The House Committee on Homeland Security and the House Select Committee on China escalated their joint investigation Tuesday into American companies’ rising reliance on Chinese AI models, a probe first announced in April that now runs directly into a market share number Washington didn’t see coming. OpenRouter data reported by CNBC shows Chinese models have accounted for more than 30% of U.S. enterprise API token usage every week since Feb. 8, peaking at 46%. Across the prior twelve months the average was 11%. In the first half of 2025 it was 4.5%.

The mechanism isn’t ideology. It’s price. Justin Summerville, who works on data and analytics at OpenRouter, told CNBC the open-source Chinese options run “60% to 90% cheaper” than leading models from Anthropic and OpenAI. That gap widened into a rout in June, when a 19-day U.S. export-control suspension of Anthropic’s Mythos 5 and Fable 5 pushed Coinbase and Uber to route workloads to DeepSeek and Z.ai’s GLM 5.2, per Nikkei Asia. Lindy shifted 100% of its Claude traffic to DeepSeek.

Committee chairmen have already sent letters to Cursor, now being acquired by SpaceX for $60 billion, and to Airbnb. Cursor’s Composer 2 was built on Moonshot AI’s Kimi.

A State Department spokesperson said “The growing use of Chinese AI models by U.S. companies raises serious concerns,” arguing the systems “are designed to advance Beijing’s narratives, censor dissent, and reflect CCP ideology and values.” A spokesperson for the Chinese embassy in the U.K. said Beijing “opposes baseless allegations and malicious smears against its AI development.”

OpenAI plans to launch GPT-5.6 Thursday after a delay tied to national-security review, Reuters reported. The export-control instrument built to slow Beijing has, for one 19-day stretch, functioned as a customer-acquisition channel for it.

Sources