OpenAI Compute Chief Sachin Katti Is Leaving: Uday Ruddarraju Takes Over

OpenAI's top infrastructure executive is stepping down. Sachin Katti, who has led the company's computing capacity for about a year, told colleagues on October 14, 2026 that he is leaving. His last day is November 15, and he will remain an adviser. CNBC reports that Uday Ruddarraju, now OpenAI's CTO of compute, will take over the whole compute organization.

This is the kind of personnel news that looks like inside baseball until you remember what the compute team does. It decides where models are trained, how fast new data centers come online, and how much capacity is left over for the ChatGPT and API limits that builders actually feel. That is why a leadership change here is worth a careful read, and why the numbers attached to it deserve a check.

TL;DR: what is confirmed

What Katti said

According to CNBC, Katti wrote on LinkedIn that "building AI infrastructure at this scale has given me an incredible view into every dimension of the compute world, from silicon and systems to datacenters, energy, supply chains, financing, and the economics of AI." He added: "I'm a builder at heart, and I'm looking forward to building something new," and said he has "total confidence in Uday and the team to carry it forward."

That is a standard, friendly departure statement. It does not say where he is going, and no outlet we checked reported a destination. We will not guess one.

Who is Uday Ruddarraju?

Ruddarraju is not a new face on the compute team. CNBC, citing his LinkedIn, says he joined OpenAI in July 2025 and previously served as head of infrastructure engineering at Elon Musk's xAI, which CNBC notes was acquired by SpaceX earlier this year. Our earlier coverage of that deal is in Nvidia's $21 billion SpaceX stake and the xAI merger.

Press summaries of an August 2026 reorganization described Ruddarraju as overseeing data center operations under Katti's infrastructure org. We could not open the original report for that, so treat the exact August titles as second-hand. The firm facts are the ones OpenAI's spokesperson gave CNBC: he is CTO of compute today and will lead the overall organization afterward.

For readers who follow where talent moves between labs, this is another data point: the person inheriting OpenAI's buildout came from a rival's infrastructure team. The same movement shows up in other places, for example in AI talent moves across browser-agent startups.

Katti's background

Katti arrived from Intel. Bloomberg reported in November 2025 that Intel's chief technology and AI officer was leaving for a hardware role at OpenAI, and Intel CEO Lip-Bu Tan took over his AI responsibilities. OpenAI President Greg Brockman said at the time that Katti would design and build OpenAI's compute infrastructure. So his one year at OpenAI covered the period when Stargate sites, chip partnerships and multi-gigawatt deals turned from announcements into construction.

The spending numbers do not agree

Two figures are attached to this story.

CNBC writes that OpenAI is "targeting roughly $600 billion in total compute spend by 2030." A separate figure of about $750 billion in computing spend through 2030 circulated in September press reports, which we covered in OpenAI reportedly plans $750 billion in compute spend.

Why might they differ? We cannot say from the sources. Plausible explanations include different time windows, committed versus planned spending, or whether partner-financed capacity counts. That is our reading, not a reported fact. The practical rule is the same one we applied to the revenue numbers in OpenAI says $50B run rate, not $70B: attribute every large figure to whoever reported it, and do not mix them in one sentence.

A string of departures

CNBC places Katti's exit in a pattern. In July, Fidji Simo, OpenAI's product and business chief, said she would move into an advisory role to focus on recovery from a serious chronic illness. In August, Brad Lightcap and Denise Dresser, the revenue chief, also left. Our report on that month is OpenAI's exodus: Lightcap out.

Brockman told CNBC at the time that he does not see the departures as unusual for a young company, saying "every departure gets scrutinized in a way that it doesn't otherwise," and that he and Sam Altman are "constants." That is the company's framing. Reasonable observers can read the same list as turnover in a company growing faster than its org chart, or as a warning sign. The facts so far support neither strongly, and an adviser arrangement suggests an orderly handoff.

Why timing matters: data centers are now a political issue

CNBC points out that the change lands while backlash against data centers over noise, electricity rates and water use has become a midterm-election issue, with candidates from both parties campaigning against the facilities. OpenAI is involved in large projects in Ohio, Texas and Georgia.

That backdrop is visible across recent coverage:

• OpenAI wrote to the Texas governor about a data center pause: OpenAI letter to Governor Abbott.
• San Francisco moved on a moratorium: San Francisco's data center moratorium.
• A Senate investigation asked whether hyperscalers misled the public about data center costs: Senate investigation into hyperscaler data center costs.

Whoever runs OpenAI's compute organization now has to deliver gigawatts of capacity while negotiating local politics, power interconnection and supplier financing at the same time. Katti himself listed "energy, supply chains, financing" in his farewell note, which reads like a job description for that problem.

What this means for people who build on OpenAI

Pricing and limits. Nothing in the reports changes prices or usage limits. OpenAI has tightened and loosened limits several times this autumn, as in the GPT-6 Astra usage limit cuts. Capacity is the underlying lever, so a compute leadership change is worth watching, but a leadership change alone is not a signal that limits will move.

Roadmap risk. The compute plan relies on partners. If a site slips, the visible symptoms are rate limits, waitlists and delayed model rollouts. A smooth handoff to an internal successor, rather than an outside hire, lowers the chance of a disruption in the next two quarters.

Vendor concentration. If your product depends on a single lab's capacity, this is a reminder to keep a second provider wired in. Our comparison of current frontier options is in Muse Wonder vs GPT-6.1 Sol vs Opus 5.5.

Timeline of the compute leadership

1. July 2025: Per his LinkedIn as cited by CNBC, Ruddarraju joins OpenAI from xAI. He therefore predates Katti at the company by a few months.
2. November 2025: Bloomberg reports Katti leaving Intel for a hardware and infrastructure role at OpenAI; Greg Brockman says he will design and build the compute infrastructure.
3. July 2026: Fidji Simo moves to an advisory role for health reasons.
4. August 2026: Brad Lightcap and Denise Dresser leave, per CNBC.
5. September 2026: The $750 billion compute-spend report circulates, covered in our earlier post.
6. October 14, 2026: Katti announces his exit effective November 15; Ruddarraju is named to lead compute.

The ordering shows something useful: the successor has been inside the organization longer than the person he replaces. That is the strongest evidence for continuity, and the reason we read this as a handoff rather than a rupture, though only OpenAI can confirm its own intent.

What we do not know

• Where Katti is going next, and whether "something new" competes with OpenAI's suppliers.
• Whether the $600 billion and $750 billion figures describe the same thing.
• Whether Ruddarraju's current CTO of compute role will be backfilled, and by whom.
• Whether any specific data center timeline moves because of the change.

We will update this post if OpenAI or the Wall Street Journal publishes more. Primary sources to check: Katti's LinkedIn post, OpenAI's newsroom, and the CNBC report.

How to follow this story

If you track AI infrastructure, three signals matter more than the headline: new site or power announcements from OpenAI and its partners, any change to published rate limits for ChatGPT and the API, and Katti's next move.

Related reading

• OpenAI reportedly plans $750 billion in compute spend through 2030
• OpenAI's exodus: Lightcap out and the full departure list
• OpenAI says $50B run rate, not $70B
• OpenAI letter to Texas governor on a data center pause
• San Francisco data center moratorium
• Senate investigation into hyperscaler data center costs
• Nvidia's $21 billion SpaceX stake and the xAI merger

Details are accurate as of October 14, 2026 and based on CNBC and Wall Street Journal reporting; check the linked sources for later changes.