Oracle CEO Clay Magouyrk appears on a media tour of the Stargate AI data center in Abilene, Texas, on September 23, 2025.
Kyle Grillot | Bloomberg | Getty Images
oracle On Friday, the company pushed back against reports that the data center for OpenAI, one of its major customers, would be completed in 2028 rather than 2027.
A Friday report in Bloomberg, citing anonymous sources, said the delay was due to a lack of labor and materials. Oracle stock closed down more than 4% on Friday.
“The site selection and delivery schedule were established in close collaboration with OpenAI and agreed upon jointly after the contract was signed,” an Oracle spokesperson said in an email to CNBC. “There are no site delays and all milestones are on track to meet our contractual commitments.”
An Oracle spokesperson declined to provide a timeline for enabling cloud computing infrastructure for OpenAI. In September, OpenAI announced a partnership with Oracle worth more than $300 billion over the next five years.
“We have a great relationship with OpenAI,” Clay Magwik, one of Oracle’s two newly appointed CEOs, said at an analyst conference in October.
Business with OpenAI is relatively new for Oracle, which has been in business for 48 years. Historically, Oracle grew through selling database software and business applications. Although Oracle’s cloud infrastructure business now accounts for more than a quarter of its revenue, Oracle remains a small hyperscaler. Amazon, microsoft and google.
OpenAI also partners with other companies to meet anticipated capacity needs.
Nvidia announced in September that it had signed a letter of intent with OpenAI to equip the San Francisco artificial intelligence startup with at least 10 gigawatts of Nvidia equipment. The first phase of the project is expected to begin in late 2026.
“We look forward to finalizing the details of this new phase of our strategic partnership in the coming weeks,” Nvidia and OpenAI said in a statement in September.
However, no announcement has come yet.
“There is no guarantee that we will enter into a definitive agreement regarding the OpenAI opportunity,” Nvidia said in a November filing.
OpenAI has traditionally relied on Nvidia graphics processing units to power ChatGPT and other products, but the company is now also looking to work with OpenAI to design custom chips. broadcom.
On Thursday, Broadcom CEO Hock Tan revealed the timeline for the OpenAI effort, which was announced in October. Broadcom and OpenAI announced they have signed a term sheet.
“2027, 2028, 2029, something like 10 gigawatts. That was the OpenAI discussion,” Tan said on Broadcom’s earnings call. “And that’s what I call the agreement, which aligns where we’re going in terms of OpenAI, which is a very respected and valued customer. But we don’t expect much in 2026.”
OpenAI declined to comment.
WATCH: Oracle says there was ‘no delay’ in OpenAI deal even after stock price drop
