OpenAI’s Stargate AI project is moving toward a much larger infrastructure footprint through Oracle. According to Bloomberg, OpenAI is set to rent an additional 4.5 gigawatts of computing power from Oracle's US data centers, tying the project to a major expansion of data center capacity.
A larger Oracle role in Stargate
The planned arrangement centers on Oracle's US data centers and the computing power needed for Stargate. The additional 4.5 gigawatts is not a small adjustment. The source describes that amount of energy as roughly equivalent to the electricity use of several million households.
That comparison matters because it puts the scale of AI infrastructure into everyday terms. Instead of treating data centers as abstract technical facilities, the figure shows how much physical capacity a project like Stargate can require when it moves beyond software and into large-scale deployment.
The deal is also part of a broader cloud contract with Oracle valued at $30 billion per year. That figure places the arrangement in the category of major long-term infrastructure commitments rather than a short-term capacity purchase.
Where Oracle plans to build
To meet the demand, Oracle plans to build new data centers across multiple states. The states named in the source are Texas, Michigan, and Wyoming.
The geographic spread is important because the Stargate project is not described as relying on one site alone. A multi-state buildout suggests that Oracle is preparing to supply large amounts of computing power through a broader data center network.
The existing Stargate facility in Abilene, Texas, is also expected to grow. According to the source, that facility is expected to expand from 1.2 to 2 gigawatts.
That expected expansion shows that the current Stargate footprint is not static. Even before counting the additional 4.5 gigawatts from Oracle's US data centers, the Abilene, Texas, site is set for a significant increase in capacity.
Why gigawatts are central to the story
AI projects are often discussed in terms of models, products, and cloud services. This report shifts attention to the infrastructure layer underneath those systems: data centers, computing power, and the energy needed to operate them.
The phrase 4.5 gigawatts of computing power captures both the technical and physical sides of the Stargate project. It points to the equipment needed to run large AI workloads, but it also points to the electricity demand attached to that equipment.
For readers, the household comparison is the clearest way to understand the scale. If the additional energy is roughly equivalent to the electricity use of several million households, then the Stargate buildout sits far beyond ordinary enterprise cloud expansion.
That does not by itself explain every operational detail. The source does not specify construction timelines, exact site counts, or how the capacity will be divided among Texas, Michigan, and Wyoming. What it does establish is the size of the commitment and the role Oracle is expected to play in supplying it.
What the $30 billion per year contract signals
The broader cloud contract with Oracle is valued at $30 billion per year. In practical terms, that number signals that OpenAI’s Stargate project is tied to a large and continuing relationship with Oracle rather than a narrow facility-by-facility rental.
Cloud contracts at this scale are not only about servers. They connect computing power, data center planning, energy requirements, and operational capacity into one business arrangement. In this case, the source links the additional 4.5 gigawatts directly to Oracle's US data centers and the Stargate AI project.
For Oracle, the plans described in the source point to new data centers across multiple states and an expanded Abilene, Texas, facility. For OpenAI, they point to access to additional computing power for Stargate through Oracle.
The essential takeaway is straightforward: Stargate is becoming a major data center power story as much as an AI story. The reported 4.5 gigawatts, the Abilene expansion from 1.2 to 2 gigawatts, and the $30 billion per year Oracle cloud contract all show how much infrastructure is being positioned behind the project.