According to Bloomberg on May 22, DeepSeek is currently conducting its first external fundraising round since inception, aiming to raise roughly 70 billion yuan (approximately $10 billion). This target surpasses earlier market expectations of $7 billion, while the pre-money valuation stands at about $45 billion — potentially making it the largest initial funding round in Chinese tech startup history. Sources familiar with the matter revealed that founder Liang Wenfeng emphasized during meetings with potential investors that DeepSeek’s primary objective remains achieving Artificial General Intelligence (AGI), and the company will continue releasing its AI models under open-source licenses rather than prioritizing short-term commercial gains. Potential investors include the National AI Industry Investment Fund, Tencent, IDG Capital, and Monolith Capital; the former reportedly intends to contribute around 10 billion yuan, while Liang himself may invest roughly 20 billion yuan from his personal funds.
This marks DeepSeek’s first external fundraising effort in three years of operation — previously, the company relied solely on financing from Liang’s quantitative hedge fund High-Flyer. The shift stems from soaring model training costs exceeding what a single hedge fund can sustain. In April this year, DeepSeek unveiled V4-Pro, an open-source MoE model boasting 1.6 trillion parameters, requiring unprecedented computational resources; both V4-Pro and V4-Flash are released under permissive open-source licenses. Discussions regarding the funding round remain ongoing, so final participants and amounts could still change. For comparison, OpenAI is currently valued at $500 billion, while Anthropic’s valuation sits at $350 billion — DeepSeek’s $45 billion valuation thus represents merely one-tenth of those figures.