DeepSeek has suspended its second external funding round, telling prospective backers on Saturday, July 25, that it would not sign investment agreements as expected, Bloomberg reported, citing people familiar with the discussions. The round would have valued the Hangzhou-based AI lab at roughly $71 billion — about 40% above the valuation it commanded just six weeks earlier.
The trigger was not a financing dispute. According to those reports, founder Liang Wenfeng grew frustrated after an account of his remarks to investors during the company's first fundraising circulated widely across Chinese tech and investor circles.
Key Highlights
- DeepSeek verbally informed prospective investors on July 25 that it is suspending the round "for now"
- The round targeted at least 10 billion yuan at a pre-money valuation of at least 480 billion yuan, roughly $71 billion
- The pause follows the online spread of remarks attributed to Liang from a closed-door investor meeting, first reported by Chinese outlet Yicai
- DeepSeek closed its first-ever external round of about $7.4 billion in June 2026, at a valuation north of $50 billion
- Talks remain fluid and the company may resume the process later
What Happened
DeepSeek opened preliminary discussions with new investors in mid-July, only weeks after closing its first outside round. Those early talks pointed to a pre-money valuation of at least 480 billion yuan and a raise of at least 10 billion yuan in fresh capital.
That process is now on hold. People familiar with the matter told Bloomberg the company verbally notified would-be backers that it was suspending the deal, without signing the agreements investors had expected in the following days.
The reported cause is a leak. Chinese financial outlet Yicai published an account of comments attributed to Liang from a lengthy closed-door meeting with investors during the first round. In that account, Liang is described as acknowledging that DeepSeek still relies heavily on Nvidia chips, that China continues to trail the United States in AI sophistication, and that the gap between the two countries comes down to computing power rather than talent. He is also described as saying DeepSeek would prioritise AGI research over short-term profit and would likely keep its strongest models open-source.
DeepSeek has not publicly confirmed the accuracy of those remarks, and the company has a long-standing practice of near-total public silence.
Why It Matters
The pause is a signal about control, not solvency. With roughly $7.4 billion raised in June, DeepSeek is not short of runway, and halting a round at this scale suggests a founder able to stop capital momentum rather than be carried by it — a governance pattern increasingly common among frontier AI labs.
It also underlines how fragile confidentiality has become around Chinese AI financings. Investor meetings at this level are tightly held; a leaked account moving through public channels is itself the kind of event that can reprice a deal, independent of the company's fundamentals.
For the open-source ecosystem, the leaked remarks matter more than the funding delay. DeepSeek's releases have repeatedly reset expectations for what open-weight models can do, and any indication about the company's commitment to publishing its strongest models carries weight for developers building on top of them — including teams across the MENA region that have adopted open-weight Chinese models to avoid per-token dependency on Western providers.
Background
DeepSeek spent most of its existence funded internally. Liang also runs High-Flyer, a quantitative hedge fund with deep enough pockets to bankroll AI research without outside capital, which allowed the lab to ship frontier open-weight models without a conventional venture track record.
That changed in June 2026, when DeepSeek raised about $7.4 billion in its first external round at a valuation above $50 billion — one of the largest startup financings China has seen. Backers included Tencent Holdings, battery maker Contemporary Amperex Technology (CATL), and the state-backed National Artificial Intelligence Industry Investment Fund. Liang himself contributed roughly $3 billion, and his net worth roughly doubled to about $36 billion on the round, making him the wealthiest founder of any frontier AI model developer.
What's Next
The private round may not be the main event. DeepSeek has begun preparing for a domestic listing on Shanghai's STAR Market, holding early discussions with accounting firms and investment banks, with reports pointing to a filing as soon as this year and a debut potentially in 2027.
If a listing is the intended exit, the urgency of a second private round drops considerably — which makes the pause readable as sequencing rather than distress. DeepSeek would join a broader wave of Chinese AI companies, including Moonshot AI, moving toward public markets.
For now, the company has kept its options open: people familiar with the talks describe negotiations as fluid, with the possibility that DeepSeek resumes the raise once the fallout settles.
Source: Bloomberg