Trump is giving data centers a pass to pollute

Former EPA officials warned that the Trump administration's rollback of dozens of environmental rules to accelerate AI data-center construction could increase health risks for Americans. EPA Administrator Lee Zeldin has framed deregulation as part of a strategy to make the U.S. the global AI leader, while former officials are urging a voluntary Data Center Health Protection Pledge. The policy shift could reduce permitting and compliance burdens for AI infrastructure developers but raises regulatory, environmental, and community-risk concerns.

Analysis

The investable implication is not the environmental headline itself but a lower permitting and compliance burden for power-intensive AI infrastructure. Over the next 6-18 months, this potentially increases the addressable pipeline for hyperscaler-linked data-center developers and utilities with dispatchable generation, while widening the valuation gap versus developers reliant on slower grid interconnection and local approvals. EQIX and DLR benefit only if incremental capacity can be energized; the more direct bottleneck beneficiaries remain power suppliers such as VST, CEG and NRG, plus gas-turbine supply chain exposure through GEV.

Near term, deregulation may paradoxically increase project delay risk rather than eliminate it. Reduced federal safeguards can shift opposition to state regulators, local zoning boards and litigation, particularly in constrained power markets such as Virginia/PJM, Texas and the Midwest; this raises the value of sites with contracted power, water access and already-completed permits. Watch PJM capacity pricing, utility interconnection queues and hyperscaler capex commentary over the next 1-3 months: these are more consequential to earnings than federal policy statements.

Consensus may overstate the benefit to AI beneficiaries broadly. Lower environmental standards do not create transformers, turbines, transmission lines or firm generation, and accelerated construction can tighten equipment and power markets further. That favors incumbent owners of energized assets and suppliers with backlog pricing power, but can compress returns for speculative data-center developers whose land banks lack secured electricity. The thesis is falsified if hyperscaler capex moderates, PJM/ERCOT power prices retreat materially, or state-level permitting injunctions halt major campus projects.

Key Decisions for Investors

• Maintain a 6-12 month long bias in VST and CEG versus a broad AI-infrastructure basket: contracted or merchant power scarcity is the binding constraint, with upside if data-center load forecasts lift forward capacity prices. Reduce if PJM capacity-auction pricing or 2027-29 load projections disappoint.
• Pair long GEV / short speculative data-center development exposure via a short position in SRVR ETF only after confirming its constituent sensitivity: turbine and grid-equipment backlog has better pricing power than unpowered land-bank economics. Target 10-15% relative return over 6-12 months; stop on evidence of gas-turbine order cancellations.
• Do not chase EQIX or DLR solely on this policy signal. Add only following disclosure of incremental megawatts under signed power contracts or a pullback that offers a clear premium-to-cost-of-capital spread; regulatory rhetoric alone does not improve energized-capacity availability.
• Set alerts for Virginia and PJM state/local permitting actions, EPA litigation outcomes, and utility large-load tariff filings over the next 90 days. A coordinated state response imposing water, emissions or grid-upgrade charges would shift economics away from merchant data-center expansion and toward regulated utilities able to rate-base upgrades.