Utah Senate President J. Stuart Adams sent a letter asking to reduce the proposed Stratos Project data center campus from 40,000 acres to about 10,000 acres and to require stronger commitments on water, conservation, environmental review, heat reduction, and public transparency, Business Insider reports. A spokesperson for O'Leary Digital told Business Insider the company "was caught off guard" and is "analyzing the letter carefully," and said Kevin O'Leary intends to respond personally before the end of the week. Kevin O'Leary told The Salt Lake Tribune he was "not walking away" from the project, Business Insider reports. State documents describe the campus as intended to support AI, cloud, and defense operations and project it would require 7.5 to 9 gigawatts, per Business Insider.
What happened
Utah Senate President J. Stuart Adams sent a letter calling for a 75% reduction in the proposed Stratos Project AI and defense data center campus, asking to cut the site from 40,000 acres to about 10,000 acres and to secure stronger commitments on water, conservation, environmental review, heat reduction, and public transparency, Business Insider reports. Adams also chairs the Military Installation Development Authority, which provided initial approval for the project in April, Business Insider reports. A spokesperson for O'Leary Digital told Business Insider "we have not engaged any Utah legislators on this. The letter caught us off guard," and said the company is "analyzing the letter carefully" and that Kevin O'Leary intends to respond personally before the end of the week, Business Insider reports. Kevin O'Leary told The Salt Lake Tribune he was "not walking away" from the project, Business Insider reports. State documents cited by Business Insider describe the campus as supporting artificial intelligence, cloud computing, and defense operations and estimate it would require 7.5 to 9 gigawatts if built at the current proposed scale.
Technical details
Editorial analysis - technical context: Large, greenfield data campus proposals at the scale described typically drive concentrated demands for electricity and cooling infrastructure, and often raise water-use and waste-heat questions for local utilities and regulators. Projects with projected loads in the single-digit gigawatt range create complex interdependencies across transmission upgrades, substation capacity, and onsite cooling strategies. For practitioners, those infrastructure dependencies translate into longer lead times, greater capital coordination with utilities, and amplified permitting risk.
Context and significance
Editorial analysis: The public letter from a state legislative leader requesting a 75% reduction is an example of growing municipal and state scrutiny on very large AI-adjacent campuses. Observers in the sector are increasingly factoring in environmental review, water allocation, and heat-management expectations when assessing the feasibility of large-scale compute sites. For AI infrastructure planners and operators, contested local approval processes can reshape siting decisions and expected timelines even when initial local authorities provide preliminary approvals.
What to watch
Observers following the project will watch for a formal written response from O'Leary, any revised permit applications or environmental assessments filed with Utah authorities, statements from the Military Installation Development Authority, and if utilities publish updated interconnection or load-impact studies reflecting the project's projected 7.5 to 9 gigawatts demand. Changes in any of these records would be the clearest indicators of whether the campus footprint or technical specifications are being renegotiated.
Scoring Rationale #
This story matters to infrastructure and operations teams because a potential 40,000-acre AI campus with 7.5 to 9 gigawatts demand would affect grid planning, permitting, and siting norms. Local political intervention on scale and environmental terms highlights material project risk for large compute deployments.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.