
When OpenAI stepped back from expanding its AI data center footprint in Abilene, Texas, Microsoft stepped right in.
The tech giant’s move in Abilene, Texas is more than a lease or a buildout as it adds two new AI data center buildings and a 900-megawatt power plant to a site already central to OpenAI and Oracle’s Stargate project, bringing the campus to 10 buildings and about 2.1 gigawatts of planned capacity. The deal shows how quickly the AI arms race is turning into a contest over land, power, and compute, and not just over software and model quality.
The Abilene campus has become one of the biggest AI infrastructure projects in the United States. Crusoe, the developer behind the site, has already completed two buildings for OpenAI and Oracle, and six more are still due by the end of this year. Microsoft’s new facilities will sit next to that work.
What Microsoft is Adding
Crusoe said Microsoft will build two “AI factory” buildings and an on-site power plant designed to generate 900 megawatts. That is a major addition to a campus that was originally planned for cryptocurrency mining before AI demand pushed the project in a new direction.
The scale matters because AI training and inference now depend on large amounts of electricity and specialized infrastructure, and the companies able to secure that capacity gain a practical advantage in the race to achieving AI supremacy.
Why OpenAI Stepped Back
OpenAI recently said it decided not to expand the Abilene site further, even though it described Stargate as one of the largest AI data center campuses in the country.
This was due to failed financing negotiations, changing infrastructure needs at OpenAI, and concerns regarding data center power stability. However, OpenAI is still developing other sites with Oracle, including a project in Wisconsin that is a part of the broader Stargate project.
What This Says About Cloud Competition
Microsoft was once OpenAI’s exclusive cloud provider and still holds about a 27% of stake in the company, but the two are now pursuing AI infrastructure on separate paths. That they are on different paths matters because it shows how the AI and cloud wars have escalated into a race to control the physical backbone of AI, from data centers, power plants, and grid access.
The firms that can lock in these physical infrastructures early are better placed to serve the next wave of model training and AI products.
As such, Microsoft’s Texas move reveals that the next phase of the cloud wars is being fought in megawatts and compute capacity.