Chinese AI start-up DeepSeek has begun raising external capital for the first time, largely to establish a market benchmark for its valuation at a time of aggressive poaching of AI talent, the South China Morning Post reported. The limited funding round is intended to give employees greater clarity on the value of their stock options, in order to limit departures to well-funded competitors, the paper said, citing unnamed investors from Hangzhou and Beijing.
Image credit: Unsplash
First Funding Round
The start-up is not currently in urgent need of funding, as its parent company, hedge fund High-Flyer, was the second most profitable in China last year, according to industry figures. Hangzhou-based DeepSeek is looking to sell no more than 3 percent of its equity, preferring to sell to state-backed and industrial investors, including local government guidance funds and affiliated tech platforms that can provide strategic resources such as AI infrastructure, the Post reported, citing three investors. Private, market-oriented funds are reportedly more likely to be included in subsequent rounds, as the capital-intensive nature of AI development requires further financing.
Infrastructure Access
High-Flyer founder Liang Wenfeng has said that DeepSeek’s main constraint is not access to capital, but access to advanced chips, amid US export restrictions. The Information earlier reported that DeepSeek was aiming to raise at least $300 million (£222m) at a valuation of at least $10bn, and that Tencent Holdings and Alibaba Group were in discussions to invest at a valuation of over $20bn. One of the investors who spoke to the Post said DeepSeek’s estimated valuation could exceed $100bn.
DeepSeek, a relatively new entrant in the Chinese AI landscape, has quickly gained attention for its advanced large language models and competitive performance against global players like OpenAI. The company was founded in 2023 by Liang Wenfeng, a prominent figure in the Chinese quantitative finance world through High-Flyer, which manages billions in assets. DeepSeek's AI models have shown exceptional capabilities in reasoning, mathematics, and code generation, often rivaling or surpassing those of larger, better-funded Western companies. This rapid progress has made DeepSeek a prime target for talent poaching, as competitors seek to lure away its key engineers and researchers with lucrative offers.
The funding round is a strategic move to retain its workforce. By offering employees a clearer picture of their stock option value through an external valuation benchmark, DeepSeek hopes to reduce turnover. The company is currently valued at around $10 billion, but the potential inclusion of Tencent and Alibaba could push that valuation to $20 billion or more. Some investors believe the valuation could eventually exceed $100 billion, reflecting the immense potential of generative AI in China's market.
DeepSeek's reliance on High-Flyer for funding has allowed it to focus on research without the pressure of external investor expectations. However, the US government's export controls on advanced semiconductors, particularly Nvidia's A100 and H100 chips, have posed significant challenges. China's AI firms have turned to stockpiling chips and developing domestic alternatives, but the gap in computing power remains a critical bottleneck. DeepSeek has been exploring ways to optimize its models for lower-precision chips and has collaborated with Chinese semiconductor companies to ensure a stable supply of AI accelerators.
The fundraise also signals a broader trend in China's AI sector, where start-ups are increasingly turning to state-backed investors and strategic partners to secure not just capital, but also access to cloud computing, data centers, and other infrastructure. This approach aligns with Beijing's push for technological self-reliance and the development of a domestic AI ecosystem independent of US influence. DeepSeek's preference for government guidance funds and tech platforms ensures it can leverage national resources while maintaining operational independence.
Meanwhile, the global AI race continues to intensify. DeepSeek's open-source models have been widely adopted by developers and researchers, giving it a strong foothold in the global community. The company's novel architecture, including mixture-of-experts (MoE) designs and sparse attention mechanisms, has allowed it to achieve competitive performance with fewer parameters and lower computational costs. These innovations have made DeepSeek a case study in efficient AI development, demonstrating that resource-constrained firms can still achieve breakthroughs with clever engineering and talented teams.
Looking ahead, DeepSeek is likely to face increasing competition from domestic rivals like Baidu, Alibaba, and ByteDance, as well as from international players like OpenAI, Google, and Anthropic. The influx of capital from the funding round will enable DeepSeek to scale its infrastructure, hire more researchers, and explore new applications in healthcare, finance, and education. However, the company's long-term success will depend on its ability to navigate geopolitical tensions, secure advanced chips, and retain its core team of experts.
The limited equity sale also reflects a cautious approach to dilution, as High-Flyer maintains majority control. This allows DeepSeek to preserve the agility and research-focused culture that has driven its early success. As the first external funding round closes, all eyes will be on how DeepSeek deploys its new resources and whether it can emerge as a leading force in the global AI revolution.
Source: Silicon UK News