OpenAI has restructured its leadership in light of its decision to rent more AI servers from cloud providers.

The company is in the midst of a major data center buildout in the US and around the world through its Stargate project, being undertaken in partnership with Oracle and SoftBank. However, reports of late have suggested that the company is instead leaning more on renting AI servers from cloud providers and scaling back its build-out plans.

As reported by The Information, and citing two people familiar with the shift in strategy, OpenAI has thus divided its compute and infrastructure teams into three groups: those designing the data centers, those working on commercial partnerships with cloud providers and chip companies, and those managing the data centers OpenAI is using.

In addition, The Information states that 'Stargate' now refers to the broader strategy of bringing AI compute online.

Sachin Katti, formerly of Intel and who joined OpenAI in November 2025, is now overseeing the Stargate groups.

“AI demand is moving fast, and our compute team has built a powerful ecosystem across clouds, chips, and infrastructure to meet it,” Katti said in a statement. “The progress has been incredible, and I’m excited about the work ahead as we keep building the systems that bring AI breakthroughs to people everywhere.”

Prior to his arrival, OpenAI leaders Peter Hoeschele and Chris Malone oversaw data center groups and reported to president Greg Brockman. Now, The Information reports, Hoeschele reports to Katti, and hands the commercial relationships with cloud and chip companies.

The technical engineering and design team is now co-led by Malone and former Microsoft engineering leader Adrian Caulfield.

Notably, OpenAI dropped plans to expand its Abilene Stargate site earlier this month. Previously, it was set to expand to 2GW, but OpenAI's use will now be capped at 1.2GW. The AI startup said: " We considered expanding it further, but ultimately chose to put that additional capacity in other locations. Today we have more than half a dozen sites under development across multiple states, including the site we're building with Oracle in Wisconsin, where the first steel beams went up just this week."

This news comes in the midst of a slurry of cloud contracts signed by OpenAI. The company signed on to use 2GW of Amazon Web Services' (AWS) Trainium chips, expanding its cloud contract by $100 billion, in late February 2026, shortly after the company said it planned to spend around $600 billion on compute over the next four years, seemingly a reduction from the $1.4 trillion planned over eight years.

OpenAI is also a known customer of Microsoft, Oracle, and CoreWeave, among others.