How the US midterm elections reflected resistance to data centres

Voters rejected or moved to limit data-centre development in 40 of 41 ballot measures, while candidates campaigning for tighter AI and data-centre oversight won several prominent races. A Federal Reserve Bank of Dallas working paper estimated data centres raised wholesale electricity prices by 6.9% as of last year. Investors repriced regulatory risk: Nvidia lost $450bn in market value on Thursday, then gained 4% in early Friday trading while remaining down 6.8% over five days; CoreWeave was down 16% over five days after a Friday 6.6% rebound.

Analysis

The investable signal is permitting and power-delivery friction, not an immediate collapse in AI compute demand. Local opposition can lengthen approval timelines, raise community-benefit and infrastructure costs, and make power access less fungible across sites. That shifts bargaining leverage toward utilities and regions able to offer reliable power with credible cost allocation; it may also redirect projects rather than cancel them, benefiting less contentious locations. Utility upside is not automatic: regulators may resist passing data-centre upgrades through to household bills.

CRWV is more directly exposed than NVDA to delays at the site and power layer; NVDA’s exposure is second-order through customer deployment schedules and eventual utilization. But ballot outcomes do not themselves establish binding rules, and a national political shift does not guarantee enactment. Near term, expect headline-driven volatility; over 1–3 months, watch permitting pauses, oversight actions, and state-level rules. Over 6–18 months, the key structural question is whether power, water, and local consent become binding constraints on commissioned capacity. Voluntary operator pledges should not be treated as proof of lower costs or faster approvals.

Contrarian read: the market may overprice near-term demand destruction from political rhetoric while underpricing execution delays and higher all-in project costs. Thesis weakens if approvals continue, power contracts and commissioning schedules remain intact, and operators show that costs are not shifted to residents. Verify CRWV’s contracted-versus-planned capacity, power commitments, customer concentration, and cash needs before sizing exposure; the supplied information is insufficient to quantify earnings sensitivity.

Key Decisions for Investors

• Prefer a relative-value expression over a broad AI short: consider a small CRWV-underweight / NVDA-overweight pair, sized for volatility, because local permitting and power constraints are more direct for operators than for the chip supplier. Reassess after company updates on contracted capacity and commissioning timelines.
• Do not treat ballot results alone as a short trigger. Set a 1–3 month watch for binding state or federal rules, project moratoria, and utility rate decisions; escalate the negative view only if these translate into delayed approvals, reduced contracted capacity, or higher operator-funded infrastructure costs.
• For NVDA, avoid extrapolating a short-term political selloff into a demand downgrade absent evidence of customer capex cuts or weaker shipment guidance. Falsify this relative thesis if NVDA’s guidance or order indicators deteriorate materially while CRWV demonstrates resilient power access and on-time deployments.