Chinese artificial intelligence start-up DeepSeek has begun raising external capital for the first time, marking a pivotal moment in the country's rapidly evolving AI landscape. According to a report from the South China Morning Post, the funding round is primarily designed to establish a market benchmark for the company's valuation, as competition for AI talent intensifies across China.
The limited funding round is intended to provide employees with greater clarity on the value of their stock options. By doing so, DeepSeek hopes to limit departures to well-funded competitors that are aggressively poaching skilled professionals. The report cited unnamed investors based in Hangzhou and Beijing, indicating that the move comes amid a broader talent war in the AI sector.
Background on DeepSeek and High-Flyer
DeepSeek is a Hangzhou-based AI start-up founded by Liang Wenfeng, who also runs High-Flyer, one of China's most profitable hedge funds. High-Flyer was the second most profitable hedge fund in China last year, according to industry figures, which means DeepSeek is not currently in urgent need of funding. Instead, the funding round serves strategic purposes: it establishes a valuation that can be used for future financing and employee retention.
The start-up is looking to sell no more than 3 percent of its equity in this initial round. It prefers 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. This approach suggests that DeepSeek values long-term partnerships over pure financial investment. Private, market-oriented funds are reportedly more likely to be included in subsequent rounds, given that the capital-intensive nature of AI development will require further financing down the road.
DeepSeek's parent company High-Flyer has been a dominant force in the Chinese quantitative trading space. The hedge fund's success has provided DeepSeek with a strong financial cushion, allowing it to focus on long-term AI research rather than immediate profitability. However, the AI industry is notoriously expensive, requiring massive computational resources and talent acquisition, which makes external funding eventually necessary.
The Chinese AI Landscape and Talent Competition
China's AI sector has been growing rapidly, with both state-backed initiatives and private enterprises vying for dominance. Companies like Baidu, Alibaba, Tencent, and ByteDance have invested heavily in AI research and development, while newer start-ups like DeepSeek are emerging with specialized focuses. The competition for top AI talent has become fierce, with salaries soaring and stock options being a key retention tool.
DeepSeek's decision to raise a limited amount of capital to set a valuation is a strategic move to prevent talent drain. Employees who hold stock options are more likely to stay if they have a clear understanding of the value of their shares. This is particularly important in a market where well-funded competitors can offer higher cash compensation or more lucrative equity packages.
The talent war is not limited to China; globally, AI experts are in high demand. However, in China, the situation is exacerbated by government policies that encourage AI development and by the presence of state-backed funds that can invest without the same profit pressures as private firms. DeepSeek's preference for state-backed investors aligns with this trend.
Access to Advanced Chips: A Critical Constraint
While capital is not a primary concern for DeepSeek at the moment, access to advanced chips is. Liang Wenfeng has stated that the main constraint for DeepSeek is not capital but the availability of advanced semiconductors, due to US export restrictions. The United States has imposed strict controls on the export of high-performance AI chips to China, targeting companies like Nvidia that produce chips used for training large language models.
These restrictions have forced Chinese AI companies to seek alternative sources or develop their own chips. DeepSeek's choice of investors that can provide AI infrastructure suggests that it is looking for partners who can help secure access to computational resources, including data centers and chip supplies. This is a critical factor for any AI start-up that needs to train increasingly large models.
The US export controls have accelerated China's push for self-reliance in semiconductors, but progress has been slow. Chinese companies are investing in domestic chip design and manufacturing, but they still lag behind industry leaders. For DeepSeek, forming partnerships with state-backed entities that can offer access to restricted technology or alternative computing solutions could be essential.
Valuation and Investor Interest
Reports have varied on DeepSeek's potential valuation. The Information earlier reported that DeepSeek was aiming to raise at least $300 million at a valuation of at least $10 billion. It also said that Tencent Holdings and Alibaba Group were in discussions to invest at a valuation of over $20 billion. However, one investor who spoke to the Post estimated that DeepSeek's valuation could exceed $100 billion, a figure that underscores the high expectations for the company.
The wide range in valuation estimates reflects the uncertainty in the AI market. While some investors are willing to bet on DeepSeek's potential, others are more cautious. The company's limited equity sale — only 3 percent — suggests that it is not looking to give away too much control at an early stage. By setting a high valuation, DeepSeek can also use it as a benchmark for future rounds and for employee stock compensation.
Tencent and Alibaba are among the largest tech companies in China and have significant AI ambitions themselves. Their interest in DeepSeek indicates that they see the start-up as a potential partner or acquisition target. However, DeepSeek may prefer to remain independent for the time being, leveraging its parent company's financial strength.
DeepSeek's Technology and Model
DeepSeek is known for its large language models, which compete with those from Western companies like OpenAI and Google, as well as Chinese rivals like Baidu's ERNIE Bot. The company has released several models that have achieved impressive benchmarks in natural language processing tasks. Its approach emphasizes efficiency, aiming to achieve high performance with fewer computational resources than some competitors.
The company's name, DeepSeek, reflects its mission to dive deep into AI research. It has published papers on topics such as reinforcement learning, model architecture, and training techniques. DeepSeek is also active in the open-source community, releasing some of its models for research purposes, which helps attract top talent who value openness.
Despite its promising technology, DeepSeek faces significant challenges. The AI field is dominated by giants with massive budgets and established ecosystems. However, DeepSeek's agility and focus could allow it to carve out a niche, particularly in specialized applications where it can outperform general-purpose models.
Regulatory and Geopolitical Context
China's regulatory environment for AI is evolving. The government has introduced guidelines for generative AI services, requiring companies to obtain licenses and ensure content safety. DeepSeek must navigate these regulations while also dealing with the geopolitical tensions between the U.S. and China.
The US export controls on AI chips are part of a broader effort to limit China's technological advancement. However, they have also spurred Chinese innovation in areas like chip design, alternative computing architectures, and software optimization. DeepSeek's reliance on advanced chips makes it vulnerable, but its ability to obtain access through strategic partnerships could mitigate this risk.
State-backed investors are likely to view DeepSeek as a strategic asset, particularly if it can achieve breakthroughs in AI that are important for national competitiveness. This aligns with China's goal to become a world leader in AI by 2030. Government guidance funds and tech platforms affiliated with the state can provide not only capital but also political support and access to key resources.
Future Prospects
DeepSeek's first funding round is a small but significant step. By limiting the equity sale, the company is signaling confidence in its future valuation growth. The focus on talent retention through stock option clarity is a smart move in a competitive market. However, the real test will be whether DeepSeek can scale its technology and secure the necessary computational resources to train larger models.
The AI industry is moving at a breakneck pace, with new models and breakthroughs emerging regularly. DeepSeek will need to continue innovating to stay relevant. Its parent company's profitability provides a buffer, but eventually, DeepSeek will require substantial external funding to compete with the likes of OpenAI, Anthropic, and Google.
The involvement of Tencent and Alibaba, if confirmed, would give DeepSeek access to vast resources, including cloud computing, data, and distribution channels. But it could also lead to conflicts of interest, as these companies have their own AI products. DeepSeek may prefer to maintain its independence and choose investors that are less competitive.
Ultimately, DeepSeek's success will depend on its ability to attract and retain talent, acquire necessary hardware, and develop models that outperform or differentiate from competitors. The funding round is a first step, but the journey ahead is long and fraught with challenges. The broader context of US-China tech competition will continue to shape the landscape, favoring companies that can navigate geopolitical risks while delivering cutting-edge technology.
Source: Silicon UK News