DeepSeek’s No-KPI, No-Overtime Playbook Isn’t a Flex — It’s the AI Race’s Quietest Cheat Code

Posted on by

(SeaPRwire) –

By: Lucas Caldwell

Everyone’s screaming that 996 is the only way to win the AI race, especially for Chinese firms going head-to-head with Silicon Valley giants. The common narrative has long been that China’s tech edge comes from cheaper labor and willingness to grind far longer hours than Western teams. DeepSeek just dropped a reality check that makes every founder bragging about 7-day workweeks look like they’re grinding for the wrong prize. A $60 billion valuation in three years, 173 million downloads since January 2025, and zero KPIs, zero mandatory overtime? That’s not a “chill startup” flex from the world’s richest AI founder. That’s a fundamental rejection of the playbook every tech leader has parroted nonstop for the last two years.

Liang Wenfeng, worth $37.9 billion as the world’s richest AI founder, laid out the logic in a nearly four-hour internal staff session transcribed by Tencent Technology. First, research needs a relaxed environment — push too hard and you kill the creative problem-solving AI work demands. Second, the team stays hyper-focused, doing very few things instead of spreading themselves thin across half-baked projects. There’s no rigid hierarchy, no employee rulebook, no written daily goals, no monthly targets. “No one manages them,” Liang says, and half of every workday is completely unallocated for employees to use however they want. He insists the company is run by “a bunch of very ordinary people,” not geniuses.

996 — 9 a.m. to 9 p.m., six days a week — was outlawed in China in 2021 but remains a tech industry staple. Alibaba, Tencent, and JD.com have all credited long hours for their rise, with Jack Ma calling 996 a “great blessing” in 2019. Michael Moritz first introduced the 996 concept to Western audiences in 2018, to widespread controversy. Now the mindset has gone mainstream in Western tech. 20VC founder Harry Stebbings says 7-day workweeks are required to win the AI race. Index Ventures partner Martin Mignot says 996 is the new startup standard, and founders no longer apologize for it. Some firms even add 996 clauses to employment contracts.

The 996 playbook works for execution-heavy businesses, like e-commerce logistics or social media feature iteration. You throw bodies at a problem, grind out incremental tweaks, and win on speed to market. For that kind of work, more hours usually mean more output, so 996 feels like a logical tradeoff. But the toll is real: young Chinese workers are quitting white-collar jobs for lower-paying manual labor to escape the grind, and youth unemployment keeps climbing. Cutting-edge AI research doesn’t work that way anyway. Breakthroughs don’t come from burnt-out researchers staring at screens for their 72nd straight hour that week. They come from unstructured thinking, casual brainstorming, and chasing weird side projects that don’t fit a pre-set roadmap.

The global AI talent pool is tiny, and every top researcher has their pick of jobs paying top dollar. What they rarely get is true autonomy and a say in what the team works on. Most firms lock their best people into rigid project roadmaps, measuring output by lines of code or quarterly model accuracy KPIs. DeepSeek’s vision-driven, unstructured model — no written mission statement, no top-down orders, decisions made by consensus — doesn’t just attract talent, it retains it. It lets researchers chase the ideas that actually move the needle, not just check boxes on a manager’s list. Those unplanned, self-directed projects are often where the biggest leaps forward come from.

Within 18 months, every top-tier AI lab in both China and the U.S. will roll out unstructured work time policies to stop their best researchers from defecting to competitors with more flexible setups.

Author bio: Lucas Caldwell, a tech opinion leader with millions of X/Twitter followers covering AI industry trends and startup strategy.