Oracle Corp. strongly denied reports on Friday that it postponed completion dates for artificial intelligence (AI) data centers being built for OpenAI, which sent its stock down more than 4%.
Bloomberg News reported Oracle had revised delivery schedules for several large AI data centers from 2027 to 2028, citing unnamed sources who attributed the delays to shortages of skilled labor and materials. The facilities are part of Oracle’s commitments under the Stargate infrastructure program, a joint initiative announced in January with OpenAI and SoftBank.
“Site selection and delivery timelines were established in close coordination with OpenAI following execution of the agreement and were jointly agreed,” an Oracle spokesperson said. “There have been no delays to any sites required to meet our contractual commitments, and all milestones remain on track.”
The disputed data centers stem from a contract signed in July between Oracle and OpenAI to expand Stargate capacity to two million AI accelerators and five gigawatts of power. In September, OpenAI announced the partnership was worth more than $300 billion over five years.
Despite the controversy over timelines, Oracle executives have expressed confidence in their expansion capabilities. Clay Magouyrk, one of Oracle’s two newly appointed CEOs, pointed to the company’s track record at an October analyst meeting, highlighting 147 active cloud regions and 64 more in development. He cited Oracle’s SuperCluster in Abilene, Texas — which houses nearly 200,000 NVIDIA Corp. GPUs — as proof the company can execute large-scale projects in months rather than years.
“Our pace of capacity delivery continues to accelerate,” Magouyrk told analysts last week. “We follow a very rigorous process before accepting customer contracts. This process ensures that we have all the necessary ingredients to deliver to customer success at margins that make sense for our business.”
The data center buildout represents a strategic shift for Oracle, a 48-year-old company historically known for database software and business applications. Cloud infrastructure now contributes over one-fourth of Oracle’s revenue, though it remains smaller than hyperscale competitors Amazon.com Inc., Microsoft Corp., and Google.
Meanwhile, OpenAI continues diversifying its infrastructure partnerships to meet surging demand. NVIDIA announced in September a letter of intent to deploy at least 10 gigawatts of equipment for OpenAI, with the first phase expected in late 2026 — though NVIDIA noted in November that no definitive agreement has been reached. Additionally, Broadcom CEO Hock Tan said Thursday that his company’s chip collaboration with OpenAI targets deployment in 2027-2029, with minimal impact expected in 2026.
OpenAI declined to comment on the Oracle dispute.
