OpenAI now expects to spend about $750 billion on computing power through 2030, and it has handed control of that budget to Uday Ruddarraju. Bloomberg first reported the promotion. Ruddarraju will lead OpenAI's infrastructure and compute businesses, including the Stargate data center effort.

The move displaces Sachin Katti after less than a year in the role. Katti joined from Intel in November 2025 and took over Stargate this March. He is expected to remain at OpenAI as an adviser.

Who is Uday Ruddarraju?

Ruddarraju arrived at OpenAI in July 2024 as head of compute and infrastructure. This July, the company named him CTO of Compute. He came from xAI, where he served as head of infrastructure engineering at the SpaceX business.

His rise has been fast. Two promotions in roughly a year put him above colleagues with deeper OpenAI tenure. The company clearly wants an operator who has built large data centers before, and Ruddarraju brings that record from xAI.

He is not the only xAI veteran in the building. OpenAI's data center team has been steadily filled with former xAI staff as that company folds into SpaceXAI, soon to be rebranded SpaceXSI. Brent Mayo, OpenAI's head of data center build and delivery, is one. Others who have joined include site lead Aakash Verma, data center infrastructure staffer Amin Golnabi, project manager Andrej Sulek, and construction project manager Johnathan Henley. This month, former xAI lead commissioning engineer Michael Roberson joined OpenAI's infrastructure division.

Why did Katti lose the job?

The mechanism matters more than the personalities. Under Katti, OpenAI leaned hard on cloud partnerships because its own data center projects have been slow to come online. In-house capacity that was supposed to arrive has slipped. Meanwhile the company raised its compute spending projection from roughly $600 billion to about $750 billion through 2030 — a $150 billion jump.

Katti's tenure also coincided with turnover below him. Data center heads Peter Hoeschele, Keith Heyde, Shamez Hemani, and Chris Malone all left OpenAI following his promotion in March. That is four senior departures in one layer of the org chart, and it left gaps that Ruddarraju's xAI hires are now filling.

So the pattern is clear: OpenAI keeps changing leaders while the underlying problem — slow in-house builds and rising external spend — persists, and each reshuffle brings in more people who built xAI's infrastructure from scratch.

What does this change for OpenAI's buildout?

For anyone tracking Stargate, the practical question is whether Ruddarraju can deliver capacity faster than Katti did. The answer will show up in two places: the pace of Stargate data centers coming online, and the balance between owned capacity and rented cloud compute in OpenAI's cost structure.

The hiring signals intent. OpenAI has effectively transplanted a working data center team — site lead, build and delivery head, project managers, commissioning engineer — rather than assembling one from scratch. That suggests the company wants proven execution speed over internal continuity.

The stakes are the $750 billion number. Every month of delay in owned facilities pushes more of that spend toward cloud partners at rental prices. The $150 billion upward revision already shows the cost of that dependence, and investors will watch whether Ruddarraju's tenure bends the curve back down or merely manages the overage.

A pattern of churn

This is the third compute leadership arrangement OpenAI has tried in about a year: Katti arrives in November 2025, takes Stargate in March, and hands it to Ruddarraju months later. Four data center heads left after the March change. Now a new team, largely ex-xAI, runs the effort.

Continuity is gone; that much is certain. What replaces it is a bet that xAI-style infrastructure execution, applied at OpenAI scale, can close the gap between a $600 billion plan and a $750 billion reality before 2030.