U.S.export controls unwittingly boost Chinese open AI ecosystems: trails of policy reveal surge in models, papers and GitHub activity

UBS says about 60% of big companies are slowing AI spending. CFOs and CTOs are very focused on rising bills, while ROI still looks uneven. So executives are adding guardrails, cutting tools, and forcing tighter usage rules. i.e. enterprise AI is leaving its trial phase and becoming an engineering budget problem. The new discipline is about routing tasks to cheaper models without hurting output quality. That shift could pressure OpenAI and Anthropic first, because usage-based revenue depends on volume. Open-source and Chinese models could gain share when tasks need cost control over peak reasoning. Last week JP Morgan research published a report saying, Chinese AI models are up to 50 times cheaper than their American counterparts on a per-token basis. The report said Chinese firms accounted for over 45% of all traffic on the AI aggregation platform OpenRouter by April 2026, up from under 2%in late 2024. Google is already pushing Gemini 3.5 Flash as a faster, efficiency-focused model. Anthropic’s Claude Sonnet 5 also arrives as buyers ask for capable, cheaper autonomy. --- businessinsider .com/ubs-enterprises-ai-spending-tokens-2026-7  > **引用原帖 Rohan Paul (@rohanpaul_ai):** > U.S. chip restrictions helped push China to build and spread open AI models. > The authors tested this by looking at policy documents, open model releases, GitHub activity, research papers, company-linked papers, and U.S. patents. > They found that after major U.S. export controls, Chinese developers increased activity around open LLM projects much more than U.S. developers did. > ---- > Link – arxiv. org/abs/2606.15999 > Title: "U.S.Policies Unintentionally Accelerated China's Open AI Ecosystems" > https://x.com/rohanpaul_ai/status/2072215261202252015