DeepSeek is seeking external funding for the first time, quickly triggering a competitive pursuit from major domestic technology firms. According to a Wednesday report by tech media The Information, four people familiar with the matter disclosed that TENCENT and Alibaba are engaged in talks regarding an investment in DeepSeek. Driven by strong market interest from preliminary discussions, DeepSeek's target valuation has been raised from an initial minimum of $10 billion to over $20 billion, with the fundraising scale potentially expanding accordingly. This funding round signifies a major shift in the long-held stance of DeepSeek's founder, Liang Wenfeng, against introducing external capital. As AI model training costs continue to escalate, the competitive landscape intensifies, and pressure mounts from delays in the development of its next-generation flagship model, this startup, which evolved from the AI lab under private equity firm幻方科技, is in urgent need of "ammunition" to replenish computing power reserves and retain core talent.
**Valuation Benchmarks and Market Pricing Debate**
The report stated that, according to one person directly involved, DeepSeek is using Kimi as a partial benchmark for its valuation—the latter is currently finalizing a new funding round with a valuation of $18 billion. Other data suggests DeepSeek's actual value may be higher. Competitors in the same field, MiniMax and智谱, both listed in Hong Kong in January of this year with valuations under $10 billion at the time; as of now,智谱's market capitalization has surpassed $50 billion, and MiniMax exceeds $30 billion. However, insiders pointed out that because DeepSeek has long operated more as an AI lab dedicated to open-source technology rather than a profit-driven commercial company, its appropriate valuation has sparked extensive debate within China's tech and venture capital circles. Unlike MiniMax and智谱, DeepSeek has not yet generated significant revenue to date, its models are entirely open-source, and its consumer-facing chatbot is free to use. Negotiations are ongoing, and both the final valuation and fundraising scale remain subject to change.
**Strategic Moves by Domestic Giants**
The potential involvement of TENCENT and Alibaba is underpinned by clear strategic considerations. Both companies have previously invested in several leading Chinese AI startups, including智谱, MiniMax, and月之暗面. Taking a stake in DeepSeek allows them to hedge risks while they concurrently advance their own proprietary AI models and opens a direct channel for deeper collaboration with DeepSeek. DeepSeek's R1 model, launched early last year, caused a global stir by achieving performance close to top US systems at a significantly lower training cost. At that time, TENCENT, Alibaba, and ByteDance all integrated the R1 model into their existing products. Among them, TENCENT, which has been relatively slower in developing its own cutting-edge foundational models, finds the strategic significance of investing in DeepSeek particularly pronounced.
**Competitive Pressures and Technical Challenges**
DeepSeek's search for external funding reflects a competitive environment that has drastically changed. The pace of AI model iteration is accelerating, training and operational costs are rising, and well-funded tech giants from both China and the US are increasingly pressuring startups. DeepSeek has seen some core researchers depart for major domestic firms in recent months. Notably, Luo Fuli has officially joined Xiaomi as the head of the MiMo team. On the product development front, DeepSeek originally planned to release its next-generation flagship model, V4, in February of this year, but this timeline has been postponed multiple times due to engineering challenges. The infusion of external capital will directly provide DeepSeek with the financial resources needed to procure computing power and stabilize its talent pool.
Comments