Oracle has decided to push back the delivery of several data centers planned as part of its strategic partnership with OpenAI, until 2028, according to reports by Bloomberg. These delays concern sites originally expected in 2027 and are said to be due to difficulties sourcing materials as well as a labor shortage. Despite these adjustments, Oracle maintains high ambitions for the development of its infrastructure, notably in the United States.
In June, Oracle signed a large $300bn contract with OpenAI to provide the computing capacity needed to train and deploy its artificial intelligence models. The first data center linked to this partnership, located in Abilene, Texas, remains on track with the original schedule. Clay Magouyrk, Oracle's co-CEO, confirmed this week that more than 96,000 Nvidia chips have already been delivered to the site.
Neither Oracle nor OpenAI has officially commented on the delays, but the American group emphasizes its ability to meet its commitments globally. The data centers concerned are among the industry's most ambitious projects, and Oracle continues to aim for a leadership position in AI infrastructure despite the logistical challenges encountered.
