Sources revealed on Friday that Oracle and OpenAI have abandoned plans to expand a major artificial intelligence data center in Texas, United States. The decision follows prolonged negotiations over financing arrangements and OpenAI's shifting computational needs. This development creates an opportunity for Meta Platforms, which is now considering leasing the expansion portion of the project. According to insiders, Meta is currently in discussions with project developer Crusoe to lease the expanded data center facility in Abilene, Texas. AI chip leader Nvidia played a significant role in facilitating Meta's engagement with the developer to ensure future expansion would utilize Nvidia's AI chips rather than solutions from competitor AMD. This adjustment highlights the complexity of constructing hyperscale AI data centers. Such projects typically require investments reaching tens of billions of dollars and extensive collaboration among developers, cloud service providers, chip suppliers, and customers. The Abilene data center campus forms part of the high-profile "Stargate" project, which was announced by former President Donald Trump at the White House last year and spans approximately 1,000 acres. While the campus remains under construction with some facilities already operational, Oracle and OpenAI ultimately decided not to proceed with the previously discussed expansion plan that would have increased the data center's capacity from 1.2 gigawatts to approximately 2.0 gigawatts. Gigawatt-scale data centers represent enormous infrastructure, with one gigawatt of power equivalent to the output of a nuclear reactor, capable of supplying electricity to about 750,000 households simultaneously. Sources indicated that the negotiation breakdown stemmed from complicated financing arrangements and OpenAI's frequently changing projections for computational requirements. Additionally, the data center previously experienced several days of downtime due to winter weather affecting liquid cooling equipment, further straining relations between Oracle and Crusoe. Despite these challenges, both parties affirmed that their partnership remains solid. Oracle stated in a declaration that it takes pride in its collaboration with Crusoe and the project's progress, while Crusoe confirmed that the companies are jointly building one of the world's largest AI computing infrastructures. Meanwhile, Nvidia has paid Crusoe a $150 million deposit to secure continued use of its chips in the project and has helped facilitate Meta as a potential tenant for the expansion. Notably, the broader collaboration between Oracle and OpenAI remains unaffected. Oracle agreed in July last year to develop 4.5 gigawatts of data center capacity for OpenAI, with related projects still advancing, including other data center initiatives near Detroit. As AI model training and deployment drive surging demand for computational power, global data center construction is experiencing an unprecedented expansion cycle. Social media giant Meta is making substantial investments in AI infrastructure, projecting capital expenditures of up to $135 billion by 2026. The company is currently building large data centers in Louisiana and Indiana and signed an agreement last month to deploy AMD equipment with 6 gigawatts of capacity. Following the news, Oracle's stock closed down 1.18% at $152.96 on Friday. Other companies associated with AI infrastructure development, including CoreWeave, AMD, and Nvidia, also saw share price declines.
Comments