China Seeks AI Technology Self-Reliance, Weakening Washington’s Leverage Over Beijing

中国寻求AI技术自主，削弱华盛顿制衡北京筹码

DeepSeek optimized its latest model for inference on Huawei chips for the first time, while two semiconductor sources said training still relies on Nvidia chips; Huawei says it plans to release a training chip this year, but matching current Nvidia performance will take another year.

Why it matters: HKR-H/K/R all pass: NYT ties DeepSeek-Huawei chip optimization and Huawei's training-chip timeline to US export-control leverage. It is not a model launch and lacks benchmark results, so it stays in the 78–84 band.