OpenAI is shaking up the leadership of the teams responsible for the largest infrastructure buildout in the company's history. Sachin Katti, the computing chief who joined in late 2025 to oversee its data center effort, will leave next month, according to The Wall Street Journal. Uday Ruddarraju, a former xAI executive, will take over the company's computing strategy.
A leadership change at the heart of OpenAI's expansion For more context on this story, see our ongoing AI trends.
According to Bloomberg, which first reported the appointment, Ruddarraju will lead OpenAI's infrastructure and compute businesses. He joined OpenAI last year and was promoted to chief technology officer for compute in July, reporting to President Greg Brockman. Katti will transition out of his role over the coming weeks and plans to remain an adviser afterward.
"Sachin has done an incredible job maintaining the commercial partnerships," Ruddarraju said in an interview with Bloomberg. "Now I have to spend more time on the commercial side."
Katti confirmed the move in a statement: "I'm proud of what we've built and have total confidence in Uday and the team to carry it forward. I'm looking forward to building something new."
Why the computing role matters so much
The job Katti is leaving has become one of the most consequential in the AI industry. OpenAI's spending plans have ballooned as it races to secure the GPUs, power and land needed to train and serve its models. According to the Journal, the company has lifted its computing-power spending forecast to roughly $750 billion through 2030, up from a projection of about $600 billion earlier this year. That sits alongside a broader commitment of more than $1.4 trillion earmarked for AI infrastructure that OpenAI pledged to investors in February.
Sam Altman has said publicly that OpenAI will likely spend trillions of dollars on the physical infrastructure needed to develop new AI models and support wider adoption of the technology.
What Katti built, and what Ruddarraju inherits
Katti, a longtime Stanford professor who previously worked at Intel, joined OpenAI in late 2025 to oversee the Stargate data center effort. Per the Journal's account, the project moved slowly at first after OpenAI failed to reach a joint deal with SoftBank and Oracle, and instead struck agreements with the two companies individually. Katti added computing agreements with other providers such as Amazon Web Services and worked to revive OpenAI's plans to lease its own data centers rather than renting capacity from others.
Among the deals signed on his watch was a 10-gigawatt data center lease with SB Energy in August that carried a financial guarantee from Nvidia, according to the Journal. Bloomberg's report also credited Katti with an 8-gigawatt project in Ohio involving SB Energy and Nvidia, as well as a Georgia site that is the first data center designed and led by OpenAI itself. Since July, Katti had also overseen an effort called GPT-Infra, which applies OpenAI models to tasks such as chip and infrastructure design. That work will now fall under Ruddarraju, who said he expects a smooth transition given his prior role working with the engineering, construction and technical sides of the infrastructure group. The strategy and business leaders who reported to Katti will remain in their roles, which Ruddarraju said should provide stability.
What GPT-Infra signals about OpenAI's direction
One of the more unusual assets Ruddarraju inherits is GPT-Infra, the program Katti had run since July. The initiative applies OpenAI's own models to the business of building infrastructure itself — tasks such as semiconductor design and data center engineering. Katti said recently that customers are already experimenting with OpenAI models to develop central processing units and edge chips, an effort that turns the company's models into tools for the very supply chain it is trying to expand.
Katti's own path to the role was unconventional. The longtime Stanford professor spent far more of his career in academia than in operations, telling an interviewer in March that he did not have what he called a real job until he joined Intel nearly five years ago.
A broader management shuffle
The change is the latest in a string of departures from the teams sourcing, financing and building OpenAI's data center capacity. Chris Malone, an executive helping oversee the data center buildout, left in August. Peter Hoeschele, who played a central role in the Stargate initiative, exited for Meta earlier this year along with two colleagues.
The turnover extends beyond infrastructure. OpenAI announced in August that Chief Revenue Officer Denise Dresser and Chief Operating Officer Brad Lightcap were leaving, according to Quartz. The company also submitted a confidential IPO prospectus to the SEC in June, a sign that its internal operations are being restructured with public markets in mind.
For now, OpenAI's leadership is betting that Ruddarraju — who has already worked with partners including Nvidia CEO Jensen Huang — can bring stability to the strategy and business teams that remain in place. With a $750 billion spending forecast and gigawatt-scale projects lined up across multiple US states, the person running OpenAI's compute operation now effectively controls the pacing of the entire company's roadmap.
Analysts will be watching whether the churn slows execution. Data center capacity has become the binding constraint on AI progress, and every rival from Google to Anthropic to Meta is competing for the same constrained supply of land, power and chips. OpenAI can ill afford missteps as it scales toward targets that, on paper, exceed the entire annual electricity generation of many countries.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
