A Chinese robotics startup engineer named Zhuhuai Jiang thinks the key to China's string of powerful AI models isn't talent or resources—it's organizational culture.
In his view, teams behind models like DeepSeek and Kimi (both Chinese AI startups) are flat. Research and engineering aren't siloed. The same people switch between algorithms, data, and training infrastructure, driven by one question: does the model work?
He criticizes Silicon Valley for anointing "researcher" as a high-status title while treating infrastructure work as grunt labor. Engineers scramble to rebrand themselves as "research engineers," leaving the infrastructure work to whoever couldn't escape. But at frontier-model scale, infrastructure directly determines experiment speed, and experiment speed determines research output.
Elon Musk weighed in, calling it "both pretentious and toxic" for for-profit companies to label themselves as "labs." He said they create a "noble versus peasant" hierarchy between researchers and engineers. When either is doing good work, he added, they're essentially all engineers.