Oracle is reportedly committing $40 billion toward Nvidia’s cutting-edge GB200 chips to support OpenAI’s sprawling new data centre in Abilene, Texas. This bold move is part of a broader effort to meet the exploding demand for AI computing infrastructure and marks one of the largest single chip purchases in history.
Inside the Stargate Project
The Texas facility is being described as the first U.S. “Stargate” data centre—an ambitious venture backed by OpenAI and SoftBank to deliver hyper-scale AI infrastructure. Once completed, the centre will generate a staggering 1.2 gigawatts of computing power, ranking it among the most powerful data centres on the planet.
Oracle will lease roughly 400,000 Nvidia GB200 chips and rent the computing capacity back to OpenAI. The project is being financed by Crusoe and Blue Owl Capital, with $15 billion raised via debt and equity. JPMorgan has provided nearly $10 billion in loans, while Oracle signed a 15-year lease to lock in its commitment.
The Broader Race for AI Compute Power
The Abilene project isn’t alone. Elon Musk is scaling up his “Colossus” data centre in Tennessee to support 1 million Nvidia chips. Amazon, too, is building a mega-campus in Virginia, pushing past 1GW in capacity. The race to build AI superclusters is heating up fast—and Oracle is now a key player.
OpenAI’s growing compute needs have driven it to diversify beyond its long-time dependency on Microsoft. While Microsoft’s investment has largely come in the form of Azure cloud credits, OpenAI is now seeking dedicated infrastructure, speed, and scale that existing providers have struggled to deliver.
Funding the AI Arms Race
To fuel this infrastructure boom, Stargate is raising up to $500 billion over four years. So far, OpenAI and SoftBank have each pledged $18 billion, while Oracle and Abu Dhabi’s MGX fund have committed $7 billion apiece. The equity-holding group aims to launch a series of facilities across the U.S. and abroad.
International expansion is already underway. During a recent visit by President Trump to the Gulf, OpenAI announced a new Stargate data centre in Abu Dhabi. That site will span 10 square miles and offer a massive 5GW of capacity—enough to run over 2 million Nvidia GB200 chips. UAE-based AI firm G42 is leading construction.
Breaking Away from Microsoft’s Shadow
The shift away from Microsoft signals a changing dynamic in AI infrastructure. Earlier this year, OpenAI and Microsoft agreed to end their exclusive partnership, leaving licensing details unresolved. But what’s clear is that OpenAI is betting on vertical integration: build the infrastructure, own the chips, and control the destiny of its models.
With Oracle stepping in as a heavyweight partner, and other tech giants scrambling to build out their own compute pipelines, the future of AI infrastructure will be defined by who can build faster, scale smarter, and spend bigger.
The chip wars aren’t theoretical—they’re breaking ground in Texas.
