Abstract
How do domestic political structures constrain a state’s capacity to build and deploy frontier compute? Aggregate measures of industrial scale and defense spending do not answer the question. This article models technological state capacity as relative compute against a moving frontier. Regime coercion enters as a tax on true feedback fidelity; reported capacity is an observation, not the state. Three basins follow from the sign of a single growth gap β. Russia occupies an analogue attractor: relative compute is globally asymptotically stable at zero. China occupies a dual-pole band: market scale supplies a logistic restoration that yields a unique finite lag if and only if scale exceeds a coercion-and-sanctions threshold. The United States occupies an escape-velocity face if and only if its net growth outranks both rivals; that face is positively invariant and is destroyed if American administrative friction or talent restriction flips the ranking. 2025–26 evidence is consistent with the Russian and Chinese sign patterns. American industrial policy has already moved U.S. friction off the theoretical zero bound. Lyapunov proofs are given in the appendix.
Supplementary materials
Title
Technical Appendix
Description
This appendix supplies the Lyapunov arguments withheld from the revised main text.
Actions



![Author ORCID: We display the ORCID iD icon alongside authors names on our website to acknowledge that the ORCiD has been authenticated when entered by the user. To view the users ORCiD record click the icon. [opens in a new tab]](https://www.cambridge.org/engage/assets/public/coe/logo/orcid.png)