AI Infrastructure Is Becoming Local Politics.

Why is AI infrastructure a local political issue?

The industrial era of AI has substations, cooling systems, land deals, transmission lines, and neighbors. Lawrence Berkeley National Laboratory estimates that US data centers used 4.4% of electricity in 2023 and could reach 6.7% to 12% by 2028.

Once that load becomes visible in electricity bills, water debates, and planning fights, AI stops being a distant software story. It becomes ordinary local politics.

Why did we predict elections would turn on data centers?

The 2025 State of AI Report put multi-gigawatt data centers and power constraints at the center of frontier competition. The US Department of Energy separately warned that data centers could consume up to 9% of US electricity by 2030, from 4% in 2023.

The 2025 off-cycle elections gave the thesis its first test. In Virginia, home to roughly 13% of global data center capacity, data centers and energy affordability were central to the governor's race; Abigail Spanberger won on a platform that requires data centers to pay their fair share for power. In Georgia, two Democratic challengers unseated incumbent utility regulators with more than 62% of the vote in races dominated by power bills and data center demand. In August 2025, Tucson's city council rejected the Amazon-linked Project Blue data center 7 to 0 after weeks of public pressure over water.

The Tucson coda is instructive: after the council vote, the project advanced anyway through county channels and a state utility approval. The politics can be loud and the build can still happen. That gap, between what voters reject and what gets built, is where the 2026 midterm fights will live.

What would count as a hit?

For data center politics, the hard evidence is a campaign ad, debate, poll, candidate platform, or post-election analysis showing that an AI infrastructure project materially shaped a 2026 midterm, state, or gubernatorial race. The 2025 Virginia and Georgia results are strong leading indicators, but the prediction names the 2026 cycle.

For AI neutrality, the clean evidence is a government or bloc formally adopting non-alignment in compute procurement, model access, cloud infrastructure, or technical standards.

Why it matters

AI infrastructure puts the industry in contact with voters who may never follow a benchmark but do care about power prices, water use, construction, jobs, and who captures the upside.

Those voters can slow projects, change utility rules, redirect investment, and reshape the geography of frontier AI. A model race that looks global from San Francisco can still be decided by a county permit.

What we are watching now

We are watching state and local permitting fights, new utility tariffs for very large loads, grid operators naming AI demand as a material planning risk, and whether data centers appear in 2026 midterm campaign advertising.

Internationally, the signal is whether governments frame cloud and model procurement as non-alignment rather than simply choosing the cheapest US, Chinese, European, or regional provider.