In a high-stakes realignment of its foundational operations, OpenAI has consolidated its compute and infrastructure divisions under a single executive leadership role. The company has appointed Uday Ruddarraju—a veteran of Elon Musk’s xAI—to lead the newly unified unit. This strategic shift marks the departure of Sachin Katti, the former head of infrastructure who oversaw the company’s ambitious $30 billion "Stargate" data center initiative in Georgia.
The move, first reported by Bloomberg, signals a fundamental pivot in how OpenAI intends to manage the staggering costs and complex logistics of AI scaling. By merging compute and infrastructure under Ruddarraju, OpenAI is signaling an intent to tighten the integration between the hardware required to train its models and the physical facilities that house the GPU clusters.
The Executive Transition: From xAI to the Heart of OpenAI
Uday Ruddarraju is no stranger to the intensity of large-scale AI development. Joining OpenAI in July 2025, he initially stepped into the role of Chief Technology Officer for compute. His background is marked by his tenure at xAI, where he was instrumental in the orchestration of the Grok 3 training run—a massive undertaking that spanned a cluster of 100,000 GPUs.
His appointment resolves a mystery that has lingered within the industry since August. At that time, when OpenAI confirmed the departure of its former data center chief, Chris Malone, the company acknowledged it had already hired a new compute CTO but remained tight-lipped regarding their identity. It is now clear that the individual was Ruddarraju, whose expertise in managing massive-scale GPU deployments makes him a critical asset as OpenAI looks to transition from experimental infrastructure to industrial-scale data center operations.
Sachin Katti, who arrived at OpenAI from Intel late last year, will transition into an advisory role. Despite the departure, OpenAI has indicated that the leadership teams previously reporting to Katti will remain in place, ensuring institutional continuity during the handoff.
A Chronology of Infrastructure Flux
The consolidation follows a period of significant volatility within OpenAI’s infrastructure wing. The company has seen a series of high-profile departures and strategic resets that have tested its operational resilience.
- Late 2024: Sachin Katti joins OpenAI from Intel, where he served as Chief Technology and AI Officer, tasked with scaling the company’s massive physical footprint.
- March 2026: Katti officially takes the lead on the "Stargate" project, the company’s first internally designed and led data center site in Georgia.
- April 2026: OpenAI formally pauses its "Stargate UK" expansion. The decision, driven by concerns over British electricity costs—which were noted as four times higher than U.S. rates—and a lack of clarity regarding copyright regulations, dealt a blow to the company’s international ambitions.
- July 2025: OpenAI reveals a 230-megawatt data center project in Norway, marking its only major announced European infrastructure venture.
- August 2026: Data center chief Chris Malone departs the company, sparking questions about the stability of the infrastructure division.
- October 2026: OpenAI announces the consolidation of compute and infrastructure under Ruddarraju. Katti begins his transition to an advisory role.
These shifts occurred alongside the exit of several key engineers, including Peter Hoeschele, a central figure in the Stargate project, who departed for Meta earlier this year. His departure was accompanied by colleagues Keith Heyde and Shamez Hemani, according to reports from Data Center Dynamics.
Supporting Data: The Scale of the $856 Billion Ambition
OpenAI’s infrastructure strategy is defined by unprecedented capital expenditure. Recent filings shared with investors revealed a projected $856 billion compute bill, a figure that underscores the massive financial risks the company is currently navigating.
The scale of the company’s American footprint is massive compared to its international efforts. The Georgia "Stargate" site, an anchor of Katti’s tenure, represents a $30 billion commitment. Even more expansive is the 8-gigawatt project in Ohio, developed in partnership with SB Energy and Nvidia. According to reports from Axios, Nvidia has guaranteed up to $105 billion in lease and power payments for this site alone.
By contrast, the company’s European footprint remains marginal. The Norway project, while significant, is dwarfed by the Ohio site, which is roughly 35 times larger in terms of capacity. The failure to launch in the UK has effectively relegated OpenAI’s primary infrastructure ambitions to the North American market, where power grid availability and regulatory frameworks are currently more favorable to massive-scale AI clusters.
Internal Strategy and the "GPT-Infra" Handover
One of the most nuanced aspects of this reorganization involves "GPT-Infra," the specialized unit that leverages OpenAI’s own models to design and optimize semiconductors. Under Katti, this unit focused on utilizing AI to solve complex chip-design problems.
Following the restructure, this division will move under the purview of Richard Ho, OpenAI’s chip chief. This transition is particularly noteworthy given Ho’s recent public stance on AI-led design. Earlier this month, Ho suggested that he does not yet fully trust the design changes proposed by the models, highlighting a tension between the company’s reliance on its own technology and the need for rigorous, human-verified engineering standards in hardware development.
Implications for the Future of AI Scaling
The appointment of Ruddarraju suggests that OpenAI is prioritizing operational efficiency and high-density GPU management over the experimental phase of data center design.
1. Operational Integration
By uniting compute and infrastructure, OpenAI is removing the silos that previously separated the "digital" needs of model training from the "physical" realities of power delivery and cooling. Ruddarraju’s experience with the Grok 3 cluster suggests a focus on maximizing GPU utilization, which is essential for maintaining the competitive edge in the ongoing "compute arms race."
2. Commercial Pressure
In his initial comments regarding the transition, Ruddarraju noted that he must now focus more heavily on the commercial side of the business. "Sachin has done an incredible job maintaining the commercial partnerships," Ruddarraju stated. This suggests that the next phase of OpenAI’s infrastructure growth will be defined less by ground-breaking engineering alone and more by the complex negotiation of energy contracts, supply chain stability, and partnerships with firms like Nvidia and SB Energy.
3. The "Americanization" of Infrastructure
The pause of the UK project and the emphasis on the Ohio and Georgia sites indicate a strategic retreat to "safe" jurisdictions. With electricity costs and regulatory hurdles proving to be bottlenecks, OpenAI appears to be focusing its resources where it can guarantee stable, long-term power delivery. For investors, this is a signal that the company is moving toward a more conservative, cost-mitigation-focused approach to infrastructure, likely in response to the massive $856 billion compute projection.
4. A Culture of Transition
The departure of Katti and the earlier exit of Malone suggest a high-pressure environment where leadership tenures are measured by the ability to execute on hyper-growth targets. Katti’s parting statement—"I’m proud of what we’ve built and have total confidence in Uday and the team"—is a standard corporate valediction, yet it masks a broader reality: OpenAI is in the midst of a transition from a research-first organization to a massive, infrastructure-heavy utility company.
As Ruddarraju steps into his expanded role, the industry will be watching closely to see if he can harmonize the company’s aggressive compute demands with the physical limitations of the power grid. For now, OpenAI is doubling down on its domestic assets, signaling that for the immediate future, the path to AGI will be paved primarily in the American heartland.
