The partnership with Oracle highlights the critical role of corporate alliances in scaling cutting-edge technologies, while the emphasis on workforce development and regulatory engagement signals a broader effort to shape the future of AI responsibly. As the company continues to expand globally, its focus on both technical and political infrastructure positions it to address the evolving demands of the AI era.