DeepSeek, the AI lab that rattled Silicon Valley with a low-cost model trained on a fraction of the compute, has completed its first ever funding round. The Wall Street Journal and The Information report it raised more than 50 billion yuan — approximately $7.4 billion — valuing it above $50 billion. The lead investors are China's National AI Industry Investment Fund and the China Integrated Circuit Industry Investment Fund, known as Big Fund III. Tencent, JD.com, CATL and NetEase also participated. The US banned its own investors from the round. China filled every seat.
This is not abstract. DeepSeek is optimising its models to run on Huawei Ascend chips, cutting dependency on Nvidia hardware the US has already sanctioned. Every Western AI company — OpenAI, Anthropic, Google DeepMind — now faces a state-funded, open-weight competitor that costs users nothing to deploy. UK and US developers building on proprietary Western APIs are pricing in a cost that their Chinese counterparts have just been handed for free by Beijing.
The Law of the Narcissist says the system presents itself as the only credible option. Western AI labs spent two years telling the market that frontier intelligence required $100 billion in compute. DeepSeek disproved it publicly. Now the state has funded the proof. Anyone still assuming Silicon Valley holds the monopoly on intelligence infrastructure is not reading the room. They are performing confidence for investors who need to believe the moat still exists.
The Sovereign One does not pick a flag. They ask: which open-weight model runs cheapest on commodity hardware, and which closed-model provider just lost their pricing power? Step 6 — the Internal Intelligence Agency — means you audit your own AI cost stack before your competitors do it for you. The lab that just raised $7.4 billion from a government is not competing on returns. It is competing on dominance.
Want the full steps? Start with The Money Bible
