Summary of Key Points
DeepSeek has just completed its first external equity financing, raising 51 billion yuan with a valuation of approximately 325-350.9 billion yuan (ranked among the top AI companies in China). Founder Liang Wenfeng made it clear during a 4-hour meeting that the company’s focus is not on short-term commercial gains (such as going public or making quick money), but rather on achieving AGI (Artificial General Intelligence) as its long-term goal. Management at DeepSeek is driven by vision, and team stability is considered essential. The company believes that open-source development and commercialization are not contradictory; they aim to earn only reasonable profits. The current biggest challenge is the lack of computing power, but they are optimistic about the potential for domestic chips to fill this gap. The selection process for investors was rigorous, with priority given to those who share the same long-term interests as the company. Liang Wenfeng personally invested 20 billion yuan, and companies like Tencent and CATL participated in the financing, although some institutions did not invest due to insufficient information.
Detailed Analysis
1. Financing: More Than Just Money – Choosing Long-Term Partners
DeepSeek’s approach to fundraising is not about accepting any offer; instead, they are looking for partners with aligned values and goals. For instance, Liang Wenfeng invested 20 billion yuan himself, while Tencent contributed 10 billion yuan and CATL 5 billion yuan, along with other investors such as NetEase and JD.com. However, there were strict requirements for potential investors, including a commitment to a long-term partnership rather than focusing solely on short-term returns. A domestic investor mentioned that despite recognizing DeepSeek’s world-class technology, they did not invest because the company did not provide clear details on how it planned to generate revenue or exit the market.
Why such selectivity? Liang Wenfeng aims for AGI, which requires significant long-term investment; if investors are eager for quick returns, they might push the company to pursue short-term projects that could undermine its long-term goals. Therefore, financing is more like finding partners who will join them on this journey rather than simply providing financial support.
2. Company Positioning: AGI as the Core Focus
Liang Wenfeng repeatedly emphasizes that DeepSeek’s purpose is not to go public or maximize profits, but to create technologies that truly benefit humanity. Last year, when AI became popular, many companies focused on capturing consumer traffic (e.g., developing apps and monetizing users), but DeepSeek stayed true to its vision. He views these efforts as secondary to the more substantial goal of achieving AGI.
How does this differ from other companies? While others were busy acquiring users, DeepSeek was focused on advancing core technology, believing that these activities are less crucial than developing fundamental AI capabilities. Liang Wenfeng believes that by strengthening the underlying technology, future applications will naturally emerge.
3. Management: Vision-Driven, Without Strict KPIs or Overwork
DeepSeek’s management style is unconventional: there are no strict performance indicators (KPIs), and employees spend about half of their time on independent research projects. The team works in two parallel tracks—ones assigned by the company (e.g., developing new models) and others initiated by the employees themselves, with the former accounting for no more than 50% of their time. There is also little overtime, as a relaxed environment is essential for innovation.
The logic behind this approach? According to Liang Wenfeng, management should be based on shared vision rather than rigid rules. Even when the company was unknown and financially struggling two years ago, a team of dedicated individuals achieved remarkable results because they believed in the potential of AGI to change the world. Team stability is key; by offering equity to employees after financing, DeepSeek reduced the risk of talent loss. He believes that as long as the core team remains committed, success in achieving AGI is guaranteed, and resources will follow.
4. Open-Source and Commercialization: Incompatible but Not Opposite
There has been confusion about how DeepSeek, being open-source (free model distribution), can generate revenue. Liang Wenfeng explains that these two goals are not mutually exclusive. For example, deploying the company’s models requires substantial computing power and expertise, making it accessible only to a limited number of companies. DeepSeek’s pricing strategy aims to recoup server costs within 10 months; they set prices reasonably to ensure profitability without excessive profit-seeking. When one of their models was initially priced too high, Liang Wenfeng lowered the price by half, which was well-received by the team.
How does this balance openness with profitability? By making their technology accessible and profitable, DeepSeek encourages more companies to build applications based on their models. Given the vast potential of the AI market (potentially accounting for 10% of GDP), it’s better to share the benefits rather than monopolize them.
5. The Challenge of Computing Power: Domestic Chips as a Solution
Liang Wenfeng highlights that the biggest obstacle in AI development is the lack of computing power, particularly compared to the United States. This limitation affects the progress of various research directions. He believes that larger models yield better results, but current computational constraints limit their scale. DeepSeek plans to build its own large-scale computing clusters and will consider developing its own chips if it proves cost-effective. He is optimistic about domestic chip technology, suggesting that the competitiveness of NVIDIA’s CUDA ecosystem may decline in the coming year, making domestic alternatives more viable. Additionally, he notes that the current fragmentation of resources among domestic AI companies will eventually lead to a consolidation, reducing the need for so many teams working on basic models and potentially lowering profit margins.
6. DeepSeek’s Unique Approach
With tens of billions in funding and a valuation exceeding 300 billion yuan, DeepSeek has chosen a less commercial path, focusing on AGI rather than short-term gains. This approach may not be the most profitable, but it makes them one of the most idealistic companies in China’s AI industry. For consumers, this means the potential for more accessible and beneficial AI technologies, rather than being dominated by a few large corporations.