Oracle has pushed back completion dates for some data centers it is developing for artificial intelligence model developer OpenAI to 2028 from 2027, according to people familiar with the matter. The delays are largely due to labor and material shortages, said the people, who asked to remain anonymous while discussing private timelines. Since signing a $300 billion contract this summer, Oracle has been working to meet its commitments to provide the computing power needed to train and run Open AI models. Despite these delays, project timelines in the United States remain ambitious for sites that will become one of the largest in the world. Oracle and OpenAI declined to comment. Oracle’s share price fell by up to 6.5% on the impact of this news. The stock fell 5.3% to $188.26 at 11:03 a.m. New York time on Friday. “We have ambitious and achievable goals to deliver capabilities globally,” Oracle co-CEO Clay McGuirk said on an earnings call this week. He added that the first data center developed by the company for “Open AI” – in Abilene, Texas – continues according to the planned path, with more than 96 thousand “NVIDIA” chips delivered, according to what he said during the call.