The left and right agree on one thing: no data centers

A bipartisan backlash to AI data centers is intensifying in the run-up to elections, highlighted by Hernando County, Florida’s unanimous approval of a one-year moratorium on new data center construction. Protestors across states cite localized harms including groundwater contamination/PFAS risks and water needs for cooling, while also arguing that AI data-center build-outs are not delivering durable local job gains. The article frames data centers as increasingly influential in electoral politics, potentially affecting Florida’s Republican primary dynamics (e.g., scrutiny of candidates’ links to AI/data-center development).

Analysis

This is less a demand shock to AI than a political entitlement tax on physical buildout. The market should distinguish between delayed megaprojects and destroyed compute demand: the former pushes revenue recognition out, the latter would justify multiple compression. That makes AMZN the cleanest read-through because AWS is the most exposed to siting friction, power interconnect bottlenecks, and local opposition to large-footprint expansion.

Second-order, the winners are incumbents with already-entitled power and shell capacity — colo owners and utility-adjacent infrastructure with existing permits — because scarcity of shovels-in-the-ground capacity should raise pricing power for existing facilities. Over 1-3 quarters, the real margin pressure is not construction cost alone but longer time-to-service, more community-benefit payments, and higher spend on water recirculation/onsite generation. That should favor operators that can monetize speed-to-capacity rather than pure scale.

The contrarian point: consensus may be overpricing a national anti-AI regime. This looks more like a local zoning backlash than a structural ban, so the base case is project migration, not cancellation. What would falsify that view is evidence that hyperscaler capex guides, especially AWS, start coming down or that delivery schedules slip materially over the next 1-3 earnings cycles.

Key Decisions for Investors

• Pair trade: short AMZN vs long EQIX or DLR for 1-3 months. Thesis: AWS bears the highest entitlement/power-friction risk while existing colocation assets benefit from scarcity and faster time-to-revenue. Cut the short if AWS capex guide or backlog remains intact through the next earnings print.
• Buy AMZN puts only on strength into earnings or any Florida/Arizona permitting escalation. Use this as a tactical hedge, not a core short; the thesis fails if management says data-center delays are being offset by new sites or if AWS growth reaccelerates.
• Do not short NFLX on this headline. The backlash is aimed at hyperscale physical infrastructure, not streaming demand; NFLX is at most a sentiment bystander. Treat as neutral unless data-center politics spills into a broader anti-tech spending slowdown.
• Set a watchlist alert for state-level preemption bills and utility/interconnect approvals. If regulators start standardizing reclaimed-water and zoning rules, the trade shifts from outright short AMZN to relative-value long existing capacity owners.