**[Sequoia Warning: The U.S. May Win with Closed-Source Models but Lose the Open Foundation]** According to monitoring by *Beating*, Sequoia partners Dean Meyer and Konstantine Buhler published an article stating that while the U.S. leads in closed-source models, Western companies are increasingly reliant on China's open-weight models. Qwen, Kimi, GLM, and DeepSeek are being used as foundational platforms, teacher models for training, and sources of synthetic data. The ATOM report shows that Qwen's share in newly fine-tuned and adapted models rose from 1% in January 2024 to 69% in February 2026. Thinking Machines' *Inkling* also used synthetic data generated by open-weight models like Kimi K2.5 during its early post-training stages. However, this only accounts for a small portion of training compute. The article argues that the issue lies in "distillation" rules. Distillation involves using the outputs of a stronger model to train another model. OpenAI and Anthropic restrict customers from using their model outputs to train competing products, but U.S. companies can legally learn from China's open models. The authors suggest that U.S. frontier labs sell controlled, delayed, and auditable training rights to qualified companies. Otherwise, the U.S. may continue to lead in closed-source models but hand over the open-model foundation to China. China has recently discussed restricting overseas access to certain advanced models, though no final policy has been established. The article warns that even if existing models remain downloadable, Western companies may gradually fall behind due to lack of access to subsequent versions. [Original Link]