OpenAI Commits $300 B to Oracle for Massive Data‑Center Power Capacity
Key Points
- OpenAI pledges $300 billion to Oracle for up to 4.5 GW of power capacity.
- Deal announced by major newspapers, set to start in 2027.
- Power amount equates to two Hoover Dams or four million homes.
- U.S. data‑center capacity nearly doubled from 2021‑2024; demand up 9% annually through 2030.
- OpenAI diversifies beyond Microsoft Azure, partnering with Oracle under its Stargate Project.
- Stargate Project involves $500 billion investment with Oracle, Microsoft, Nvidia and Softbank.
- Construction of a Stargate facility underway in Abilene, Texas.
- ChatGPT faces competition from Google Gemini, Anthropic Claude, and Perplexity.
- CEO Sam Altman warns of a possible AI bubble amid rapid growth.
OpenAI has reportedly pledged $300 billion to Oracle over a five‑year period to secure up to 4.5 gigawatts of power for new data‑center capacity, a deal announced by major newspapers and slated to begin in 2027. The agreement, part of OpenAI’s broader push to diversify beyond Microsoft Azure, ties into the company’s Stargate Project, a multi‑billion‑dollar effort to build AI infrastructure with partners including Oracle, Microsoft, Nvidia and Softbank. The deal reflects soaring demand for generative‑AI compute, with U.S. data‑center numbers having nearly doubled since 2021 and power consumption expected to double again by 2035.
OpenAI’s $300 B Commitment to Oracle
OpenAI is reported to have locked in a $300 billion commitment with Oracle spanning five years, aimed at delivering as much as 4.5 gigawatts of power capacity for new data‑center facilities. The scale of the contract, described by industry observers as one of the largest ever for cloud computing, was disclosed by both The Wall Street Journal and The New York Times and is slated to take effect in 2027. Oracle and OpenAI have not provided comment on the arrangement.
Power Scale and Impact
The 4.5 gigawatt figure translates to roughly the output of two Hoover Dams or the electricity used by four million homes, underscoring the massive energy appetite of modern AI workloads. Industry data shows U.S. data‑center capacity almost doubled between 2021 and 2024, with annual demand projected to climb about 9 percent through 2030. Analysts anticipate that by 2035 data‑center facilities will consume twice the electricity they do today, highlighting the strategic importance of securing reliable power sources.
Strategic Shift Beyond Microsoft Azure
Historically, OpenAI relied exclusively on Microsoft’s Azure cloud platform for its compute needs. The new Oracle partnership signals a deliberate diversification of its cloud portfolio, aligning with broader corporate initiatives to broaden infrastructure options.
Stargate Project and Wider Partnerships
Earlier in the year, OpenAI announced the Stargate Project, a $500 billion, four‑year investment aimed at building AI infrastructure at scale. The project brings together a consortium of partners—Oracle, Microsoft, Nvidia and Softbank—to develop the necessary data‑center capacity. In July, OpenAI and Oracle detailed an agreement to supply 4.5 gigawatts of power for Stargate, supplementing a prior pledge of 10 gigawatts announced in January. Construction of a Stargate facility is underway at a large site in Abilene, Texas.
Market Context and Competitive Landscape
ChatGPT remains the world’s most widely used chatbot, competing with offerings such as Google’s Gemini, Anthropic’s Claude and Perplexity’s software. The rapid expansion of generative‑AI products has prompted OpenAI CEO Sam Altman to voice concerns about a potential AI bubble, reflecting the high‑stakes environment surrounding AI development.