Utah Approves 9GW AI Data Center Campus as Infrastructure Race Accelerates
Key Takeaways
- ▸Utah approved a 9 GW AI data center campus consuming more than 2x the state's current electricity supply
- ▸O'Leary Digital built the facility off-grid using natural gas generation to bypass years-long utility interconnection delays
- ▸Project reflects a growing industry trend of hyperscalers building dedicated power infrastructure as AI demand outpaces utility capacity
Summary
Utah's Military Installation Development Authority approved a development agreement for Stratos, a hyperscale AI data center campus in Box Elder County being developed by O'Leary Digital, the infrastructure division of Shark Tank investor Kevin O'Leary. The project would eventually reach 9 GW of power capacity—more than double Utah's current average electricity consumption—spanning 40,000 acres of private land plus 1,200 acres of state and military property.
The facility will generate all its power on-site via a connection to the Ruby Pipeline natural gas line, avoiding the multi-year utility interconnection delays that have become a critical bottleneck for major data center projects. Phase 1 calls for 3 GW, with full buildout reaching 9 GW. To attract major hyperscale cloud operators, Utah authorities reduced the energy tax from 6% to 0.5% and agreed to rebate 80% of property tax revenue. Economic projections include $30 million annually in initial phases and over $100 million at full capacity, along with 2,000 permanent jobs.
Stratos joins a growing wave of AI-era data center projects where hyperscalers build their own power infrastructure rather than waiting for utility capacity. SoftBank is developing a 10 GW campus in Ohio, while Meta recently committed to funding seven natural gas plants for a 7 GW Louisiana facility. Kevin O'Leary framed the urgency in geopolitical terms, noting that China built 400 GW of new power in the past 24 months largely for AI data centers. No primary tenant has been publicly named, though Amazon, Microsoft, and Google—the three largest cloud operators—are considered likely candidates, along with Meta and Apple.
- Expected to generate $30M–$100M+ annually for Box Elder County and create 2,000 permanent jobs; also driven by competitive pressure from China's aggressive AI infrastructure expansion
Editorial Opinion
The race to build AI-capable infrastructure is intensifying as hyperscalers recognize that data center capacity, not just model innovation, is the bottleneck to scaling AI services. While O'Leary Digital's Stratos campus represents a significant infrastructure commitment, the sheer scale (9 GW—more than double Utah's current electricity use) underscores the energy-intensive nature of modern AI and raises serious questions about environmental sustainability. The competitive pressure from China's aggressive infrastructure expansion appears to be driving U.S. investment, but whether this infrastructure race ultimately serves innovation or merely inflates capital requirements remains an open question.
