OpenAI's Infrastructure Chief Sachin Katti Steps Down After Securing Double-Digit Gigawatts

Former xAI Executive Takes Control of OpenAI's Infrastructure Empire

OpenAI leadership change has reshaped the company's AI infrastructure division as Sachin Katti announced his departure from his role as OpenAI's infrastructure and compute lead on Wednesday1

1

. Uday Ruddarraju, who joined OpenAI in July 2025 from Elon Musk's xAI, will take over the entire compute organization after serving as CTO of compute since July 20262

2

. Katti will remain with OpenAI as an advisor after his final day on November 15, stating on LinkedIn that he "decided to move on to build something new"1

1

.

This transition matters because OpenAI's infrastructure operations underpin the company's ability to train and deploy increasingly powerful AI models. The compute organization develops the roadmap for massive-scale infrastructure projects that CEO Sam Altman believes will require trillions of dollars in investment5

5

. Ruddarraju brings direct experience from xAI, where he worked on the Grok 3 training run across 100,000 GPUs2

2

, positioning him to handle OpenAI's escalating infrastructure demands.

Major Compute Capacity Expansion Under Katti's Tenure

During his year at OpenAI, Sachin Katti secured double-digit gigawatts of new compute capacity across multiple projects, fundamentally expanding the company's infrastructure footprint4

4

. The most significant achievement was an 8-gigawatt OpenAI data center project in Ohio with SB Energy and NVIDIA, where NVIDIA guaranteed up to $105 billion in lease and power payments according to securities filings2

2

. Katti also oversaw a $30 billion Georgia facility, the first OpenAI designed and led itself, marking a strategic shift toward direct infrastructure control2

2

.

The scale of these projects reflects OpenAI's aggressive infrastructure strategy. The company has lifted its OpenAI compute spending forecast to roughly $750 billion by 2030, up from an earlier projection of $600 billion3

3

. This represents a separate commitment beyond the more than $1.4 trillion earmarked for AI infrastructure broadly3

3

. Katti credited partnerships with NVIDIA, Broadcom, AMD, Oracle, Microsoft, Google, Amazon Web Services, Cerebras, and CoreWeave for enabling this expansion4

4

.

Stargate Data Center Initiative and International Setbacks

Katti joined OpenAI in late 2025 specifically to oversee the company's Stargate data center effort, though the project initially moved slowly after OpenAI failed to reach a comprehensive deal with SoftBank and Oracle3

3

. He successfully pivoted to strike individual agreements with both companies and expanded OpenAI's strategy to include leasing its own data centers rather than solely renting capacity from providers3

3

. In August, this approach culminated in securing a 10-gigawatt data center lease with SB Energy backed by NVIDIA's financial guarantee3

3

.

However, what Uday Ruddarraju inherits is almost entirely American infrastructure. OpenAI paused Stargate UK in April, citing British electricity prices more than four times US levels and unresolved copyright rules2

2

. The UK government called the decision disappointing, and reports later emerged that OpenAI had never visited the announced north-east England site2

2

. OpenAI's sole announced European presence is a 230-megawatt data center in Norway revealed in July 2025, roughly 35 times smaller than the Ohio project alone2

2

.

GPT-Infra and Semiconductor Design Ambitions

Beyond traditional data center expansion, Katti launched GPT-Infra in July, a business applying OpenAI's models to semiconductor design and infrastructure development4

4

. This initiative represents OpenAI's push into designing its own computing hardware, including the custom Jalapeno inference chip and a reported $20 billion deal to use Cerebras AI chips4

4

. Katti recently noted that customers were already testing the models to develop central processing units and edge chips5

5

.

GPT-Infra will now move to Richard Ho, OpenAI's chip chief, who has expressed caution about fully trusting the models' design changes2

2

. This handoff raises questions about how aggressively OpenAI will pursue AI-designed semiconductors given Ho's reservations. The long-term implication is significant: if OpenAI can successfully use its own models to design better chips, it could reduce dependence on external suppliers like NVIDIA while accelerating hardware development cycles.

Wave of Executive Departures Continues

This OpenAI leadership change adds to a string of high-level exits throughout 2026. In July, Fidji Simo, OpenAI's product and business chief, moved into an advisory role to focus on recovery after a severe exacerbation of a chronic illness1

1

. The following month, Brad Lightcap, a longtime OpenAI executive, and Denise Dresser, who served as revenue chief, also departed1

1

. Infrastructure executives Chris Malone and Peter Hoeschele left earlier this year, with Hoeschele joining Meta alongside two colleagues4

4

. Safety lead David Robinson resigned in October4

4

.

OpenAI President Greg Brockman has downplayed concerns about the departures, telling CNBC he doesn't believe the wave is atypical for a young company and that heightened scrutiny comes with being in the spotlight1

1

. Brockman emphasized that different eras bring different leadership configurations and that he and Sam Altman provide continuity1

1

. Nevertheless, the concentration of senior departures in infrastructure roles specifically suggests potential challenges in executing OpenAI's ambitious compute expansion plans.

What to Watch: Infrastructure Challenges and Political Headwinds

Ruddarraju assumes control during a tense moment for AI infrastructure in the United States. Backlash against data centers has mounted over concerns about noise, electricity rates, and water consumption, becoming a flashpoint in upcoming midterm elections where both Republican and Democratic candidates have campaigned against the facilities1

1

. This political opposition could complicate OpenAI's plans to deploy double-digit gigawatts of new compute capacity across multiple states.

The short-term question is whether Ruddarraju can maintain the commercial partnerships Katti established while expanding OpenAI's infrastructure footprint. He acknowledged that Katti "has done an incredible job maintaining the commercial partnerships" and expects a smooth handover, noting he must now spend more time on the commercial side2

2

. Longer-term, watch whether OpenAI can successfully execute its international expansion beyond the single Norwegian facility, particularly if energy costs and regulatory environments in Europe remain unfavorable. The company submitted a confidential IPO prospectus to the SEC in June3

3

, suggesting infrastructure reliability will become even more critical as public market scrutiny intensifies.