The AI election: Wall Street braces for midterms amid mounting opposition to data centers

The next Congress could shape policy for the AI buildout, putting an estimated $5 trillion in capital spending through 2030 at stake as public opposition to data centers grows. Wall Street is monitoring the election but is reassured by the prospect of a presidential veto of any bold legislation passed by Congress.

Analysis

The key mispricing risk is treating a congressional veto as a shield for AI infrastructure. It may constrain sweeping statutes, but it does little to neutralize state and local zoning, utility commission decisions, grid interconnection queues, or executive-agency implementation. Those channels can delay projects without a headline federal law—and time-to-power, not chip availability, may increasingly determine which planned data centers earn returns.

Exposure is uneven: a delayed site can hurt data-center developers and slow near-term orders for cooling and electrical equipment, while hyperscalers may redirect capital to less-constrained regions or lease capacity rather than abandon AI investment. Grid equipment suppliers could retain demand as projects shift, though utility ratepayer pushback and cost-recovery rulings are a material offset. The broad capex number should not be treated as uniformly at risk; distinguish announced plans from funded projects with secured power and permits.

Over days, election headlines can widen dispersion more than change aggregate AI demand. Over 1–3 months, watch local approvals, utility rate cases, interconnection timelines, and hyperscaler capex commentary. Over 6–18 months, policy-driven power costs and siting delays could redirect investment geographically and compress returns on capital. The contrarian gap: veto arithmetic may reassure investors about federal legislation while understating cumulative subnational friction. Thesis weakens if permitting and energization timelines improve, and strengthens if projects slip or utilities disclose rising costs that regulators will not allow them to recover.

Key Decisions for Investors

• Avoid a broad short of AI infrastructure solely on election risk. Prefer selective relative value: favor electrical-grid and power-management suppliers such as Eaton and Vertiv over data-center landlords with concentrated exposure to contested markets, while sizing for the possibility that project delays defer supplier orders too.
• Treat hyperscaler capex guidance as the near-term validation point: monitor Microsoft, Amazon, Alphabet, and Meta for evidence of project deferrals, rising power costs, or a shift from owned buildouts to leased capacity. A sustained reduction in planned capex would invalidate the view that this is mainly a siting and timing issue.
• Set an alert, rather than adding a directional position, for state and local permitting outcomes, utility rate-case rulings, and interconnection delays. Escalate the risk if multiple large projects are delayed or power-cost recovery is denied; de-escalate if approvals and energization timelines improve despite election uncertainty.