OpenAI’s massive Stargate project, an AI infrastructure partnership with Oracle and SoftBank, just announced five new sites, setting the stage for AI data centers to start popping up across the U.S. over the next decade.
OpenAI said on Tuesday that the new data centers, along with the project’s flagship site in Abilene, Texas and other ongoing projects, would bring Stargate to nearly 7 gigawatts of planned capacity and more than $400 billion in investment over the next three years. For context, one gigawatt can power roughly 750,000 homes.
Three of the new data centers are being developed by Oracle and will be located in Shackelford County, Texas; Doña Ana County, New Mexico; and an unnamed site in the Midwest.
The other two are being developed by SoftBank in Lordstown, Ohio, and Milam County, Texas.
Stargate was first announced this past January during a White House press conference with President Donald Trump. The plan called for $500 billion in investments over four years to build out AI infrastructure nationwide.
At the time, top executives from the three companies joined Trump to make the case that new infrastructure is essential for training and operating powerful AI models that could, for example, speed up the search for a cure for cancer.
Originally, the Stargate initiative was pitched as a joint venture that would be helmed by SoftBank’s CEO Masayoshi Son. It has now become an umbrella term for new data center projects between OpenAI and either Oracle or SoftBank. Recent reporting alleged the project had hit some stumbling blocks, but with this announcement, it now seems to be ahead of schedule.
Still, Stargate serves as an example of the circular nature of AI investments. For instance, details are still murky, but Wired reported that the already-in-the-works flagship Stargate site in Abilene is primarily owned and run by Oracle, with OpenAI serving as its main tenant. And just this week, Nvidia said it plans to invest up to $100 billion in OpenAI while also supplying the GPUs that power these data centers.
“AI can only fulfill its promise if we build the compute to power it. That compute is the key to ensuring everyone can benefit from AI and to unlocking future breakthroughs,” OpenAI CEO Sam Altman said in a press release announcing the new sites.
On his personal blog, Altman made a more dire warning, claiming that if the industry can’t reach 10 gigawatts of compute, it may have to decide which AI applications take priority. He pointed to examples like discovering a cancer cure versus providing customized tutoring for every student on Earth. The exec said it’s his goal to “create a factory that can produce a gigawatt of new AI infrastructure every week.”
“If we are limited by compute, we’ll have to choose which one to prioritize; no one wants to make that choice, so let’s go build,” Altman wrote.