- The AI startup has reportedly delayed signing a planned funding round after confidential remarks by founder Liang Wenfeng were circulated online
- The incident highlights growing tensions between secrecy, talent retention and investor expectations as China’s AI race intensifies
China’s AI upstart DeepSeek has reportedly put its second fundraising round on hold after founder Liang Wenfeng became reportedly dissatisfied with the leak of confidential discussions from a private investor meeting, according to Bloomberg.
The reversal comes just weeks after reports that the Chinese company was seeking to raise at least 10 billion yuan at a pre-money valuation of about $71 billion.
The controversy stems from a transcript of what was described as a nearly four-hour closed-door investor session in mid-July.
The transcript, which was widely circulated online and later verified by multiple Chinese media outlets, contained Liang’s candid views on topics ranging from U.S.-China AI chip competition to the future of China’s large language model industry and DeepSeek’s long-term AGI strategy.
Among the most closely watched remarks, Liang reportedly argued that Huawei’s Ascend 950 SuperPoD could serve as a viable alternative to Nvidia’s GB200 systems, saying “four Huawei chips can match one Nvidia chip.”
He also suggested the technology gap was “fourfold and about two years” and claimed Nvidia’s CUDA software ecosystem was gradually being weakened by advances in AI.
Bloomberg, citing people familiar with the matter, reported that Liang was upset that internal strategic discussions had been leaked by investors and circulated publicly.
As a result, DeepSeek has informed some prospective investors that signing of the new financing agreements would be postponed.
The company has not abandoned the fundraising altogether, however, and negotiations could resume once the controversy subsides.
DeepSeek completed its first external fundraising round in June, raising about $7 billion in what was reported as the largest first-round financing for a Chinese technology startup.
Investors included Liang himself, Tencent, CATL, JD.com, NetEase and IDG Capital, valuing the company at roughly $52 billion post-money.
Talent pressures
Liang has previously maintained that DeepSeek would avoid external financing, public listings and aggressive commercialization.
But the company has softened that stance as competition for AI talent intensified.
Several key researchers reportedly left DeepSeek between late 2025 and mid-2026 to join ByteDance, Tencent, Xiaomi and other technology companies.
With leading AI firms worldwide offering increasingly lucrative compensation packages, equity incentives have become an important retention tool.
According to the leaked transcript, Liang repeatedly described team stability as the company’s “only non-negotiable core interest,” suggesting that fundraising would allow employees to receive meaningful stock options.
Why it matters globally
The episode illustrates the delicate balance China’s leading AI startups must strike between attracting capital and maintaining secrecy around long-term strategy.
DeepSeek has become one of the world’s most closely watched AI companies, and its reported $71 billion fundraising target reflects investor confidence in China’s ability to produce a global AI leader.
At the same time, Liang’s public emphasis on AGI research over rapid commercialization appears to diverge from investors’ expectations for near-term financial returns.
That a leaked investor discussion could disrupt a multibillion-dollar financing underscores both the market’s intense interest in DeepSeek and the challenges facing frontier AI companies as they navigate fundraising, governance and long-term research ambitions.
