The article argues that the “AI arms race” framing is colliding with the economics of AI. Governments want scale and interoperability, but also sovereignty: control over data, compute, models, standards and talent. Since the full stack is too costly for most states, sovereignty becomes modular risk management, and energy constraints make compute a strategic bottleneck. Cloud regions still sit under jurisdiction, so access can become a bargaining chip.

Collaboration still pays where externalities cross borders: safety science, benchmarking, incident sharing and interoperable standards. This creates layered coexistence: open coordination at the bottom, control at the frontier. The U.S. pairs safety cooperation with export controls, the EU pools capacity via the AI Act and AI Factories, China enforces tight domestic rules and India bets on sovereignty-through-access and open ecosystems. The takeaway: treat access risk, energy and standards as first-order strategy variables.

In normal times, markets lean on two anchors: steady data and clear central-bank guidance. In 2024–2025, both feel loose. Measures of U.S. policy uncertainty have hit multi-decade highs, and investors are reacting less to trends than to each new twist in the story.

What looked like a near-certain December rate cut suddenly became “less clear-cut and less certain” after Fed officials publicly split; the October 2025 meeting produced rare dissents in both directions, and Chair Powell cautioned that another cut was “not a foregone conclusion.” The policy path has been a moving target, and markets are trading on that uncertainty rather than clarity.