Data centers have moved from a local zoning fight to a primary campaign plank across the Midwest, according to NPR's reporting. Two months after OpenAI broke ground in Salem Township, Michigan, Democratic Senate candidate Abdul El-Sayed used the site as a backdrop. Francesca Hong, running for governor in Wisconsin as a Democrat, called for a moratorium. So did Vivek Ramaswamy, running for governor in Ohio as a Republican. In Michigan's 7th Congressional District, Democrat William Lawrence did not plan to run on the issue and made it a plank because voters kept raising it. A Gallup poll earlier this year put seven in ten Americans against an AI data center being built in their area.
When two candidates from opposite parties converge on the same policy, the issue has stopped being partisan and started being local. That changes the risk profile for anyone with a backlog. Eleven states now carry moratorium exposure that shows up in municipal bond pricing. Wisconsin's debate has been running on cooling water since last year, and Ohio's 25 megawatt ban fell short of the 2026 ballot in June and is aiming at 2027. A CDU order booked against a campus that gets frozen in a governor's first hundred days is a receivable, not revenue.
Nobody wins a primary talking about PUE. They win talking about a tower plume over a subdivision, a water bill, and a fan wall running at two in the morning. All three are outputs of the cooling design, which makes the mechanical spec the most politically exposed line item in a data center budget and the one least often discussed in public. Congressional staff put the odds of a federal framework this year at roughly zero, so the binding constraint stays at the state and county level, where the arguments are physical and specific.
Our read: the vendors with a credible zero-evaporation story spend 2027 selling into a market that has been pre-sold by its own opposition. Closed loop, dry cooling, and warm-water direct-to-chip all cost electricity relative to an evaporative tower, and that trade is now worth paying, because a project that clears a hostile county eighteen months earlier is worth more than a project with a better PUE that never clears at all. Operators still specifying towers in contested counties are pricing a political risk at zero. It is not zero anymore.