DeepSeek Founder Declares AGI Goal as State-Led $10B Funding Round Advances — First External Capital for China's Most Disruptive AI Lab

DeepSeek founder and CEO Liang Wenfeng publicly declared artificial general intelligence (AGI) as the company's primary long-term objective on May 22, 2026, as DeepSeek's first-ever external funding round advanced toward close. Bloomberg reported that Liang told prospective investors in the ongoing 70 billion yuan (~$10 billion) round that DeepSeek will prioritize groundbreaking AI research over near-term commercialization — pledging to continue releasing open-source models while pursuing AGI as its central mission. SCMP separately reported the round's valuation has reached approximately $50 billion, up from the $45 billion figure reported when the round launched in early May 2026, and sharply above the $20 billion+ estimate from weeks prior. China's national semiconductor investment fund — the China Integrated Circuit Industry Investment Fund (CICIIF, commonly known as the 'Big Fund') — is leading the round, with Tencent participating and additional strategic Chinese investors in discussions. This marks DeepSeek's first external capital raise since founding; the company had been entirely funded by its parent quantitative hedge fund High-Flyer (幻方科技) until now. The decision to accept state capital with the Big Fund as lead investor formalizes Beijing's treatment of DeepSeek as a strategic national AI asset — not merely a commercial venture. Liang noted that the primary driver for accepting funding is talent competition: DeepSeek's researchers are being aggressively recruited by better-capitalized rivals, and the funding is needed to retain the team that produced its breakthrough models. The AGI declaration carries significant geopolitical weight: it is the first time a major Chinese AI lab has publicly declared AGI as its operational target, placing DeepSeek in direct competition with OpenAI, Anthropic, and Google DeepMind at the frontier mission level — not merely on benchmark scores. Context: DeepSeek V3 (December 2024) matched GPT-4o performance at reported training costs of $5.6M. DeepSeek R1 (January 2025) triggered a $600B Nvidia market cap drop by demonstrating competitive reasoning at a fraction of US compute costs. DeepSeek V4 (April 24, 2026) was the first major Chinese frontier model running inference natively on Huawei Ascend chips, validating the domestic AI stack. The Big Fund's AGI investment — alongside ByteDance's $5.6B Huawei chip order, Alibaba's Zhenwu M890 launch, and Cambricon's first profitable year — completes a week in which China's domestic AI ecosystem received its highest-level state and commercial endorsements simultaneously.