Is DeepSeek Transforming? First Funding Talk Shows $100B Valuation and $3B Raise

According to reports from The Information and Reuters, DeepSeek is in talks with investors to raise at least $3 billion at a valuation of roughly $100 billion.

Last year DeepSeek gained attention with its high‑performance, low‑cost inference model DeepSeek R1, which sparked market volatility and a reassessment of large‑model technology and AI infrastructure. Despite its impact, the startup has never taken external venture capital; its early compute resources and R&D funding came entirely from its parent, the quant‑hedge fund Fantasia Quant.

DeepSeek has previously turned down multiple financing offers from top Chinese venture firms and tech giants. The current fundraising effort highlights the massive capital required to develop and run cutting‑edge large models and emerging AI agents.

Recent job postings show DeepSeek hiring a data‑center operations engineer in Inner Mongolia, marking its first direct recruitment for infrastructure staff and indicating a move toward a "heavy‑asset" model that relies on owning data‑center capacity to lower long‑term compute costs.

The company is also preparing to launch its next‑generation model, DeepSeek‑V4, which will further increase funding needs due to higher compute demands.

DeepSeek’s APIs are priced at a fraction of those offered by U.S. peers such as OpenAI and Anthropic. Sufficient external capital would allow the firm to maintain this aggressive pricing strategy in the global AI price war.

If the deal proceeds at the reported terms, the equity dilution would be minimal, leaving the parent Fantasia Quant and the founding team with full control, making a shift to a closed‑source model unlikely.

Talent turnover is evident: core researcher Luo Fuli has joined Xiaomi to lead its AI business, and post‑95 researcher Guo Dayi, a key contributor to DeepSeekMath, DeepSeek‑V3, and DeepSeek‑R1, has moved to ByteDance’s Seed team as an agent lead (L8). Media reports claim Guo’s total compensation approaches ¥100 million per year, a claim ByteDance’s VP Li Liang partially refuted, explaining that salaries are composed of cash, ByteDance stock options, and Doubao options, with full vesting over four years and potential multi‑hundred‑million‑yuan payouts for top performers.

The strong pull of AI talent by major tech firms reflects a broader industry shift from early‑stage large‑scale research toward engineering validation and ecosystem construction. Future financing could give DeepSeek more resources for commercial deployment.