Zeo & Creekstone: How to Power Delta Gigasite AI Data Centre
In a quiet corner of Utah's Black Rock Desert, there is something historic taking shape.
Millard County, which can be found on the westernmost edge of the Beehive State, is a flat, remote and sparsely populated area.
While these attributes do not lend themselves well to most industries, they are ideal for housing the infrastructure that powers AI.
Creekstone Energy is building what it calls a "gigasite" in Millard County – a next-generation power and data infrastructure campus designed specifically for AI workloads.
The site first broke ground in December 2025 and, by the first half of 2027, Creekstone plans to have more than 300 MW of gas-powered capacity online. This is only the tip of the iceberg.
When construction is finished, this site is expected to be the world's largest AI-optimised data centre campus, boasting an eye-watering power capacity of 10 GW.
To put it in context, this would be enough electricity to power 15 million homes in Great Britain, according to the Low Carbon Contracts Company.
Not leaving renewable energy opportunity behind, On 18 February, Creekstone announced it had signed a memorandum of understanding (MoU) with Zeo Energy, a Florida-based clean energy company, to develop around 280 MW of solar and storage capacity for the site.
When it acquired energy storage firm Heliogen in August 2024, Zeo set itself up to commercialise its offerings at a far larger scale.
With the data centre sector booming, demand for round-the-clock power to support AI operations has increased, with Zeo rising to the challenge.
The kind of energy storage systems Zeo now has at its disposal are vital technologies for AI infrastructure, simply because they allow for the uninterrupted flow of renewable energy, even when generation is intermittent.
Meeting AI's power demands
Ray Conley, CEO of Creekstone Energy, acknowledges the massive pressure AI workloads exert on infrastructure providers.
"AI workloads are driving unprecedented demand for power," he explains.
"At Creekstone, we plan to deliver more than 600 MW of baseload power to our Gigasite customers in 2027 in Phase 1 of our project. Our collaboration with Zeo reflects the market urgency of using all available energy sources to rapidly provide baseload power.
"With solar power and Zeo's long-duration energy storage solution, we plan to significantly expand the amount of clean power we offer our hyperscalers and artificial intelligence data centre customers."
The Gigasite's first confirmed tenant is Blue Sky AI, which has been allocated 50 MW of energy. As things stand, the firm is targeting multiple gigawatts of total capacity, with Zeo's clean energy solutions expected to form a key part of supporting AI operations.
Validating clean power solutions
Under the terms of the MOU, the company has already begun a pre-feasibility study to determine the optimal solar and storage configuration for the site, with its engineering team drawing on Heliogen's thermal and chemical storage know-how.
Tim Bridgewater, CEO of Zeo says: "Since our acquisition of Heliogen, we have been actively seeking to apply our long-duration storage expertise to the unprecedented power demand in the data centre space.
"Our MoU with Creekstone is a milestone in this effort, and we are in discussions with several other projects that we believe can benefit from our clean baseload power solutions.
"The Creekstone collaboration is an opportunity to validate the application of our expertise in renewable power generation and long-duration storage to increase power delivery for data centre customers in a cost-effective, low-emissions manner. We expect our ability to access the public capital markets to provide project financing could give us a competitive edge in our business development efforts."
The MoU also suggests that Zeo could be providing engineering services for the project, as well as potentially arranging financing for the development.
What is clear is that the exponential growth of AI is creating unprecedented infrastructure requirements.
As AI models become more sophisticated and their applications more widespread, the demand for dedicated, reliable power continues to surge. Projects like Creekstone's Gigasite could represent the future of AI infrastructure – purpose-built facilities designed from the ground up to support the computational demands of artificial intelligence whilst incorporating sustainable energy solutions.
As AI continues its rapid evolution, the infrastructure to support it must evolve just as quickly.
