The Utah AI data center approved by the state’s Military Installation Development Authority on April 24, 2026, represents the largest infrastructure project in the agency’s history and a watershed moment for how the U.S. builds AI computational capacity. Kevin O’Leary’s O’Leary Digital and partner West GenCo secured permission to develop a 9-gigawatt campus across 40,000 acres of privately owned land plus 1,200 acres of military and state property in Box Elder County. To put that in perspective: one gigawatt powers roughly 1.8 million people annually, meaning this single facility will generate and consume more electricity than Utah’s entire state population currently uses.
Key Takeaways
- Utah AI data center approved April 24, 2026, by state’s Military Installation Development Authority as its most ambitious project.
- 9-gigawatt total capacity (Phase 1: 3 gigawatts) will exceed entire state power consumption by roughly 2x.
- On-site power generation targets 100% energy self-sufficiency using natural gas infrastructure access.
- Project marketed as Wonder Valley; supports Pentagon, Department of Defense, and U.S. Air Force national security objectives.
- Twin Wonder Valley campus in Alberta, Canada, targets additional 6 gigawatts, bringing combined AI capacity to 15 gigawatts.
Why This Utah AI Data Center Matters Now
The approval arrives amid explosive global demand for AI infrastructure. Data centers currently consume roughly 4% of U.S. electricity, a figure that will climb sharply as artificial intelligence workloads expand across government and commercial sectors. MIDA Executive Director Paul Morris called the project a national security imperative, signaling that the Pentagon views AI computational capacity as critical defense infrastructure. The facility will support U.S. Air Force operations, the Utah National Guard, and Department of Defense initiatives—not just commercial AI vendors. This framing matters: the project isn’t positioned as a profit center but as essential national capability.
The timing also reflects frustration with traditional data center siting. Existing facilities face community opposition over power consumption and water use. O’Leary and MIDA pushed back on these criticisms, claiming the Stratos Project (the formal development name) will generate 100% of its power on-site and use less water than the ranching operations currently occupying the land. Whether those claims hold up under construction scrutiny remains open—but the framing signals a shift toward data center projects that promise energy independence rather than relying on regional grids.
The Utah AI Data Center’s Scale and Architecture
The facility spans roughly 41,200 acres total in unincorporated Box Elder County, with hyperscale data center infrastructure, advanced manufacturing, housing, and commercial facilities planned. Phase 1 targets approximately 3 gigawatts of power generation and capacity, with the full 9-gigawatt vision rolling out over time. Access to major interstate natural gas infrastructure supports the on-site generation strategy, though specific details on energy technology—whether natural gas turbines, combined-cycle plants, or hybrid approaches—remain undisclosed. West GenCo will handle permitting and regulatory coordination, a critical role given the project’s scale and interstate implications.
This architecture differs sharply from traditional data center development, which typically relies on purchasing power from regional grids or signing long-term contracts with utilities. The Utah AI data center’s self-generation model eliminates grid dependency but requires massive upfront capital investment in power plants. It also concentrates infrastructure risk: if on-site generation fails, the entire facility loses power simultaneously rather than experiencing partial outages. The trade-off appears intentional—national security objectives demand reliability that grid-dependent facilities cannot guarantee.
Comparison to the Alberta Wonder Valley Campus
The Utah project is not standalone. A parallel Wonder Valley campus in Alberta, Canada, targets an additional 6 gigawatts of AI-ready power capacity, with a reported development cost of approximately 70 billion dollars. Together, the two Wonder Valley facilities aim to deliver 15 gigawatts of integrated AI infrastructure spanning North America. This binational approach suggests a coordinated strategy to build AI capacity outside traditional data center clusters in California and Virginia, potentially offering lower power costs and regulatory advantages in less congested regions.
The Alberta project’s scale and pricing context hint at the capital requirements: if 6 gigawatts in Canada costs roughly 70 billion dollars, the 9-gigawatt Utah facility likely demands comparable or greater investment. Neither O’Leary nor MIDA has disclosed total project cost for the Utah campus, but the infrastructure footprint and power generation requirements suggest multi-billion-dollar expenditure. Comparing the two projects reveals a broader strategy: build massive, self-powered AI capacity in underutilized regions with political support, rather than competing for grid capacity in saturated tech hubs.
What Critics and Skeptics Should Watch
The project’s claims deserve scrutiny. MIDA’s assertion that the facility will use less water than current ranching operations is difficult to verify without detailed water-use modeling—data centers typically require significant cooling water, and claims of reduced consumption compared to agriculture warrant independent auditing. Similarly, the 100% on-site power generation promise depends entirely on whether West GenCo can deliver the promised infrastructure on schedule and budget. Delays in permitting, construction, or power plant deployment could force the project to rely on grid power initially, undermining the national security independence argument.
Labor and housing claims also merit monitoring. The project includes residential and commercial facilities, implying job creation and workforce development. But how many permanent jobs will the facility actually generate once operational? Data centers are notoriously capital-intensive and labor-light—a 9-gigawatt campus might employ hundreds, not thousands. If housing and commercial development outpace actual job creation, the project risks becoming a speculative land play rather than a genuine economic engine for Box Elder County.
Is the Utah AI data center a national security asset or a speculative venture?
Both. The Pentagon’s involvement and MIDA’s framing as critical defense infrastructure are genuine—AI computational capacity is legitimately important for military operations and intelligence work. But Kevin O’Leary’s involvement and the project’s scale also suggest significant profit potential. The two motivations are not mutually exclusive, but they do create incentives for aggressive claims about power self-sufficiency, water use, and economic benefit.
When will the Utah AI data center become operational?
Phase 1, targeting approximately 3 gigawatts, is the near-term focus following the April 24, 2026 approval, but specific operational timelines have not been disclosed. Full build-out to 9 gigawatts will take years, likely stretching into the early 2030s depending on permitting, construction, and power generation deployment.
How does the Utah AI data center compare to existing U.S. data center capacity?
The 9-gigawatt facility will be one of the largest concentrated data center campuses in North America once completed. For context, it exceeds the total power consumption of Utah’s current population, making it a transformative infrastructure project rather than an incremental addition to existing capacity.
The Utah AI data center represents a genuine shift in how the U.S. approaches critical computational infrastructure. Rather than building data centers within existing power grids, the Stratos Project attempts to create self-contained energy ecosystems that can operate independently and serve national security objectives. Whether the execution matches the ambition remains to be seen—but the approval signals that regulators and policymakers now view massive AI capacity as essential national infrastructure worthy of extraordinary resources and expedited permitting. That framing alone reshapes the competitive landscape for data center development across North America.
Edited by the All Things Geek team.
Source: Tom's Hardware
