NextFin News - The most expensive construction project in human history is being managed by a man who, until recently, spent more time in lecture halls than in server rooms. Sachin Katti, a former Stanford professor and Intel executive, has emerged as the primary architect of OpenAI’s "Stargate" initiative, a trillion-dollar infrastructure buildout that aims to redefine the physical limits of artificial intelligence. As of March 11, 2026, Katti’s role has shifted from theoretical advisor to the high-stakes operator of a project that consumes more capital than the annual GDP of most G20 nations.
The scale of the ambition is staggering. According to Bloomberg, Katti is overseeing a multi-phase expansion that includes the flagship Stargate site in Abilene, Texas, and five newly unveiled data center locations across the United States. These facilities are designed to house hundreds of thousands of Nvidia’s latest Blackwell-successor chips, requiring power loads that have forced OpenAI to negotiate directly with nuclear energy providers. The project is no longer just about software; it is a massive bet on the industrialization of intelligence, where the winner is determined by who can secure the most gigawatts and the most silicon.
However, the path to a trillion-dollar footprint is proving volatile. Just this week, reports surfaced that OpenAI and Oracle have scrapped plans for a massive expansion at their Texas campus, a move that sent ripples through the supply chain. While Katti publicly maintained that Stargate remains "one of the largest AI data center campuses in the United States," the pivot suggests a tactical retreat from over-centralization. Instead of one monolithic "AI factory," Katti appears to be steering the company toward a more distributed model, likely to mitigate the immense strain on local power grids and to hedge against the rising costs of specialized cooling infrastructure.
The financial engineering behind this buildout is as complex as the physical engineering. Oracle recently launched a $50 billion financing plan to support its AI cloud ambitions, a move that draws intense scrutiny as the industry’s capital expenditure reaches unprecedented levels. For Katti, the challenge is balancing the academic pursuit of "scaling laws"—the theory that more data and more compute inevitably lead to smarter models—with the cold reality of a 70% surge in DRAM prices and a tightening market for high-end RF components. The "trillion-dollar" figure is not just a marketing slogan; it represents the cumulative cost of energy, land, and the specialized hardware required to train the next generation of frontier models.
U.S. President Trump’s administration has signaled a "drill and build" approach to energy, which may provide the regulatory tailwinds Katti needs to secure the massive power allocations required for Stargate. Yet, the reliance on a single hardware provider remains a strategic bottleneck. While Nvidia remains the primary partner, Katti’s background at Intel suggests he is acutely aware of the risks of vendor lock-in. There are growing indications that OpenAI is exploring custom silicon designs to reduce its multi-billion-dollar annual payments to Jensen Huang’s firm, a move that would transform OpenAI from a software lab into a vertically integrated hardware giant.
The stakes for Katti extend beyond the balance sheet. If Stargate succeeds, it will provide the compute foundation for Artificial General Intelligence, potentially centralizing the world’s most powerful technology within a handful of American data centers. If it fails, or if the returns on scale begin to diminish, it could trigger a capital exodus that would dwarf the dot-com bust. For now, the former academic is betting that the laws of physics will continue to reward those with the deepest pockets and the largest power plants.
Explore more exclusive insights at nextfin.ai.
