DeepSeek Closes $7.4 Billion Debut Funding Round Under Founder-Control Structure
Chinese AI startup DeepSeek has completed its first-ever external funding round, raising more than RMB 50 billion yuan (US$7.4 billion) through an unconventional deal structure designed to preserve founder control — a milestone that underscores both the company's rising stature and the intensifying pressures it now faces.
According to a report by The Information published on June 15, 2026, the funding round values DeepSeek at more than $50 billion. The deal was reported by journalists Jing Yang, Qianer Liu, and Juro Osawa, citing two people with direct knowledge of the matter.
Rather than channeling investor capital directly into DeepSeek, the structure requires all investors — except one — to contribute to a limited partnership controlled by CEO Liang Wenfeng, effectively insulating him from external shareholder influence. All investors in that partnership are subject to a five-year lockup period, barring them from selling their stakes. Such restrictions are rare in venture capital, where secondary-market trading of shares in high-profile startups is commonplace.
The sole exception is China's National Artificial Intelligence Industry Investment Fund, which invests RMB 1 billion yuan directly into DeepSeek, is exempt from the lockup, and holds voting rights at the company. All other external investors receive no voting rights, though they are granted access to privileged financial disclosures and priority participation rights in future funding rounds.
Liang himself is the largest single contributor, committing RMB 20 billion yuan to the round. Tencent follows with RMB 10 billion yuan. Contemporary Amperex Technology, widely known as CATL, is investing RMB 5 billion yuan, while JD.com, NetEase, and venture firm IDG Capital each contribute RMB 3 billion yuan. None of the named parties immediately responded to requests for comment.
The fundraising marks a significant strategic pivot for DeepSeek, which was founded in 2023 as an AI division of Liang's hedge fund High-Flyer Capital Management and had previously operated without any outside capital. Its research-first ethos was widely credited as a key factor behind the global success of its R1 model, released in early 2025. However, that model has proven difficult to sustain: rising computing costs and an increasingly competitive talent market have forced the company's hand.
Talent attrition has already become visible. Luo Fuli, a key contributor to DeepSeek's V3 model, departed to lead Xiaomi Corporation's AI division, while researcher Guo Daya joined ByteDance earlier in 2026 at a significantly higher compensation level.
The deal's architecture — combining state capital with strict investor vetting and an unusually long lockup — reflects DeepSeek's carefully managed position at the center of China's AI ambitions. How the company balances state influence, founder autonomy, and the commercial pressures now bearing down on it will be closely watched across the global AI industry.
Related Coverage:
DeepSeek Weaponizes Compute Costs to Force Global AI Consolidation in 2026