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. AI router, up from about 30% a year ago. The firm warns that many businesses are using these models through vendors and shadow AI without realizing customer data may be leaving U.S. oversight.

Why it matters: - Chinese AI models are now handling a major share of U.S. business AI traffic. - Layer3Labs says that creates data-sovereignty, compliance, and breach-risk exposure for companies that do not know which model is processing their prompts. - The risk is highest for regulated industries such as healthcare, legal, and finance.

What happened: - Layer3Labs said as much as 46% of token traffic on the largest U.S.-based AI model router is now routed to Chinese models. - The firm said U.S. models handled about 70% of that traffic a year ago and now account for closer to 30%. - The company issued its warning on July 22, 2026, from Miami.

The details: - Business owners often buy an AI feature rather than a specific model, which lets the underlying model be routed, resold, or white-labeled without notice. - Cost-optimizing routers can send prompts to the cheapest available provider, and Layer3Labs says Chinese endpoints are often 60% to 90% cheaper for inference. - Andreessen Horowitz estimates 80% of U.S. startups build on Chinese base models, with Airbnb, Uber and Cursor named as adopters in the source material. - Research cited by Layer3Labs says 70% of enterprise AI activity sits outside IT oversight, often through small Chinese models that can run on a laptop. - Layer3Labs said its own 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. - 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.

Between the lines: - The shift is partly invisible because the AI layer is increasingly abstracted away from the buyer. - That makes vendor risk and model provenance a bigger issue than model choice alone. - The warning also reflects a broader U.S.-China competition over who supplies the infrastructure behind enterprise AI.

What’s next: - Layer3Labs is urging businesses to run a provenance check on every AI tool touching customer data. - The firm is offering a free AI Stack Sovereignty Check that identifies which models are actually processing a company’s data and flags data-sovereignty and compliance risk. - Layer3Labs said it is especially important for firms in healthcare, legal and finance to map their AI stacks now.

The bottom line: - Many U.S. companies may already be using Chinese AI models without realizing it, and the real risk is not just cost — it is control over where sensitive data goes.