WSJ: How China's AI talent network closes the gap with the US

创业邦 · wechat · 2026-08-25

Based on a Wall Street Journal report, this article analyzes the reasons behind China's rapid AI catch-up:

Talent Network: A Tsinghua-based lineage (e.g., Tang Jie of Z.AI and Yang Zhilin of Moonshot) and the "Yao Class" create a "virtual Silicon Valley."
Tech Diffusion: Open-source papers, code, and models (like DeepSeek's MLA/MoE) enable rapid iteration and cross-pollination among Chinese firms.
Engineering Efficiency: With 1/5th the chips and funding of US rivals, Chinese firms are forced to optimize engineering efficiency aggressively (e.g., distillation, low-cost architectures).
Commercial Gap: While model capability gaps have narrowed to "months," the scale of commercialization lags massively (Anthropic's revenue is 65x that of Z.AI).
Future Outlook: Export controls can't sever knowledge flows, but chips remain a critical bottleneck; competition is shifting from raw compute to efficiency.

More from Companies & People