Voters mostly don’t like AI and data centers, but neither party seems to have an edge

A New York Times/Siena University poll of 1,503 likely voters found 61% oppose construction of data centers supporting AI technology, while only 14% strongly support them. The opposition signals increasing political and permitting risk for AI infrastructure buildouts, though views were more evenly divided among 2024 Trump voters, with 49% supporting and 45% opposing data centers.

Analysis

The investable issue is not aggregate AI demand but the conversion rate of announced capacity into energized load. Local opposition raises permitting duration, interconnection costs, and mitigation requirements, favoring operators with existing generation, transmission rights, brownfield sites, or contracted power over pure data-center equipment vendors whose valuations assume uninterrupted construction schedules. The likely near-term pressure point is project timing rather than cancellation: a 6-12 month commissioning slip can defer revenue recognition for VRT, ETN, PWR and construction-exposed private peers while leaving hyperscaler capex intentions intact.

Power scarcity becomes more valuable if new greenfield campuses face greater friction. CEG, VST and NRG can benefit where they can contract existing dispatchable capacity or structured long-term supply, while regulated utilities with large planned rate-base programs face a more mixed outcome: higher capex opportunity but elevated political risk around residential rate impacts. This also increases the value of smaller, politically feasible deployments—retrofit capacity, colocated generation, and regions with surplus power—relative to headline megacampus announcements.

Consensus appears too focused on whether AI capex survives and insufficiently focused on its geographic redistribution. A permitting backlash could shift spend toward states and municipalities offering expedited approvals, producing uneven utility load growth and a widening gap between announced and operational data-center capacity over the next 6-18 months. The thesis is falsified if hyperscalers demonstrate that opposition is not affecting permit issuance, interconnection queues, or commercial-operation dates in upcoming earnings and utility filings.

Key Decisions for Investors

• Avoid treating NYT as a read-through; the survey has no direct earnings mechanism for the publisher and is not independently sufficient for a position.
• Initiate a 1-3 month relative-value watch: long CEG or VST versus short VRT, sized modestly, only if upcoming company disclosures show delayed data-center energization or reduced equipment delivery schedules. The trade captures rising value of existing power versus deferred incremental build; exit if VRT backlog conversion and hyperscaler commissioning dates remain on schedule.
• Favor selective exposure to existing-power and transmission-constrained markets through CEG, VST and PWR rather than broad data-center infrastructure beta. Add only after verifying contracted load, power-price escalation terms, and interconnection status; a decline in forward power prices or weaker contracted-load disclosure would invalidate the thesis.
• Monitor quarterly utility commission dockets and local permitting outcomes over the next 3-6 months. Escalate the short equipment/buildout basket only if approval timelines lengthen materially or utilities begin citing customer-bill protection as a reason to defer incremental data-center load.