DeepSeek, the Hangzhou lab that startled the industry early last year, is close to raising about 50 billion yuan, or roughly $7.4 billion, in one of the largest startup financings China has seen. The deal, first reported by Bloomberg on June 3, would value the company at around $45 billion. That is a smaller raise than the $10 billion the company had originally targeted, but it still ranks among the country's biggest.

Who is backing it

The investor group includes Tencent and China's government-backed National Artificial Intelligence Industry Investment Fund, with founder Liang Wenfeng putting in his own money as well. The mix of a major platform company and a state fund is a fair summary of where China's AI ambitions sit right now, part commercial, part national project.

A different bet from its rivals

What makes the round worth watching is the strategy attached to it. DeepSeek's leadership has said it will prioritise groundbreaking research over short-term commercialisation, keep releasing open-weight models, and continue working toward artificial general intelligence. That is close to the opposite of the path its largest competitors are walking. Anthropic has filed for a possible trillion-dollar IPO, and OpenAI has spent the year building out revenue lines and enterprise products. DeepSeek is raising private money specifically so it does not have to do that yet.

The open-weight commitment is the throughline. DeepSeek made its name in early 2025 by matching the performance of leading American models while using far fewer Nvidia chips, a result that rattled assumptions about how much compute frontier AI really requires. Releasing weights openly has since become a recognisable Chinese strategy, seen again in MiniMax's recent open-weight M3 model.

The pressure behind the raise

The timing is not incidental. After two years of enormous infrastructure spending, AI companies everywhere are under pressure to show tangible results, and Chinese firms face the added constraint of restricted access to the most advanced chips. A research-first posture is easier to maintain with $7.4 billion in the bank than without it. Whether DeepSeek can keep funding open models while chasing AGI, without the recurring revenue its rivals are racing to lock in, is the question this round leaves open. For now it has bought itself room to try.

Sources

  1. i. www.pymnts.com
  2. ii. www.bloomberg.com

Commentarii · 0

Add · a · Comment