Before the Stack Hardens: AI Infrastructure, Public Options, and Governance Under Deep Uncertainty

This working paper argues that democratic AI governance should secure public leverage while the artificial-intelligence infrastructure stack is still being built. Governments are making decisions about compute, cloud services, procurement, data centres, electricity, water, land, and public-sector AI before the long-run consequences of AI are knowable. The paper develops a public-options framework for governing infrastructure under deep uncertainty and asks whether public action preserves the public's ability to observe, audit, interoperate, exit, revise, allocate, condition, and recover value after infrastructure commitments have been made.

The paper builds on three earlier articles in the AI-for-All sequence: Sovereign Compute and the Physical Stack; AI for All, Infrastructure for Whom?; and AI for All, By What Standard? It connects deep uncertainty, adaptive governance, procurement, compute strategy, electricity planning, data-centre development, cloud dependence, and public bargaining power into a single framework focused on preserving public leverage before dependence hardens.