DeepSeek nears $45B as China's Big Fund joins AI stack

deepseek is reportedly nearing a $45bn valuation, up from around $20bn only weeks ago, in its first major fundraising round, according to ft

the round may be led by china’s “big fund”, the country’s biggest state-backed semiconductor investment vehicle

that matters because the big fund usually backs strategic chip companies, not ai model labs. it has previously backed smic (china’s top chip foundry, basically its closest tsmc equivalent) and ymtc (china’s leading nand flash memory-chip maker). now it may back deepseek, moving from chips into the model layer

tencent is also in talks to invest. liang wenfeng, deepseek’s founder, may personally invest in the round. he reportedly controls 89.5% of the company through personal holdings and affiliated entities

backing from the big fund would tie deepseek more directly into china’s national ai + semiconductor self-sufficiency strategy. deepseek’s latest v4 model was optimized for inference on huawei ascend 950pr chips, which fits the broader china strategy: domestic models running well on domestic chips

that is the real story here. china is trying to connect the stack:

• chips – huawei ascend
• capital – state-backed semiconductor fund
• models – deepseek
• distribution – tencent and domestic platforms
• policy – self-sufficiency under us export controls

the ai race is no longer just about who has the best model. it is about whose models run best on whose chips, whose capital funds the stack, and whose platforms distribute it

deepseek is not just another chinese ai lab anymore. it is becoming the model layer of china’s national ai stack