DeepSeek requires investors to agree no-poaching clause
The real story in DeepSeek's reported no-poach demand is not the request itself but what it reveals about power and governance inside a leading Chinese AI lab: founder Liang Wenfeng appears to have traded capital access for hiring control, a reversal of the usual Western pattern where investors extract board seats and information rights rather than surrender their own hiring freedom. The claim, raised during a four-hour investor call in May and first reported by a fundraising-focused outlet owned by 36Kr, has not been independently verified by CNBC, and DeepSeek has not publicly confirmed it. It fits a broader pattern: senior researchers have left for Xiaomi, ByteDance and Tencent since late 2025, and the newly closed, roughly $7.4 billion round reportedly gives most investors no voting rights at all. For AI/ML practitioners, it is a concrete signal of how scarce frontier research talent has become, and how far a leading Chinese lab will go, contractually, to protect its team.
DeepSeek's reported demand that its own investors promise not to poach staff is a small contract term with an outsized signal: in a financing round reportedly stacked with founder-protective terms, hiring restraint became something investors evidently had to buy their way into rather than something they could negotiate for. That is close to a reversal of how AI financing usually works in the West, where investors extract board seats, information rights and sometimes informal access to a portfolio's talent pool in exchange for capital, rather than surrendering their own hiring freedom just to get an allocation.
What was reported
CNBC, citing a fundraising-focused outlet owned by 36Kr, reports that founder Liang Wenfeng raised the no-poach condition during a four-hour virtual investor call in May, telling participants not to poach DeepSeek's people, not to let their other portfolio companies poach them, and not to encourage staff to leave and start rival firms. CNBC says it could not independently verify the 36Kr account, and DeepSeek has not publicly confirmed or denied it. The company closed its first-ever external funding round this month at a valuation CNBC puts above $50 billion on a roughly $7.4 billion raise, reportedly the largest single funding round in Chinese AI to date.
Reading the governance signal
The no-poach clause did not arrive alone. Other reporting on the same round describes a structure controlled by Liang personally through a limited partnership, a five-year lock-up on investor capital, and voting rights withheld from nearly every backer except China's state-backed AI investment fund. Together, those terms describe a founder treating this raise as a capital-for-non-interference trade rather than the usual capital-for-governance trade, a structure that would be a hard sell to Western institutional investors accustomed to board seats and information rights in return for nine-figure checks. It also reflects a China-specific dynamic: strategic backers like Tencent operate their own competing AI labs, so a no-poach pledge functions as a more direct, more enforceable-sounding ask than the equivalent request would be for a US venture fund with no operating stake in a portfolio company's rivals.
The talent math behind the ask
The request tracks a real attrition problem, though maybe not the scale the headlines imply. Coverage of the round ties the clause to a string of senior-researcher exits to Xiaomi, ByteDance, Tencent and other rivals since late 2025. Research published alongside HTX's fundraising coverage puts overall attrition across DeepSeek's credited paper authors at roughly 6.4 percent, with only two of the fifteen most frequently credited researchers having left, which suggests the no-poach ask is less about stopping a mass exodus than about protecting a handful of hard-to-replace names and signaling stability to the staff and recruits DeepSeek still needs for its stated AGI ambitions.
Where accounts diverge
Deal math varies across outlets and should be read as a moving target rather than a settled number. CNBC's figures, a valuation above $50 billion and a $7.4 billion raise, describe the round's close; The Information's earlier reporting had Tencent and Alibaba in talks at a $20 billion-plus valuation, which shows how quickly the number moved as the round progressed. The no-poach claim itself still rests on one unverified account relayed through Chinese business media, not on a leaked term sheet or an on-the-record DeepSeek statement, and should be read with that caveat attached.
What to watch
Whether the no-poach language surfaces in executed agreements or other Chinese AI financings, and whether senior departures continue despite it, will show whether this was a one-time negotiating flex or an early marker of a market where scarce researchers, not scarce capital, have become the binding constraint on Chinese frontier labs.
Key Points
- 1DeepSeek reportedly required prospective investors to pledge not to poach its staff or fund rival startups, a condition raised during a May investor call.
- 2The request follows senior-researcher departures to Xiaomi, ByteDance and Tencent and accompanies a roughly $7.4 billion round with unusually founder-protective terms.
- 3It signals that scarce AI talent, not capital, is now the binding constraint for Chinese labs, reversing the usual investor-founder power balance in Western financing.
Scoring Rationale
Kept in the 'notable' band: this is a well-corroborated governance and talent-market signal tied to a record roughly $7.4 billion, $50 billion-plus valuation round, reportedly the largest single Chinese AI financing to date, and it connects to a documented wave of senior-researcher departures across Chinese AI labs. It stays below 'major' because the central no-poach claim rests on one anonymous-sourced, CNBC-unverified account rather than a confirmed document or on-the-record statement, and its direct relevance is organizational and governance-focused rather than a technical or product development. Adjusted modestly up from 6.8 given the breadth of independent corroboration found across trade press, business media and financial outlets during verification.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
