Khanna to introduce 'Data Center Bill of Rights' as fight over AI power plants heats up

Rep. Ro Khanna introduced a “data center bill of rights” resolution aimed at slowing AI data center expansion via community veto power and siting limits (e.g., barring facilities within 2,500 feet of homes/schools/health sites) while protecting against higher electricity bills and water depletion. The proposal adds to a growing wave of state moratoriums and related legislation as voters resist the AI boom’s heavy energy and water demands, especially ahead of the midterms. Overall, the news increases near-term regulatory and permitting risk for hyperscalers, with potential electricity- and water-cost pressures feeding back into operational planning.

Analysis

This is less a demand-shock story than a permitting-shock story: the bottleneck shifts from “can AI capacity be financed?” to “can it be sited and energized fast enough?” That tends to compress multiples for publicly traded infrastructure names that rely on smooth entitlement cycles, especially data-center REITs and adjacent power-buildout names, while leaving the biggest hyperscalers relatively insulated because they can pre-buy land, power, and political influence.

The second-order winner is existing capacity. If local approval becomes harder, scarce, already-permitted campuses should command higher lease spreads and utilization, which helps incumbents versus new entrants and smaller colocation operators. In power markets, the near-term loser is not generators broadly but the incremental load-growth narrative embedded in utility and grid-equipment capex; any delay in new data center starts pushes out revenue recognition for transformers, switchgear, and interconnect vendors.

Over 1-3 months, the key catalyst is whether this stays a rhetoric trade or becomes state-level permitting friction that shows up in delayed announcements and slower absorption. Over 6-18 months, the more durable effect is geographic concentration: data-center buildout migrates to friendlier jurisdictions, which raises land, transmission, and tax costs but can actually support pricing power for the best-located assets.

Consensus may be underestimating how little this matters to the largest platforms and overestimating the hit to the whole AI stack. The market reaction could be overdone if investors sell the entire AI infrastructure basket instead of separating scarce-capacity beneficiaries from new-build losers; the thesis is falsified if states fail to pass follow-on restrictions or if leasing/rent metrics at incumbents keep accelerating despite the headline noise.

Key Decisions for Investors

• Short SRVR or a similar data-center REIT basket on regulatory headline risk; best entry is into any 1-2 day relief rally, with a 1-3 month thesis that entitlement delays pressure forward growth multiples. Cover if permitting language fails to spread beyond a few states.
• Pair trade: long MSFT/AMZN vs short DLR/EQIX for 1-3 months. The longs have balance-sheet and political optionality to self-secure power and land; the shorts are more exposed to local approval friction and valuation compression if build timelines slip.
• Watch-list only: long ETN or VRT on pullbacks if the market starts pricing a slower but more expensive grid expansion cycle. Falsify this if data-center capex guidance rolls over meaningfully or transformer lead times normalize faster than expected.
• No direct trade in DJT or STT on this headline alone; use them only as sentiment proxies if the policy debate broadens into a broader anti-tech campaign.