DeepSeek Takes Outside Money for the First Time — at $10B+

DeepSeek is seeking roughly $300 million in its first-ever external funding round, at a valuation north of $10 billion — a major shift for a lab that had turned away virtually every investor since its founding by quant hedge fund High-Flyer Capital Management.

Founder Liang Wenfeng long resisted outside capital to preserve DeepSeek's research-first culture. That stance is cracking under real pressure: explosive global demand for its models and API has triggered outages and exposed compute limits. The lab needs more GPUs, servers, and inference capacity — and self-funding isn't scaling fast enough. Chinese domestic investors are the likely participants; U.S. firms face regulatory headwinds and national security scrutiny.

The raise is striking context against DeepSeek's own origin story — the R1 model reportedly trained for just $5–6 million on export-restricted H800 chips, sending shockwaves through Silicon Valley in early 2025. Efficiency can only go so far when usage is global.

Why it matters: DeepSeek entering the funding game signals that even efficiency-obsessed labs can't outrun infrastructure costs at scale — and that Chinese AI is consolidating serious capital.