Kevin O'Leary Proposes $115 Billion in AI Data Centers for Utah and Alberta

Investor and television personality Kevin O'Leary has proposed the Stratos Project in Box Elder County, Utah, spanning 40,000 acres and designed to generate up to 9 gigawatts of power. According to a Wall Street Journal report mentioned in Benzinga, this capacity would equal more than 20% of all data-center capacity currently operating in the U.S. To power the site without straining the national grid, the plan involves burning natural gas on-site using the Ruby pipeline.

The first phase of the Utah project targets 3 gigawatts at a cost of about $45 billion. O'Leary stated that the motivation includes competition with China on economic and military superiority, framing it as both geopolitical and financial. The project could more than double the 4 gigawatts of power currently used by the entire state of Utah.

In Alberta, Canada, O'Leary has proposed a $70 billion AI data center campus with 7.5 gigawatts of capacity. The project is exempt from provincial environmental impact assessment but requires local permits, including approval from the First Nation. O'Leary told CBC News that obtaining the permit would trigger financing, engineering, and design activities.

He described Alberta as a suitable location due to its natural gas resources, water availability, and cold temperatures for cooling AI clusters. The combined investment for both projects totals $115 billion, focused on physical infrastructure for AI technology.

the project has encountered local resistance.

Benzinga reported that during a recent county commission meeting, more than 80 residents attended with signs reading “People before profits,” leading officials to delay a vote. Concerns include water usage in a state facing the disappearance of the Great Salt Lake.

Broader industry challenges include a capital spending gap, as hardware suppliers have increased investments by only half compared to cloud giants, according to The Economist cited in Benzinga. This mismatch could limit equipment availability even if projects receive approval.