Yesterday, U.S. export controls on advanced AI chips-intended to stall China's ascent-have unintentionally spurred a wave of cost-effective Chinese models that now lag leading American companies by just months.
In response, Chinese developers embraced open-weight architectures and frugal training methods, producing models that compete with OpenAI and Anthropic while operating at a tenth of the cost, a shift driven by market forces rather than a coordinated government plan.
This week, startups like DeepSeek showcased AI models that matched much of the performance of U.S. rivals, boosting confidence after years of perceived lag, while leading American firms remain far ahead in valuations and patent impact.