Report: DeepSeek Raises $7 Billion from Alibaba, Tencent, and China's Big Fund

DeepSeek has reportedly closed a massive $7 billion (around 50 billion yuan) first funding round, with Alibaba, Tencent, and China's National Integrated Circuit Big Fund each putting in $1.4 billion (10 billion yuan). Founder Liang Wenfeng contributed the remaining $2.8 billion (20 billion yuan) himself, according to sources. The deal values the AI lab at roughly $50 billion (350 billion yuan). Huawei also reportedly participated in the negotiations, but run-of-the-mill top VCs and PEs were shut out.

The injection of outside cash directly addresses the twin crises of talent and compute that have been plaguing DeepSeek. Earlier, key researcher Guo Daya jumped ship to ByteDance, frustrated by a salary inversion and insufficient computing power. To hold onto its team, DeepSeek has doubled salaries and is now offering core researchers eight-figure stock option packages. On the compute side, while DeepSeek's V4 model reportedly approaches GPT-5.4 in raw capability, the company has struggled to deeply integrate with chips like Huawei's Ascend — to the point where DeepSeek engineers have had to write some low-level operators for Huawei. The company urgently needs more capital to buy high-end hardware directly.

Now flush with cash, DeepSeek is pivoting from a pure research outfit toward AI product development and commercial deployment. The company is accelerating its release cadence, with plans to ship a multimodal model, V4.1, in June. Recent job postings show a clear shift toward hiring for multimodal and agent roles — skills that translate directly to real-world applications, as seen in models like Zhipu AI's GLM-5V-Turbo and assistants like Meta's Hatch.