Chinese AI lab DeepSeek is telling investors it will prioritize the costly pursuit of artificial general intelligence over near-term revenue, a bold pitch in its first-ever external funding round.
Chinese AI lab DeepSeek is telling investors it will prioritize the costly pursuit of artificial general intelligence over near-term revenue, a bold pitch in its first-ever external funding round.

(P1) Chinese AI startup DeepSeek is in the final stages of its first external funding round, telling prospective investors it will prioritize frontier research into artificial general intelligence over short-term commercialization. The Hangzhou-based lab is targeting a valuation of 70 billion yuan, or roughly $10 billion, in a deal that could see Tencent Holdings Ltd. participate as both a strategic investor and a key customer.
(P2) "The lab intends to pursue artificial general intelligence as a primary goal, and will keep releasing open-source models rather than chase short-term commercialisation," founder Liang Wenfeng told prospective investors, according to a Bloomberg report. This research-first identity has defined DeepSeek, which until now has been entirely funded by High-Flyer Quant, the profitable quantitative trading firm Liang also founded.
(P3) The company is seeking to raise at least $300 million in the round, with sources close to the matter indicating a state-backed vehicle, the National AI Industry Investment Fund, is in talks to invest around 10 billion yuan. IDG Capital and Monolith are also reportedly close to finalizing their participation. The move to accept outside capital reflects the immense scale of funding required to train frontier models, which now exceeds what a private hedge fund can self-finance.
(P4) The fundraising round highlights a strategic divergence in China's AI sector, with DeepSeek betting that a focus on open-source, foundational AGI research will create more long-term value than the enterprise-focused, revenue-driven models of competitors like Alibaba Group and Baidu. The strategy is validated by Tencent's dual role; its cloud division just announced it will commercialize DeepSeek's V4-Pro model as a paid service, a move that sent Tencent shares up more than four percent on the news.
DeepSeek has made significant waves by releasing powerful models under permissive open-source licenses. In April, the lab launched its V4 family, including a 1.6-trillion-parameter Mixture-of-Experts (MoE) model and a smaller 284-billion-parameter variant. These models are notably optimized to run on domestic hardware from Huawei and Cambricon, a crucial signal to a Chinese market facing restricted access to high-end US accelerators from companies like Nvidia.
The company's pitch to investors leans heavily on the credibility it built with the January 2025 release of its R1 model, which demonstrated that a Chinese lab could develop frontier-level reasoning models at a fraction of the cost spent by US counterparts.
Tencent's involvement is a significant endorsement of DeepSeek's approach. Beginning May 27, Tencent Cloud will shift the DeepSeek-V4-Pro and its own Hy3 Preview models from a free beta to a usage-based pricing structure. This move signals Tencent's confidence in enterprise demand for high-end AI services and provides DeepSeek with a clear path to monetization, even as it prioritizes research.
For investors, Tencent's decision to both fund DeepSeek and integrate its models into its commercial cloud offerings provides a powerful proof point. It shows a direct path from DeepSeek's frontier research to recurring revenue through a major cloud partner, mitigating risks associated with the lab's non-commercial focus. The market's positive reaction to Tencent's announcement underscores the perceived value of this symbiotic relationship in China's competitive AI market.
This article is for informational purposes only and does not constitute investment advice.