Gary Gensler warns US AI firms face challenges from Chinese models

Gary Gensler, the former SEC chair who spent years as the crypto industry’s most polarizing regulator, has turned his attention to a different kind of market disruption. In a Bloomberg Radio interview, Gensler warned that Chinese AI models are creating serious headwinds for American AI companies, both on the competitive front and on the question of whether hundreds of billions in spending will ever pay off.

Gensler, now a professor at MIT, called the situation a “big, consequential wrinkle.” The wrinkle, in plain terms: US firms are trying to build AI responsibly while Chinese competitors are scaling fast, playing by different rules, and rapidly eating into global market share.

The numbers behind the anxiety

The scale of US investment in AI infrastructure is staggering. Capital expenditures related to AI are projected to surpass $795 billion in 2026, with estimates suggesting that figure could blow past $1 trillion in 2027.

The competitive shift has been breathtakingly fast. Chinese open-weight AI models, developed by companies including DeepSeek, Alibaba, Moonshot AI, and MiniMax, accounted for roughly 1% of global AI workloads in late 2024. By the end of 2025, that share had surged to approximately 30%.

Safety meets geopolitics

Gensler referenced an incident involving OpenAI models and a breach connected to Hugging Face, using it to illustrate the real-world risks that come with rapid AI deployment.

Washington has already started paying attention. In April 2026, the House committees on Homeland Security and China launched a joint investigation into the implications of Chinese AI development. A particular focus of the probe is the unauthorized distillation of US models, essentially the practice of using outputs from advanced American AI systems to train competing Chinese ones.

The broader policy landscape

In September 2026, US and Chinese officials held discussions aimed at establishing communication channels around AI safety.