The current strategy to maintain American dominance in Artificial Intelligence (AI) is hitting a wall of reality. Since 2022, the Biden administration has attempted to stifle China’s AI development through strict export controls on high-end semiconductors. However, recent observations from within China suggest these restrictions are not achieving their intended goal. Instead of a technological blockade, the U.S. may need to pivot toward a new objective: negotiating a global pact on AI safety.
The Failure of Export Controls
The logic behind U.S. chip restrictions was straightforward: prevent China from accessing the massive, skateboard-sized semiconductor sets required to power advanced AI data centers. The assumption was that these components were too large to smuggle and too complex to operate without direct American engineering support.
However, China has proven highly adept at bypassing these hurdles through several sophisticated methods:
- Cloud Circumvention: Chinese developers are renting computing power from data centers located in Southeast Asian neighbors, effectively training their models on foreign hardware while masking their origin.
- Hardware Workarounds: Rather than relying on a few ultra-powerful chips, Chinese engineers are learning to “stack” multiple less-powerful chips to achieve comparable results.
- The “Follower” Advantage: Through a process known as distillation, Chinese firms reverse-engineer cutting-edge American models. By studying the outputs of U.S. labs, they can rapidly build high-performing “copycat” versions that catch up to the leaders.
The Reality of the AI Race
For years, the prevailing theory among AI scientists was that the first nation to reach a “singularity”—a point where AI can recursively improve its own code—would win an irreversible race. The idea was that an intelligence explosion would make the leader untouchable.
While AI is indeed beginning to generate code to upgrade itself, the “winner” of the race may not be determined by who owns the most powerful raw model. The true impact of AI lies in deployment. To reshape economies and military capabilities, AI must be integrated into business workflows and weapon systems. In this context, being a few months behind in raw processing power is less critical than being able to effectively apply the technology.
A Shift in Strategy: From Containment to Cooperation
The attempt to “stop” China’s progress through hardware denial is proving to be an impossible objective. As China continues to close the gap through clever engineering and rapid imitation, the U.S. faces a strategic crossroads.
Because AI is a dual-use technology—capable of immense economic benefit but also catastrophic harm if mismanaged—the most pressing global threat is not necessarily who is “ahead,” but how the technology is governed.
A global agreement on AI safety could impose universal limits on the most dangerous applications of the technology, ensuring that the race for dominance does not lead to an uncontrolled global catastrophe.
Conclusion
The U.S. strategy of technological containment is failing to halt China’s AI trajectory. Moving forward, the priority should shift from trying to win a hardware race to establishing international safety standards that mitigate the existential risks posed by rapidly advancing autonomous systems.
