DeepSeek Pauses Second Fundraising Round! Liang Wenfeng Upset Over Leaked Internal AI Remarks

According to sources, Chinese artificial intelligence (AI) startup DeepSeek has notified potential investors involved in its second fundraising round that it will temporarily pause the process. This comes just days after remarks widely believed to be from founder Liang Wenfeng on U.S.-China AI competition rapidly spread online, drawing market attention.

Bloomberg reported that sources revealed DeepSeek has verbally informed some potential investors that the investment agreements originally scheduled for signing in recent days will not proceed as planned. However, DeepSeek may still restart fundraising in the future.

Part of the reason for the fundraising pause stems from founder Liang Wenfeng’s dissatisfaction with the online circulation of reports about his remarks to investors during the first fundraising round regarding U.S.-China tech competition.

DeepSeek completed its first fundraising round in June, raising $7 billion.

Bloomberg stated it has not yet been able to verify the authenticity of the circulated online content. The related posts primarily involve a purported verbatim transcript of a meeting between Liang Wenfeng and an unidentified individual.

This week, Chinese media outlets including First Financial Daily reported that Liang Wenfeng mentioned in the conversation that DeepSeek currently still relies on NVIDIA (NVDA-US) chips for its AI development and believes China continues to lag behind the U.S. in AI technology maturity.

DeepSeek did not immediately respond to Bloomberg’s request for comment via email regarding the transcript and fundraising progress.

Fundraising negotiations remain highly uncertain. DeepSeek may ultimately resume this round of fundraising, but it is unclear whether the company has notified all potential investors involved in this round of the pause decision.

Notably, these fundraising talks began just weeks after DeepSeek completed its first round. The company successfully attracted heavyweight investors in its initial round, including Tencent Holdings (00700-HK) and CATL (03750-HK).

According to Bloomberg’s prior reporting, DeepSeek originally aimed to raise at least 10 billion yuan in its second round, though the final amount could increase depending on the number of participating investors.

DeepSeek’s pre-investment valuation target for this round is at least 480 billion yuan, significantly higher than the approximately $50 billion valuation during its first fundraising round.

Additionally, DeepSeek has already begun preparing for an initial public offering (IPO), potentially filing an application as early as this year, paving the way for one of the most anticipated tech IPOs in China in recent years.

As DeepSeek is viewed as a key Chinese player in the global AI race, it has attracted significant interest from potential investors.

The company’s large language model, launched last year, shocked the industry by successfully building a world-class, highly efficient AI model despite relatively limited computing resources. This breakthrough demonstrated that Chinese firms can still compete with top Silicon Valley AI companies, even under U.S. export controls on advanced chips.

Currently, DeepSeek is seeking new funding to support more aggressive expansion plans, including enhancing computing power. As the global AI race intensifies, major AI labs are investing heavily in data center infrastructure to ensure sufficient computing resources for model training and AI service support.

Founded in 2023, DeepSeek is held by Zhejiang Jiuzhang Asset Management Co., Ltd. Its first fundraising round set a new record for the largest initial funding round by a Chinese tech startup, attracting investments not only from Tencent and CATL but also from China’s National AI Industry Investment Fund. This fund is one of Beijing’s key official platforms for promoting AI industry development.

Bloomberg previously reported that DeepSeek executives told potential investors the company would prioritize breakthrough AI research over short-term commercial gains.

Liang Wenfeng reportedly committed at least once during an investor meeting that DeepSeek would continue developing open-source AI models while advancing toward achieving Artificial General Intelligence (AGI), highlighting the company’s focus on advancing AI technology rather than short-term profitability.

With the completion of the latest fundraising round, Liang Wenfeng’s personal wealth has surged. According to the Bloomberg Billionaires Index, his net worth has jumped from approximately $16.7 billion to $36 billion—more than doubling and surpassing Anthropic founder Dario Amodei and OpenAI’s Greg Brockman, making him the wealthiest AI model developer globally.