OpenAI, Oracle Drop Data Center Expansion Plan in Texas ‘Stargate’ Project

OpenAI, Oracle Drop Data Center Expansion Plan in Texas ‘Stargate’ Project
Mar 8, 2026 00:48

Amid a massive surge in investments in artificial intelligence (AI) infrastructure, OpenAI and Oracle have cancelled a major expansion plan. The two companies have stepped back from expanding the flagship data center of the ‘Stargate’ project located in Abilene, Texas, United States. Funding constraints and fluctuations in OpenAI’s demand forecasts have been cited as the main reasons for the decision, according to a report by Reuters.

Citing a report by Bloomberg News, the news further states that the project was part of the $500 billion Stargate initiative announced in January 2025 by then U.S. President Donald Trump. Under the initiative—developed in partnership between SoftBank, OpenAI, and Oracle—a company named Crusoe is building a 1.2-gigawatt capacity data center campus in Abilene.

Since September last year, Oracle, Crusoe, and OpenAI had been discussing plans to double the facility’s capacity to nearly 2.0 gigawatts. However, the talks eventually failed due to funding challenges and frequent changes in OpenAI’s demand projections. Tensions in the Oracle–Crusoe relationship also played a role, particularly following incidents such as cooling system failures during recent winter weather, which caused service disruptions for several days.

However, the entire project has not been cancelled. Sachin Katti, Head of Infrastructure at OpenAI, said on social media that the Stargate main site remains one of the largest AI data center campuses in the United States. While expansion had been considered, the company ultimately decided to shift the additional capacity to other locations.

In this situation, Meta has shown interest in leasing the vacated expansion portion. The potential agreement is reportedly being facilitated by Nvidia, the world’s leading AI chip manufacturer.

According to the report, Nvidia has already provided Crusoe with a $150 million deposit and is assisting efforts to attract Meta. Nvidia’s strategy behind the move is to ensure that its chips are used in the expanded data center, rather than those of its rival AMD.

Meanwhile, Oracle’s agreement to build 4.5 gigawatts of data center capacity for OpenAI remains in effect. The two companies said the capacity will be developed at other locations, including another project near Detroit.

DBTech/BMT/OR