AI Author:EqualOcean News , Leci Zhang Editor:Yiran Xing Updated 2 hours ago (GMT+8)

On April 24, multiple media outlets confirmed that the Chinese large-scale model company DeepSeek (深度求索) has officially launched its first external equity funding round. Currently, Alibaba (阿里巴巴) and Tencent (腾讯) are in talks for a joint investment in DeepSeek, with the total investment expected to reach $1.8 billion, targeting a post-money valuation of $20 billion.

Deepseek

Although rumors previously suggested that Shunwei Capital (顺为资本), backed by Xiaomi (小米), intended to participate, only Alibaba (阿里巴巴) and Tencent (腾讯) are currently identified as clear participants. This move is seen as another critical strategic move by tech giants to bind themselves to DeepSeek's high-performance model capabilities through capital ties.

The core pressure behind DeepSeek's financing stems from talent attrition. As tech giants like ByteDance (字节跳动) launch aggressive poaching campaigns with salaries two to three times higher, DeepSeek's core team has experienced significant turbulence. Former core author Wang Bingxuan (王炳宣) has confirmed his move to Tencent (腾讯); multi-modal contributor Ruan Chong (阮翀) joined DeepRoute.ai (元戎启行) as Chief Scientist; and "AI prodigy" Luo Fuli (罗福莉) joined Xiaomi (小米) late last year, where she led the release of the MiMo-V2 series models in March. Most notably, Guo Daya (郭达雅) has been reported to have joined ByteDance (字节跳动). In this context, DeepSeek urgently needs external financing to establish a clear market price for employee options to counter the "financial firepower" of big tech.

Technically, DeepSeek has not stalled despite the talent loss. On the afternoon of April 24, DeepSeek officially released and open-sourced its next-generation flagship model, DeepSeek-V4 (including Pro and Flash versions), debunking rumors of an "indefinite delay" due to computing power and talent issues. The V4 model pioneers a 1M (one million tokens) ultra-long context window and significantly leads the open-source community in Agent capabilities. Meanwhile, DeepSeek's competitors have already moved ahead in the capital markets: Zhipu AI (智谱AI) and MiniMax officially listed on the Hong Kong Stock Exchange in January 2026, with market caps maintained between 50 billion and 70 billion HKD. DeepSeek's entry into the financing market at this time is not only to retain talent but also to secure "ammunition" for new frontiers such as "Physical AI" and trillion-parameter models.