OpenRouter: Chinese AI models have drawn 30%+ of token use by US companies each week since February 8, peaking at 46%, up from 11% over the previous 12 months
Chinese-built AI models are gaining traction among U.S. companies as they narrow the performance gap with leading American rivals …
Context & Ripple Effects
Related coverage traces a progression from Chinese developers releasing open models to broaden access and enlist outside refinement, to those models gaining download share and adoption among U.S. startups. Cost, customizability, and capable-enough performance have been the recurring reasons cited for uptake.
The OpenRouter usage data makes that shift more consequential: it moves the story from model availability and downloads to sustained production usage by U.S. companies, even as Chinese models remain far behind ChatGPT in global consumer downloads.
First-order effects
- Chinese model providers gain a materially larger foothold in U.S. enterprise inference workloads on OpenRouter, while U.S. model vendors lose relative token share on that channel.
- U.S. companies using OpenRouter have stronger evidence that Chinese alternatives can be deployed for meaningful workloads, particularly where price and customization matter.
Second-order effects
- Frontier U.S. model providers face greater pressure to defend enterprise demand through pricing, model efficiency, open-weight offerings, or clearer performance differentiation.
- Open-model tooling, hosting, evaluation, and integration providers stand to benefit as customers compare and mix models rather than standardize on a single U.S. vendor.
Third-order effects
- If sustained across platforms, enterprise AI demand could become more multipolar: model origin matters less than the trade-off among capability, cost, and deployability.
- The growing use of Chinese-built models by U.S. firms is likely to keep policymakers and corporate procurement teams focused on how openness and cross-border model adoption interact with existing technology restrictions.
The trend: Enterprise generative-AI adoption is shifting from a frontier-model race toward a competitive, multi-provider market in which lower-cost open models can win production workloads.