DeepSeek has paused its $1.5 billion funding round following founder Liang Wenfeng's concerns over media coverage of his comments.
Chinese AI startup DeepSeek has paused its second funding round, which was set to raise at least $1.5 billion. This suspension follows concerns raised by founder Liang Wenfeng regarding media representation of his comments during discussions with investors, as reported by Bloomberg.
The company has informally informed some potential backers that it will not finalize agreements in the immediate future, although it might revisit the fundraising process later.
Meeting Insights
A significant meeting took place on May 20, lasting about four hours. Portions of the meeting minutes were leaked to Yicai, which shared excerpts on July 23, but neither DeepSeek nor Liang has confirmed their authenticity.
During the meeting, Liang noted that China's main disadvantage in AI compared to the U.S. is not a lack of talent, but the restricted access to computing resources. He emphasized the challenges in acquiring advanced graphics processors and the comparatively lower investment in the sector in China.
The leaked minutes suggest that DeepSeek is trailing behind leading U.S. competitors by 12 to 18 months, despite achieving similar results with only about 5% of the processing power used by these rivals. The startup aims to close this gap to within three to six months.
Furthermore, DeepSeek is working with Huawei to enhance its models for Ascend chips, which would lessen its dependence on Nvidia's CUDA software. Liang mentioned that while the V3 model runs on American chips, it employs DeepSeek's own TileLang compiler, reducing reliance on CUDA. He pointed out that there is potential for the Chinese chip industry to replace imports, although current production volumes pose challenges.
Funding and IPO Plans
In the halted funding round, DeepSeek aimed to secure a minimum of 10 billion yuan ($1.5 billion), with a pre-deal valuation approximated at 480 billion yuan ($71 billion).
The startup's previous funding round concluded in late May, raising over 50 billion yuan ($7.4 billion) at a valuation of about 367.5 billion yuan ($54.3 billion). Notable investors included Tencent, CATL, NetEase, JD.com, IDG Capital, and the National AI Industry Investment Fund, with Liang himself making the largest single investment of 20 billion yuan.
Simultaneously, DeepSeek is preparing for an initial public offering, potentially submitting its application this year. However, Liang remarked in the leaked transcript that considerations about going public or maximizing profits were not initial motivations for the project. He noted that the early employees did not approach their work with such concerns.
Additionally, reports from early July indicated that DeepSeek is in the process of developing its own processor for its AI models.
