The 2028 Democratic primary is coalescing around AI as a defining issue faster than any candidate has a workable framework for governing it. A New York Times/Siena College poll shows 61% of voters oppose datacenter construction; a CNN/SSRS poll finds 71% of Americans believe the government isn't doing enough to regulate AI. These numbers are creating gravitational pull — and eight potential candidates are orbiting at different distances. The field sorts into three tiers. Alexandria Ocasio-Cortez and Bernie Sanders occupy the hardest position: a moratorium on datacenter construction until Congress legislates guardrails on environmental, safety, and economic impacts. This is the only position that would actually halt the current buildout. Everyone else is proposing regulation of a train that's already moving. The middle lane is crowded. Kamala Harris calls for stricter federal regulations and echoes slowdown language. Gavin Newsom has signed incremental California-level safeguards — whistleblower protections, youth safety rules, a study on employee retention — and an executive order urging companies to consider an AI kill switch. Josh Shapiro champions investment while demanding federal regulation and datacenter environmental standards. Andy Beshear signed an executive order requiring datacenter developers to prove they won't harm the environment or taxpayers, and called for ending taxpayer incentives for datacenters. Mark Kelly and Ro Khanna represent the techno-optimist regulatory wing. Kelly published a roadmap proposing an "AI Horizon Fund" backed by AI companies to finance education, training, energy infrastructure, and displaced-worker support. Khanna wants an international agreement with China on AI safety inspections — comparing an unregulated AI race to "winning a nuclear war" — but explicitly resists the Sanders-AOC moratorium on datacenter expansion. Pete Buttigieg has warned about AI harms since 2017 and accused Congress of "sitting on its hands," calling for imagination comparable to the responses to the Great Depression or World War II. What's missing across all eight positions is enforcement architecture. Executive orders are suggestions. Congressional legislation doesn't exist yet. California's incremental approach relies on company compliance. Kelly's fund requires industry buy-in with no mandatory mechanism described. The moratorium proposal at least has a concrete verb — stop — but no viable path through a Republican-controlled Congress. Donald Trump has brushed off concerns entirely, insisting the US must lead in AI, and early Republican candidates are following suit. This creates a clean contrast for Democrats but also a trap: the party risks becoming defined by what it opposes (unchecked AI development) rather than what it would build (an actual governance regime). The 61% opposition to datacenters is a sentiment, not a coalition. Converting it into durable policy requires institutional machinery none of these candidates have yet proposed. The real extraction story here isn't partisan — it's temporal. AI companies are building capacity at a pace that makes regulatory catch-up structurally unlikely. Every month of positioning without legislation is a month of fait accompli. By 2028, the datacenters will be built, the energy contracts signed, the workforce restructured. The question isn't whether Democrats want regulation. It's whether any regulation they propose will arrive before the thing they're regulating becomes irreversible.
