Shark Tank’s Kevin O’Leary wants to build a massive $100 billion data center in rural Utah. Residents are revolting

A $100 billion, 9-gigawatt Utah data center project backed by Kevin O’Leary faces organized local backlash, with residents pushing a referendum after county commissioners unanimously advanced the plan. The facility would occupy 40,000 acres and, according to reports, consume more electricity than the entire state of Utah does in a year, while residents cite climate and water risks tied to the Great Salt Lake. O’Leary argues the project supports AI capacity, local jobs, and national security, but the controversy reflects widening resistance to large-scale data center development.

Analysis

The real market signal here is not the local politics; it is the widening gap between headline AI demand and the physical bottlenecks required to monetize it. Projects of this scale force the market to reprice power procurement, gas midstream, and grid-adjacent infrastructure because the first derivative of AI capex is no longer chips but electrons, permits, and cooling water. That shifts bargaining power away from software beneficiaries and toward owners of dispatchable generation, pipeline capacity, and regulated utility assets that can sell “speed to power.”

The second-order risk is that opposition moves from nuisance to financing friction. Even if the project ultimately proceeds, referendum efforts, litigation, and media scrutiny can add 6-18 months to execution, which matters because valuation for AI infrastructure is heavily back-end loaded. Any delay raises the probability of developer markdowns, higher cost of capital, and a rerating lower for private-data-center platforms with aggressive land banking or speculative power commitments.

This is also a potential negative for ORCL only at the margin: the issue is not demand for compute, but whether hyperscale and colo customers start demanding stronger contractual protections around land, water, and energy pass-throughs. The overbought consensus is that all AI infrastructure is a pure beneficiary; in practice, projects with the weakest permitting moat and highest local externalities will clear at a discount, while incumbents with existing power interconnects should capture share. The contrarian view is that public backlash may actually accelerate consolidation, because only the largest operators and utilities can navigate community opposition and regulatory complexity at scale.

Key Decisions for Investors

• Long XLU vs short a basket of speculative AI-infra developers for 3-6 months; the trade captures rising 'speed-to-power' scarcity while limiting exposure to permitting delays and community backlash.
• Buy KMI or TRP on any pullback over the next 1-2 weeks as a second-order beneficiary of natural-gas-fired load growth; asymmetric upside if data-center developers keep choosing behind-the-meter gas solutions.
• Maintain a tactical underweight in ORCL for 1-3 months until the market can distinguish contract-driven cloud demand from discretionary AI infrastructure enthusiasm; limited direct downside, but sentiment spillover can cap multiples.
• Consider a long-duration call spread on utility names with large service territories and grid interconnect optionality, paired against shorting unproven data-center REIT concepts; risk/reward favors regulated cash flows over execution risk.
• For event risk, buy cheap put spreads on data-center adjacent local landowners or small-cap infrastructure names with concentrated project exposure; a referendum or lawsuit headline can hit these names 15-25% in days.