Oracle and OpenAI abandon Texas data center expansion plans; Meta may take over the project

According to media reports, Oracle and OpenAI have abandoned plans to expand a flagship AI data center in Texas, USA.

Sources say that the project was ultimately shelved due to prolonged deadlock in financing negotiations and changes in OpenAI’s own needs.

The project is part of the “Stargate” initiative, a joint effort involving SoftBank Group, OpenAI, and Oracle, with a total investment of up to $500 billion and a capacity of 10 gigawatts. President Trump announced this project in January 2025.

This adjustment highlights the complexity of building AI data centers. Such projects typically involve investments of hundreds of billions of dollars and require coordination and support from multiple partners.

Last September, the aforementioned companies announced plans to further expand the existing Stargate core campus in Abilene, Texas, with a maximum data center capacity of 600 megawatts.

However, a knowledgeable source said on Friday that the additional computing capacity originally planned for this expansion will now be handled by other data center campuses under construction.

Currently, the Abilene data center campus has eight buildings operated by Oracle Cloud Infrastructure, with two already in operation. The source added that the overall plan for OpenAI and Oracle to build an additional 4.5 gigawatts of data center capacity is still underway.

In recent years, large tech companies have invested hundreds of billions of dollars into data centers to support generative AI services like ChatGPT and Copilot, which require enormous computational resources.

Meanwhile, the breakdown of negotiations between Oracle and OpenAI has created an opportunity for Meta. Reports indicate that Meta is considering leasing an expanded data center project in Abilene from developer Crusoe, with Nvidia assisting in the negotiations.

The data center Crusoe is building in Abilene is one of the most notable projects currently. Oracle is rapidly deploying servers within the campus, which are used by OpenAI to train and operate its AI products.

Currently, Oracle and OpenAI are using Nvidia’s AI chips in the Stargate project. Reports say Nvidia’s involvement in the negotiations aims to ensure that the expanded data center continues to use its chips rather than those of competitor AMD.

Regarding the above reports, OpenAI Infrastructure Lead Sachin Katti responded on social media: “Our flagship Stargate campus is one of the largest AI data center campuses in the United States. We have indeed considered further expansion, but ultimately decided to deploy additional capacity elsewhere.”

(Source: Cailian Press)