Chinese AI models now handle nearly half of U.S. router traffic

Layer3Labs says Chinese-developed models now process as much as 46% of token traffic on a major U.S.-based AI model router, up from about 30% a year ago. The shift raises data-sovereignty, security and compliance concerns because many companies may not know which model is handling their customer data.

Why it matters: - Layer3Labs says a large share of U.S. business AI traffic is now routed to Chinese-developed models, creating potential exposure for customer and company data. - The firm says many businesses are using AI features without knowing which language model is actually processing the prompts. - The risk is especially relevant for healthcare, legal and finance companies that handle sensitive information.

What happened: - Layer3Labs said as much as 46% of token traffic on the largest U.S.-based AI model router is now funneled to Chinese models. - The firm said U.S. models handled about 70% of that traffic a year ago and are now closer to 30%. - The company tied the shift to routing defaults, vendor backends and so-called shadow AI. - Layer3Labs published the findings as part of an effort to push businesses to check the provenance of their AI tools.

The details: - Cost-optimizing routers can send prompts to the cheapest available provider, which Layer3Labs says is increasingly a Chinese endpoint. - Layer3Labs said inference at those endpoints can be 60% to 90% cheaper. - Andreessen Horowitz estimates 80% of U.S. startups build on Chinese base models, with Airbnb, Uber and Cursor listed as named adopters. - Research cited by Layer3Labs puts 70% of enterprise AI activity outside IT oversight, including models small enough to run on a laptop. - Any prompt processed by a Chinese-provider endpoint can fall under China’s National Intelligence Law, which can compel data disclosure to the state. - In July 2026, the House Select Committee on the CCP opened a probe into U.S. firms’ use of these models. - The NDAA already bars DeepSeek from Defense Department systems. - Firms with high shadow AI exposure pay an estimated $670,000 breach premium per incident. - Layer3Labs said its review of dozens of small- and mid-sized business AI stacks found 70% contained at least one Chinese model the company had not knowingly approved. - Layer3Labs is offering a free AI Stack Sovereignty Check that identifies which models are processing a company’s data and flags data-sovereignty and compliance risk. - Layer3Labs is based in Miami and says it helps small and mid-size businesses deploy AI in 2 to 4 weeks.

Between the lines: - The bigger issue is not just model choice, but visibility. Many firms appear to be buying AI capabilities through vendors that can quietly swap or route traffic behind the scenes. - The report suggests the fastest path to cheaper AI may also be the hardest to govern, because shadow deployments can bypass security review entirely. - The policy response is starting to catch up, but the operational problem inside businesses may be outpacing internal oversight.

What’s next: - Layer3Labs is urging companies to run provenance checks on every AI tool that touches customer data. - The firm says the first step is identifying what is actually in an AI stack before setting governance or compliance rules. - Businesses with sensitive data are likely to face more pressure to document where AI prompts are routed and which jurisdictions apply.

The bottom line: - For many U.S. businesses, the AI supply chain is becoming a data-sovereignty issue as much as a cost issue.