Lay out the funding data for China's major AI labs and one number is bigger than expected.
Cumulative funding raised: approximately $9 billion. DeepSeek's new round pushing through has about $7.35B RMB more on the way.
That's before counting internal AI investments from Alibaba, Tencent, Baidu — just the publicly raised portion is already significant.
What Each Lab Is Doing
The money isn't evenly distributed. A few top players took most of it:
DeepSeek is the most eye-catching. Founder Liang Wenfeng personally contributed ~¥3B, 40% of the round, while holding 90% of the company. This founder-absolute-control structure is extremely rare in AI — most model companies see founder equity diluted to single digits after multiple rounds.
Moonshot AI (Kimi) raised about $2B backed by Alibaba, valued at $20B. Kimi's positioning is clear: long-context, document-heavy enterprise workflows. This is a direction that competes directly with GPT and Claude but with a different focus.
Zhipu GLM keeps iterating. The GLM-5 series and GLM-5V-Turbo tech reports show heavy investment in multimodal agent direction.
MiniMax takes a different route — Hailuo AI is performing well among C-end users while also maintaining an open-source presence.
China Strategy vs US Strategy
Put the US and China numbers side by side, and the gap is obvious. Top US AI companies' valuations and total funding far exceed their Chinese peers. OpenAI, Anthropic, xAI are all valued in the hundreds of billions, with individual rounds reaching tens of billions.
But Chinese labs' strategies are shifting:
Open-source first. Qwen, DeepSeek, GLM are all aggressively open-sourcing. Open-source isn't idealism, it's strategy — build developer ecosystem through open weights, then monetize through API services and commercial products.
Price competition. DeepSeek pushed API pricing to fractions of frontier models. This isn't just "burn cash to grab market" — low-cost models force the entire industry to reprice.
Talent density. Chinese AI labs' talent density is increasing. The number of researchers and engineers flowing back from overseas is growing, and domestic top universities' training systems continue to produce talent.
A Signal Worth Noting
More funding doesn't guarantee winning. The competitive dimensions in the model industry are many: compute reserves, data quality, engineering capability, ecosystem building, commercialization — each is a marathon.
But one judgment is clear: the window for Chinese AI labs is still opening.
If DeepSeek's $7.35B comes through, combined with the $9B already raised, Chinese AI labs' total funding will approach the $20B level. The gap with top US companies is narrowing.
The next key variable isn't money — it's who can first make their model "good enough" — run through in real scenarios, get users willing to pay.
→ Further reading: DeepSeek $7B Round Founder Contribution | Moonshot AI $2B Funding
Primary sources: