DeepSeek founder invests $2.7B personally, V4.1 due in June, valuation $51.5B
DeepSeek's first funding round just got a major twist. According to The Information, founder Liang Wenfeng plans to personally contribute up to 20 billion RMB (roughly $2.7 billion) to the round, taking about 40% of the shares. That means he's replacing the previously rumored state-backed mega-fund as the lead investor in this $7.35 billion (50 billion RMB) financing, pushing China's National Integrated Circuit Industry Investment Fund to second place. The post-money valuation has also jumped from the $45 billion speculated a couple days ago to over $51.5 billion.
With the capital structure clearer, DeepSeek's commercialization plans are coming into focus:
- V4.1 core update: Set for June, it will beef up enterprise tools, natively support the Model Context Protocol (MCP), and add image and audio understanding capabilities (though generation remains text-only).
- Pay and options defense: The fresh funds will go toward expanding compute power, raising salaries, and issuing stock options. This is a direct response to talent poaching from tech giants — Luo Fuli, a key contributor to DeepSeek V3, has jumped to Xiaomi to lead its AI department, and another core researcher, Guo Daya, was poached by ByteDance with a fat paycheck.
Liang Wenfeng has gone from a "token follow-on investor" to writing the biggest check, securing maximum control over the company. But with skyrocketing compute costs and a brain drain, DeepSeek can no longer afford its "no revenue pressure" pure research model. The startup now has to figure out how to sustain itself.