Can the US AI Boom Survive Data Center Backlash?

Data centers have become a political flashpoint in US election campaigns, with voters focused on rising power costs amid the rapid AI infrastructure buildout. The tech industry, via ITI CEO Jason Oxman, is facing growing calls to restrict new data-center projects, arguing continued investment is necessary to compete in AI with China. Market implications are likely more sentiment-driven for affected utilities and data-center REITs than a single immediate earnings shock.

Analysis

The immediate market reaction should be a multiple air-pocket for power-hungry infrastructure names, but the bigger mechanism is bargaining power: once voters frame data-center load as a local bill shock, developers lose pricing leverage and utilities gain leverage on interconnect terms, deposits, and timeline extensions. That is negative for colocation/REIT economics and any AI infra name whose thesis depends on fast time-to-power, while grid bottlenecks make switchgear, transformers, and EPC capacity more valuable.

Over the next 1-3 months, the key catalyst is not national policy but state and municipal behavior: permit friction, moratorium language, and utility-rate hearings. If those start showing up, the trade shifts from sentiment to earnings through delayed revenue recognition, higher project carry, and more expensive self-generation. That creates second-order winners in equipment vendors and behind-the-meter power solutions, while hyperscaler capex can still go up but with lower ROI and slower monetization.

The contrarian view is that this may be more of a headline risk than a demand shock. AI compute demand is still early, so even a hostile local backdrop may simply reroute projects to friendlier jurisdictions rather than kill spend; in that case, the selloff in the crowded data-center complex would be overdone. What would falsify the thesis is clean permitting data, no rise in utility connection deposits, and unchanged hyperscaler capex guides over the next quarter.

Key Decisions for Investors

• 1-3 month pair: short EQIX/DLR against long ETN or PWR. Expression: the local politics overhang should compress data-center landlord multiples faster than it hurts grid-equipment demand; cover if state PUC decisions stay benign or if data-center rent growth re-accelerates.
• Conditional hedge: buy a 3-6 month put spread on XLK or SMH only if we see actual state-level moratoriums or utility backlash broaden beyond rhetoric. Avoid an outright short unless interconnect queues and power costs start moving higher in the data.
• Watchlist long: XLU or a regulated utility with heavy data-center load growth only after confirmation that regulators allow rate-base expansion tied to incremental load. The upside is measured in allowed ROE and capex recovery, not immediate multiple expansion.
• Tactical long basket: ETN/PWR on any evidence of delayed time-to-power or higher utility deposit requirements. This is the cleaner second-order beneficiary if political scrutiny forces more spending into grid hardware and project execution.