1. Home
  2. / Science and Technology
  3. / 9-gigawatt AI megacampus approved in Utah will occupy 40,000 acres, consume more than twice the energy of the entire state, and operate off-grid with natural gas, promising to generate up to $250 million per year and 2,000 jobs amid the global race for data centers.
Reading time 5 min of reading Comments 0 comments

9-gigawatt AI megacampus approved in Utah will occupy 40,000 acres, consume more than twice the energy of the entire state, and operate off-grid with natural gas, promising to generate up to $250 million per year and 2,000 jobs amid the global race for data centers.

Written by Carla Teles
Published on 30/04/2026 at 16:04
Updated on 30/04/2026 at 16:05
Be the first to react!
React to this article

Natural gas places Utah at the center of the global race for AI data centers with a 9-gigawatt campus, 40 thousand acres, self-generated energy, and potential to move hundreds of millions of dollars per year

Natural gas will be the energy base for one of the most ambitious projects ever approved for artificial intelligence infrastructure in the United States. The Utah Military Installation Development Authority, MIDA, approved a development agreement for the Stratos campus, a mega data center complex in Box Elder County that could reach 9 gigawatts of capacity and operate entirely off the traditional power grid.

The project involves O’Leary Digital, the infrastructure arm of investor Kevin O’Leary, and draws attention for its size, projected consumption, and energy model. At a time of global race for computational capacity for AI, the venture was designed to produce all the energy on-site, through a connection with the Ruby Pipeline, and could consume more than double the average electricity currently used in the entire state of Utah, estimated at about 4 gigawatts.

What is the Stratos project and why it draws so much attention

Natural gas powers artificial intelligence data centers in Utah and raises the energy debate.
 Image: Getty

Stratos is a hyperscale data center campus planned for Box Elder County, Utah. The proposal was approved by MIDA on Friday and envisions a gigantic structure aimed at the new generation of digital infrastructure required by the expansion of artificial intelligence.

The impact draws attention because the complex will occupy 40 thousand acres of private land, in addition to 1,200 acres of military and state properties. It is not a conventional data center, but a campus designed to operate on an extreme scale, with self-generation and energy consumption sufficient to place it among the largest projects in the sector.

The numbers that explain the dimension of the AI megacampus

The numbers of the venture help to understand why the project gained immediate prominence. When fully completed, Stratos could reach 9 gigawatts of capacity. This volume far exceeds the current average electricity consumption of Utah, approximately 4 gigawatts.

The initial phase alone foresees a generation capacity of about 3 gigawatts. In the final scenario, the campus would position itself among the largest projects linked to the era of artificial intelligence, standing between SoftBank’s 10-gigawatt plan in Ohio and Meta’s 7-gigawatt structure in Louisiana.

How natural gas will sustain the off-grid operation

The central element of the project is precisely natural gas. All the campus’s energy will be produced on-site from a connection with the Ruby Pipeline, a 1,094-kilometer interstate line that crosses northern Utah on its route from Wyoming to Oregon.

According to MIDA’s executive director, Paul Morris, the facility should not consume “a single electron” from the existing power grid. The proposal also foresees that, in the future, the excess energy could be returned to the grid. In practice, this transforms the venture into a clear example of parallel infrastructure, with self-generation to avoid waiting for new connections with utilities.

Why AI data centers are migrating to self-generated energy

The strategy adopted in Stratos is not isolated. The project joins a growing list of AI-era data centers that are building their own power generation instead of waiting years for access to the conventional grid.

This movement reveals how the computational demand of new AI centers is pushing the limits of traditional electrical infrastructure. In Utah’s case, the use of natural gas appears as a solution to ensure scale, speed of implementation, and energy autonomy in a sector that urgently competes for capacity.

What this project can generate in revenue and jobs

Even with tax incentives granted to attract hyperscale operators, the revenue expectation is high. MIDA projects $30 million annually for Box Elder County during the initial phase of the campus and more than $100 million per year when the structure reaches its full capacity.

At the state level, the forecast is even higher. The authority estimates that the sales tax revenue generated solely by the data centers will reach $250 million per year. Additionally, the expectation is to create 2,000 permanent jobs in the county after the completion of the work.

Tax incentives and competition for hyperscale operators

To make the project more competitive, MIDA reduced the energy use tax for the enterprise from 6% to 0.5%. It also agreed to reimburse 80% of the property tax revenue generated by the development to O’Leary Digital.

These measures show the extent of the competition for large cloud and AI infrastructure operators. So far, no hyperscale tenant has been publicly announced, but the market is dominated by names like Amazon, Microsoft, and Google, with Meta and Apple close behind.

What the global AI race has to do with Utah

During the MIDA board meeting, Kevin O’Leary defended the project as a response to China’s advancement in energy infrastructure linked to artificial intelligence. According to him, the Asian country built 400 gigawatts of new energy in the last 24 months, and a large part would be linked to supplying AI data centers.

This comparison helps explain why projects like Utah’s have ceased to be merely regional ventures and have come to be treated as pieces of an international dispute for computing power, energy, and industrial capacity.

The next steps for the approved megacampus in Utah

Although the development agreement has been approved by MIDA, the process is not yet fully completed. The county commission postponed the final vote that would occur on Friday to this week, and at the time of the publication of the base text, no hyperscale operator had been officially confirmed.

This means that the project has already made significant progress but still depends on relevant definitions to be fully realized. O’Leary Digital is also developing a complementary campus in Wonder Valley, Alberta, Canada, announced for 2024, whose construction has not yet begun.

Why this project could become a reference for the new generation of infrastructure

The Stratos campus brings together almost all the elements that today define the new global race for data centers. It has extreme scale, self-generation, use of natural gas, promise of billion-dollar revenue over time, and the capacity to create permanent jobs in a specific region.

More than a local project, the approved megacampus in Utah symbolizes the shift in the level of infrastructure required by artificial intelligence. When a single complex can consume more than twice the energy of an entire state, the debate ceases to be merely technological and involves energy, territory, revenue, and economic strategy.

Do you think giant projects powered by natural gas will be the inevitable path to support the expansion of AI data centers?

Sign up
Notify of
guest
0 Comments
most recent
older Most voted
Built-in feedback
View all comments
Carla Teles

Produzo conteúdos diários sobre economia, curiosidades, setor automotivo, tecnologia, inovação, construção e setor de petróleo e gás, com foco no que realmente importa para o mercado brasileiro. Aqui, você encontra oportunidades de trabalho atualizadas e as principais movimentações da indústria. Tem uma sugestão de pauta ou quer divulgar sua vaga? Fale comigo: carlatdl016@gmail.com

Share in apps
0
I'd love to hear your opinion, please comment.x