DeepSeek ends its «small team» era — doubles every department, closes first $7.4B external round
TL;DR
DeepSeek announces a company-wide doubling of staff, just as it closes its first external round: ~$7.4B raised at a $50B+ post-money valuation. Tencent reportedly takes ~20%.
DeepSeek announced on June 26 (WeChat) that every department will at least double in size, with hiring focused on data engineers, software engineers, and «AI cross-disciplinary technical» roles. The Hangzhou lab that built V3 and R1 with 150–170 people — and helped push OpenAI toward IPO — is dismantling its most famous identity: «small and tight».
The timing matters. DeepSeek's first external funding round is in final stages: ~¥50B (~$7.4B), post-money $50B+. Tencent (reportedly ~20%), CATL and others are in. This is the first outside capital founder Liang Wenfeng has accepted since splitting DeepSeek out of High-Flyer in 2023.
The term sheet is what's interesting. Investors cannot wire money to the company directly; funds first go into a limited partnership controlled by Liang personally. 5-year lockup, no voting rights. DeepSeek takes the money without giving up governance — the most aggressive founder-control term sheet among frontier AI labs. Compare OpenAI (Microsoft commercial rights), Anthropic (Amazon cloud lock-in): Liang keeps control hard.
Scale-wise, even at double size DeepSeek lands at 300–400 — still the smallest frontier lab (OpenAI ~6,000, Anthropic ~2,500). The doubling isn't a catch-up; it's acknowledgement that the next-gen model is no longer something 150 engineers can build.
The hiring list signals direction: «AI cross-disciplinary technical» = biology, materials, drugs — the same path Anthropic has been building, the same path John Jumper joined Anthropic for last week. DeepSeek's expansion isn't aimed at chatbots; it's aimed at the scientific compute market.
For China's second-tier AI labs — MiniMax, Moonshot, Zhipu — this is a timer. The «small team papers its way to fame» path is closed once DeepSeek holds $7.4B raised at $50B valuation with 400 people.
via Bloomberg
The timing matters. DeepSeek's first external funding round is in final stages: ~¥50B (~$7.4B), post-money $50B+. Tencent (reportedly ~20%), CATL and others are in. This is the first outside capital founder Liang Wenfeng has accepted since splitting DeepSeek out of High-Flyer in 2023.
The term sheet is what's interesting. Investors cannot wire money to the company directly; funds first go into a limited partnership controlled by Liang personally. 5-year lockup, no voting rights. DeepSeek takes the money without giving up governance — the most aggressive founder-control term sheet among frontier AI labs. Compare OpenAI (Microsoft commercial rights), Anthropic (Amazon cloud lock-in): Liang keeps control hard.
Scale-wise, even at double size DeepSeek lands at 300–400 — still the smallest frontier lab (OpenAI ~6,000, Anthropic ~2,500). The doubling isn't a catch-up; it's acknowledgement that the next-gen model is no longer something 150 engineers can build.
The hiring list signals direction: «AI cross-disciplinary technical» = biology, materials, drugs — the same path Anthropic has been building, the same path John Jumper joined Anthropic for last week. DeepSeek's expansion isn't aimed at chatbots; it's aimed at the scientific compute market.
For China's second-tier AI labs — MiniMax, Moonshot, Zhipu — this is a timer. The «small team papers its way to fame» path is closed once DeepSeek holds $7.4B raised at $50B valuation with 400 people.
via Bloomberg
