TL;DRDeepSeek founder Liang Wenfeng says the $60 billion AI lab runs with no KPIs and no overtime while Western VCs push 996
Liang Wenfeng, the founder of DeepSeek, told staff in a nearly four-hour discussion that the company does not do overtime and does not use KPIs. “No one manages them,” he said of his researchers, according to a transcript obtained by Tencent Technology. The remarks describe a workplace culture that is the opposite of the 996 grind that Western tech founders are now rushing to copy.
Liang gave two reasons for the approach. “Research needs a relaxed environment; if you push too hard, you can’t do research,” he said. The second is extreme focus, meaning DeepSeek does very few things at once.
Employees are told what to do for only half their working day, and the rest is unstructured time they can spend however they choose. There is no formal hierarchy, no written rulebook, and no monthly targets. “We don’t really have an organization, we’re driven by a vision, organized by a vision,” Liang said.
The company he is describing is not a quiet research lab. DeepSeek has been downloaded 173 million times since it launched to the public in January 2025, and a funding round in June valued the Hangzhou startup at roughly $60 billion. Liang is now the wealthiest AI model creator in the world, with a fortune Bloomberg estimates at roughly $36 billion, ahead of Anthropic co-founder Dario Amodei and former OpenAI president Greg Brockman.
