Open Source

DeepSeek's First Outside Round: Valuation Talk Hits $45B

DeepSeek is negotiating its first-ever outside round per FT and Bloomberg — a state chip fund to lead, Tencent and Alibaba in talks, valuation up from $20B to $45B.

DeepSeek's First Outside Round: Valuation Talk Hits $45B — article cover

On May 6, 2026, TechCrunch, citing the Financial Times and Bloomberg, reported that DeepSeek is negotiating the first outside investment round in its history — and that the valuation under discussion has jumped from $20 billion to $45 billion in a matter of weeks. The round is reportedly being led by the China Integrated Circuit Industry Investment Fund, the state-backed vehicle known as the “Big Fund,” with Chinese cloud giants Tencent and Alibaba in talks to join.

The significance is not the money; it is the identity of the money. The world’s best-known open-weights lab, previously self-funded by founder Liang Wenfeng’s hedge-fund capital, is now preparing to bring state capital and cloud incumbents onto its cap table. That will reshape the competitive structure of China’s AI camp.

From $20 Billion to $45 Billion in Weeks

Talks are ongoing and details can move. The FT reports that the valuation being discussed climbed from $20 billion to $45 billion within a few weeks — a figure that, if it holds, would put DeepSeek among the most valuable independent AI labs anywhere. A two-plus-times jump inside a month is not a financing; it is an auction, and it says as much about buyer scarcity as about the asset: there is exactly one lab of DeepSeek’s stature still available, and everyone negotiating knows it. Bloomberg adds that the Big Fund, the core investment vehicle of China’s semiconductor self-sufficiency push, is expected to lead, with Tencent and Alibaba negotiating follow-on positions. TechCrunch could not immediately reach DeepSeek for comment.

Why Raise Now

DeepSeek has never taken outside money — its sharpest contrast with every other frontier lab. Per the FT, founder Liang Wenfeng, who controls nearly 90% of the company, decided to raise not because the lab lacks compute, but because of people: competitors have started poaching DeepSeek’s researchers, and the company needs a structure that lets it grant employees equity.

Read that carefully. This round is a talent-retention war first and an expansion war second. The lab that shocked the industry with absurdly cheap training runs has finally bumped into the market reality that every star researcher can be bought — a warning for any team whose edge lives in a handful of heads. It also means the round’s structure matters more than its size: an equity plan for employees implies governance changes, dilution for the founder, and reporting expectations that a purely self-funded lab never faced.

State Capital on the Term Sheet

The Big Fund’s role pushes this round beyond commercial territory. The fund is the centerpiece of China’s semiconductor strategy; if it leads, state capital will sit directly behind the flagship of open-weights models. The reported interest from Tencent and Alibaba adds a second layer: cloud giants want DeepSeek’s models bound tightly to their compute, producing a structure where the state fund supplies credibility and the clouds supply distribution. For anyone deploying DeepSeek weights in production, that composition cuts both ways — deeper pockets and long-term stability on one side, geopolitical exposure and potential usage strings on the other.

Open Weights and Huawei Silicon

Since detonating onto the scene in early 2025 with frontier-adjacent training costs, DeepSeek has kept pace in reasoning and coding while staying open-weight — V4-Pro is freely available on Hugging Face. Its models are also optimized to run on Huawei silicon. That software-plus-hardware pairing is China’s mainline answer to building autonomous AI capability under export controls: if Nvidia’s best chips are hard to obtain, the combination of a top open model and domestic accelerators becomes the plan of record, not a fallback.

Contrast that with Meta’s reported drift toward closed frontier releases — the open-source divergence that opened 2026 — and open weights now mean different things on either side of the Pacific. In the US it is a strategic option; in China it has become the state team’s calling card. Whatever the final terms, developers should not expect DeepSeek’s open-weights policy to reverse any time soon — that openness is precisely what the incoming capital is buying into.

Sources

AI-assisted summary compiled from the sources above, reviewed by a human before publishing.

SHAREXEMAIL