China’s AI sector is not short of cash. If anything, American pressure is pumping more in.
DeepSeek shows the pattern best. The Hangzhou lab closed a $7.4bn round, the biggest first-time raise by a Chinese startup, at a valuation above $50bn. It had run for three years on founder Liang Wenfeng’s own wealth and never touched outside money. The Information reports what changed his mind: Anthropic’s April preview of Mythos, a model so capable it can find and exploit software flaws.
The Mythos effect
Liang concluded that matching Mythos-class models would take a far bigger war chest of data and compute. So he raised one. DeepSeek closed the round through a partnership he controls, with a five-year lockup that keeps him firmly in charge. The irony is sharp. A US campaign of export limits and model bans, aimed at slowing China, instead pushed its strongest lab to amass the largest funding pile in its history.
A unicorn every five days
