AI development follows the game theory of the nuclear arms race. If the U.S. slows down, it risks creating an asymmetric power dynamic where another nation like China could dominate. The goal is not to stop, but to achieve a global balance of power to ensure stability.

In the AI arms race, placing excessive constraints on domestic AI development while adversaries like China operate without them is a form of unilateral disarmament. This could leave the entire nation's digital infrastructure, from consumer data to government secrets, vulnerable to attack by more advanced, unrestricted foreign AIs.

Maintaining a significant technological lead over China is not just about competition. It allows US policymakers the time and space to develop a thoughtful, robust, and predictable domestic AI regulatory framework without the pressure of a neck-and-neck race, which forces chaotic, reactive measures that harm innovation.

The proposed data center moratorium, while intended to address safety, would create a strategic advantage for China and other nations if enacted unilaterally. An American slowdown without global agreement allows adversaries to catch up or surpass the US in AI, highlighting the prisoner's dilemma inherent in global AI regulation.

Gurley argues against heavy-handed U.S. AI regulation, like banning models with Chinese open-source components. He fears this could create a "fence around the U.S.," leading to a scenario where Chinese AI platforms, not American ones, dominate the global market, reversing the dynamic of the internet era.

Pausing or regulating AI development domestically is futile. Because AI offers a winner-take-all advantage, competing nations like China will inevitably lie about slowing down while developing it in secret. Unilateral restraint is therefore a form of self-sabotage.

Gurley posits a critical risk of heavy-handed US AI regulation. In the internet era, a 'fence' was built around China while US firms served the world. Over-regulation could reverse this, creating a fence around the US and allowing Chinese open-source AI models to dominate and serve the rest of the world.

A US policy that slows down its own AI labs for safety is logically flawed and self-defeating unless it also effectively slows China's progress. Unilateral deceleration doesn't make the world safer; it simply cedes ground to a less safety-conscious competitor, increasing net risk.

The US faces a paradox: restricting frontier AI models for domestic safety could push global customers and allies towards unregulated foreign alternatives, like China's. This effort to control AI risks forfeiting the long-term strategic advantage of having US technology become the global standard.