Americans hate AI so much that politicians are starting to lose their jobs over it

The article highlights mounting local opposition to data center buildouts in Michigan—over 30 proposals in two years—driven by water, electricity, land-use, and environmental concerns. It notes states have courted tech demand tied to AI compute, including Michigan laws that exempt data center operators from sales and use taxes, while power demand claims (e.g., Google seeking 2.7GW in one township) raise utility investment and ratepayer cost sensitivities. Overall, the regulatory and community friction suggests incremental project delays and higher ESG-related execution risk even as AI-related capacity demand continues.

Analysis

The market should read this less as a secular hit to AI spend and more as a tax on the physical buildout path. The constraint is not demand for compute; it is permitting, water, and grid interconnection, which shifts economics toward operators with existing franchise rights and away from greenfield projects that need fresh local approvals. That makes regulated utilities and grid equipment suppliers the incremental winners, while hyperscalers absorb delays, higher carrying costs, and a greater chance of having to pay up for land, substations, or behind-the-meter solutions.

For DTE, the bull case is not volume growth alone but rate-base expansion tied to transmission, distribution, and generation upgrades needed to serve large loads. The risk is political: if customer backlash becomes a ratepayer issue, regulators can force cost sharing or slower recovery, turning a seemingly accretive load story into a capital-at-risk story. For GOOGL, the effect is mostly second-order: higher capex intensity and longer time-to-online for capacity additions, which matters more for AI monetization timing than for near-term revenue.

The contrarian miss is that this is a bottleneck, not a cancellation. In the next 1-3 months, local hearings and zoning fights create headline risk but limited P&L impact; the real pressure shows up over 6-18 months through project delays and higher utility investment needs. Falsifiers would be rapid approvals, clean cost-recovery rulings, or evidence that data center operators are successfully shifting to states with faster permitting and cheaper power, which would blunt the scarcity premium.

There may be no clean single-name trade here, but the setup favors owning the infrastructure side of the AI buildout rather than the compute layer. The best risk/reward is to wait for pullbacks in regulated utilities with visible load pipelines and avoid chasing hyperscaler optimism until permitting friction is resolved.

Key Decisions for Investors

• Watch DTE for dip-buy opportunities on any political headline selling; thesis is 6-18 month rate-base growth from grid buildout, but only if regulators permit timely cost recovery. Falsify on adverse rate-case language or explicit customer-rate backlash.
• Avoid adding to GOOGL solely on the data-center build narrative; the near-term effect is timeline slippage and higher capex, not a demand shock. Reassess only if project approvals accelerate materially or AI capex guidance improves.
• Pair idea: long DTE / short a hyperscaler basket proxy on permit-friction spikes, using a 3-6 month horizon. This is a timing trade on physical bottlenecks, with upside if utility investment is allowed to earn returns and downside if regulators cap recovery.
• Set an alert on state-level utility commission decisions and zoning rulings in major data-center markets; the trade reverses if approvals start clearing faster than expected, signaling the bottleneck premium is overstated.