Chinese AI startup DeepSeek raised $15 billion ahead of a potential IPO. The company reached a $74 billion valuation after a funding round that ran from July through this month. Big names like Tencent and Contemporary Amperex Technology joined in. DeepSeek also hired Citic Securities to advise on its future.
DeepSeek has been on investors’ radar since impressing Silicon Valley with a high-performance, low-cost AI model last year, setting off a wave of fundraising as a growing group of Chinese labs challenge U.S. dominance in AI.
Prior to the latest round’s close, DeepSeek was valued at around $74 billion, people familiar with the process said. China’s biggest battery maker, Contemporary Amperex Technology, and tech giant Tencent both participated in the round after having invested in a previous fundraise, the people said.
According to some of the people, DeepSeek’s latest round took longer to complete than other recent AI fundraises, running from July through this month as the talks drew in additional investors along the way.
Other existing shareholders, including investment firms Monolith Management and Shixiang, also injected new capital into the AI startup, The Wall Street Journal has reported.
New backers include Stony Creek Capital, a semiconductor-focused private-equity firm, people with knowledge of the matter said. Legend Capital, backed by computer and server maker Lenovo, as well as carmaker Geely, also invested in the round, the people said.
DeepSeek has hired state-owned investment bank Citic Securities to advise on its IPO, people familiar with the matter said. In September, it recruited Yan Wentao, a former venture capitalist, to serve as its chief financial officer, the people said.
Bloomberg and Chinese financial magazine Caijing had previously reported DeepSeek’s latest financing round.
The fundraising efforts come as AI companies’ profitability draws heightened scrutiny from some investors who question whether massive capital spending will translate into sustainable profits and if soaring valuations are justified.
DeepSeek’s founder, Liang Wenfeng, told investors in September that the company’s annualized revenue run rate—a key metric for subscription-based services—had hit $1 billion, according to people familiar with the matter.
The company is spending on domestic computing infrastructure to reduce its reliance on foreign chips. While it still relies on Nvidia’s chips for the most complex AI training, DeepSeek is teaming up with Huawei Technologies to pivot more training workloads on domestic chips.
Write to Raffaele Huang at raffaele.huang@wsj.com and Tracy Qu at tracy.qu@wsj.com