This deal is the latest in a string of data center investments involving OpenAI and other technology companies. Just after his inauguration in January, US President Donald Trump announced that OpenAI, SoftBank and Oracle will invest $100 billion in building AI data centers in the US and over time will contribute up to $500 billion to AI infrastructure.
The data center gold rush is partly based on the assumption that models will improve according to the laws of scaling – provided they are trained on more data and computation. “The basic story is that scale matters, and that’s been true for some time and may continue to be true,” Jonathan Koomey, a visiting professor at the University of California, Berkeley who studies computer and data center performance, told WIRED in September, talking about the OpenAI-Oracle deal to create three modern data center sites. “Many U.S. AI companies are making this bet.”
Derek Thompson, a prominent economics reporter, noted in a recent post that the tech industry is expected to spend about $400 billion on AI infrastructure this year, while U.S. consumer demand for AI services is only about $12 billion annually according to one study.
Altman noted on stage that the industry’s shift toward inference models is putting pressure on AI modelers in terms of long-term performance and implementation, and that OpenAI needs “tons of compute, tons of memory, and tons of processors” in addition to the Nvidia GPUs that the generative AI industry relies on so heavily.
At the event, Su described Altman as a “great friend” and an “artificial intelligence icon.”
Lauren Goode contributed to this report.
Update 10/02/25, 15:00 ET: This story has been updated to include additional comment from AMD.