Oracle has pushed back the delivery timelines for several major AI data centres it is building for OpenAI, shifting expected completion from 2027 to 2028, according to Bloomberg, which cited people familiar with the matter. These facilities form part of the large-scale Stargate AI infrastructure initiative announced earlier this year by Oracle, OpenAI, and SoftBank.
Sources told Bloomberg that shortages of skilled labour and certain “materials” are behind the delays, though it remains unclear whether the shortages refer to construction supplies, specialised data centre hardware, or supporting infrastructure needed for facilities of this magnitude. Despite the setback, the scope of Oracle’s commitments to OpenAI remains unchanged.
The delayed projects stem from a July agreement in which Oracle and OpenAI outlined plans to expand Stargate capacity to two million AI accelerators and roughly 5 GW of power consumption. The campuses—designed to be among the largest AI data centers in the world—were already considered ambitious in scale, making schedule adjustments unsurprising given the complexity involved.
Oracle Cloud Infrastructure CEO Clay Magouyurk acknowledged ongoing bottlenecks this week but insisted the company’s broader expansion plans remain viable.
He highlighted Oracle’s 147 active cloud regions, 64 additional regions under development, and roughly 400 MW of new data center capacity delivered in a single quarter. Magouyurk cited the company’s SuperCluster in Abilene, Texas—housing nearly 200,000 Nvidia GPUs and constructed within months—as an example of Oracle’s ability to scale rapidly.
“Our SuperCluster in Abilene, Texas, is on track with more than 96,000 Nvidia Grace Blackwell GB200 delivered,” Magouyurk said. “We also began delivering AMD MI355 capacity to customers this quarter.”
Magouyurk reaffirmed that demand for AI compute remains strong across both training and inference workloads, but emphasized that Oracle is selective about taking on new commitments. “We follow a very rigorous process before accepting customer contracts,” he said. “This process ensures that we have all the necessary ingredients to deliver to customer success at margins that make sense for our business.”
Despite the revised timelines, Oracle and OpenAI appear committed to executing their joint infrastructure roadmap. Once complete, the Stargate data centers are expected to rank among the most powerful AI computing sites globally, underscoring the growing scale and urgency of next-generation AI infrastructure development.
