Effective AI Pushback Must Scale Beyond Local Action to State Policy, Interstate Compacts, and Eventual National Regulation
Source: Aziz Huq. "An Effective AI Pushback Has to Go Beyond the Local." November 9, 2026. jacobin.com
The Gist
The author argues that the AI boom hurts people whether it succeeds or fails: if it works, workers lose income; if it fails, taxpayers get stuck with the bill. Since local fights over individual data centers can't address these national problems and Washington is blocked for now, states should end tax breaks, make AI firms pay for their social costs, and team up in interstate compacts to push toward national rules.
Conclusion
Because hyperscale AI's harms are national in scope and local bans or pauses are inadequate, progressives should escalate resistance beyond the local level by using state-level policy (ending tax breaks, requiring bonds and cost accountability) and interstate compacts as stepping stones toward eventual national regulation.
Premises
- Hyperscale AI is a single Silicon Valley strategy: a few firms take on massive debt to build computing power that is commercially viable only if sold at scale, most likely as a low-cost substitute for white-collar labor.
- If the strategy succeeds, it shifts income from labor to capital and shrinks the state's fiscal capacity (since capital is taxed less than labor); if it fails, the debt-laden firms will likely demand a public bailout, transferring risk to public balance sheets.
- Either way, trillions in AI investment divert resources from urgent social needs, so public anxieties about AI (water, grid, jobs, debt) reflect real, interconnected, national-scale problems.
- Local responses (permit denials, bonds) are limited by cities' narrow powers and geographic reach, and cannot address labor-market effects, fiscal risk, or most grid-related environmental costs.
- Federal action is currently blocked (Trump's opposition, the December 2025 executive order, and supermajority hurdles), so it cannot be the immediate vehicle.
- State pauses and bans are merely reactive and negative; they fail to communicate the full harms, and data centers just get built elsewhere.
- States can instead end data center tax exemptions, require bonds and reporting on environmental and labor effects, fund guarantee funds, require nondebt financing, and bar uses such as mass job automation, and can coordinate through interstate compacts to regulate at scale without Congress.
Assumptions
- Hyperscale AI firms are commercially viable only through labor substitution or else face financial distress requiring a bailout.
- Voters' diffuse anger at data centers can be channeled by the Left into a coherent, comprehensive critique and policy agenda.
- States have the legal authority to impose these conditions despite the executive order, whose legal theories the author considers flawed.
- Interstate compacts can be formed without congressional approval and would withstand legal challenge and industry pressure.
- State-level policies will build momentum toward national regulation within a couple of years.
- Democratic trifectas in states are sufficiently unified and willing to adopt these measures.
- Capital is taxed at lower rates than labor, so labor-share losses reduce fiscal capacity.