Harvard scholar: the data-center backlash is just getting started

The article highlights growing local and state pushback against data center expansion, including moratoriums, ballot measures, zoning limits, and lawsuits. It cites more than 1,000 pending U.S. data center proposals, along with concerns over higher electricity prices, air pollution, water use, and noise. While the AI buildout remains strategic, regulatory friction could slow project timelines and raise costs for developers and utilities.

Analysis

The key market implication is not that data centers are being blocked outright, but that the permitting path is getting longer, noisier, and more state-dependent. That pushes risk from a simple “AI demand = infrastructure buildout” narrative into a more nuanced winners/losers framework: developers with the strongest land, power, and water optionality can still grow, while marginal projects face delay risk, legal costs, and higher hurdle rates. For ORCL, the issue is less demand destruction than execution drag on large campus builds and a higher probability that promised capacity converts later than expected.

Second-order, the pressure is upstream on utilities, gas peakers, transmission gear, cooling systems, and water-constrained municipalities rather than on the hyperscalers alone. If local resistance continues, the bottleneck shifts from compute demand to interconnection and site permissions, which tends to favor incumbents with existing megawatt access and penalize fast followers that need greenfield entitlements. Over the next 6-18 months, this can create a bifurcation: a few “approved” mega-sites absorb even more capital while a larger set of announced projects stays in limbo, increasing headline risk without immediately denting aggregate AI capex.

The contrarian miss is that public pushback may actually improve project economics for the surviving winners by reducing speculative overbuild and tightening supply. That means the near-term negative for developers can become a medium-term positive for pricing power in colocation, power procurement, and utility infrastructure. GOOGL is neutral here operationally, but any parent-level read-through is modestly positive for hyperscalers that can self-fund and self-permit versus standalone developers that rely on local goodwill.

The main catalyst path is local-election and litigation-driven: delays can show up in weeks to months, while zoning preemption or state overrides would re-rate the space over quarters. The tail risk is a broader anti-AI political narrative that spills from land use into tax policy, water pricing, or utility rate cases, which would raise project IRRs and compress multiples across the ecosystem.

Key Decisions for Investors

• Underweight ORCL into data-center headline risk for 1-3 months; use rallies to trim because the market may be underpricing entitlement and litigation slippage on large AI infrastructure commitments.
• Long a basket of regulated utility and grid-equipment beneficiaries versus short data-center developers: pair FE/EIX/ETN against infrastructure-heavy AI enablers where project timing is most exposed; 3-6 month horizon.
• If you want expressed downside in ORCL, buy 3-6 month put spreads rather than outright puts to isolate timing risk from broad market beta; risk/reward improves if local moratoriums keep spreading.
• Stay neutral GOOGL short-term, but use any drawdown tied to AI-infrastructure sentiment as a relative long versus ORCL because hyperscalers can internalize permitting, power procurement, and capital allocation better.
• Watch for a state-level preemption catalyst; if multiple states override local zoning, cover shorts in developers quickly since the market will reprice project conversion probabilities within days.