DeepSeek’s “three no-determination” comes to an end: Liang Wenfeng invests 200 billion, pushing talent and computing power to force a transition

梁文鋒200億融資DeepSeek

According to a June 17 article by ThinkAI, DeepSeek has completed approximately 50 billion RMB in fundraising, with founder Liang Wenfeng contributing 20 billion RMB. A year earlier, Liang Wenfeng had declared the principles of “no fundraising, no IPO, and no commercialization.” The shift this time comes from the combined effects of three major pressures: the resignation of at least five core R&D personnel one after another, the intensifying and highly competitive race over compute power, and the IPO window pressure caused by a sharp rise in competitors’ stock prices after they listed.

Investor composition and valuation data for the 50 billion RMB funding round

Based on the reported funding structure: Liang Wenfeng contributed 20 billion RMB (the largest single investor), Tencent contributed 10 billion RMB, CATL contributed 5 billion RMB, and JD.com, NetEase, and IDG Capital each contributed 3 billion RMB. The National AI Industry Investment Fund contributed 1 billion RMB (holding voting rights). Total funding was $74 billion (about 50 billion RMB), and the total post-funding valuation exceeded 350 billion RMB.

The fundraising plan was first disclosed in early April 2026. The capital market’s reaction was strong; at one point, a fund manager publicly offered a reward of 500,000 yuan in a community post to seek an introduction channel.

Liang Wenfeng’s background: from 806-scoring top gaokao student to the founder of High-Flyer Quant

In 2002, Liang Wenfeng won the top gaokao spot in Zhanjiang with 806 points. Although he met Tsinghua University’s admission threshold, he chose Zhejiang University’s electronic information engineering program. In 2008, he began quantitative trading with 80,000 yuan in principal; in 2015 (at age 30), he founded High-Flyer Quant.

In 2019, High-Flyer Quant became China’s first quantitative hedging fund to raise more than 100 billion RMB. Related institutions speculate that its net profit in a single year was 50 to 60 billion RMB. After a drawdown in its quant performance in 2021, Liang Wenfeng made a major purchase of NVIDIA A100 high-end graphics cards to position himself in the large-model race ahead of time. By 2023, High-Flyer had become the only independent entity outside the big tech firms that had reserved more than 10,000 A100 units.

Three driving factors that overturned the “three no theorems”

Core talent drain: Luo Fuli was recruited by Lei Jun to Xiaomi, and Guo Dayaya switched jobs to ByteDance to serve as head of the Agent business, with at least five core R&D personnel leaving in succession. DeepSeek’s long-standing refusal to raise funding means it lacks equity incentive tools, making it unable to compete in the AI talent race with top firms that have equity incentive mechanisms.

Compute consumption exceeds its own capital: The upcoming new version’s training and daily operations require two to three times more compute power than V3; the procurement price of a single high-end H100 GPU exceeds 200,000 RMB. The existing 10,000 A100 units are no longer enough, and relying solely on self-owned funds from quantitative returns cannot keep up with the big tech firms’ boundless compute spending.

The IPO valuation window is limited: After Zhipu AI and MiniMax completed their listings, their share prices surged significantly, while Jietiao Xingchen simultaneously pushed forward its IPO preparations. If DeepSeek misses this round of capital market upside, the risk is that valuation in the primary market could shrink dramatically.

User growth data after the release of DeepSeek V3

DeepSeek V3 was officially released on December 26, 2024. In January 2025, it added a cumulative 125 million users, including 100 million new users within seven days—its growth rate far exceeded ChatGPT (which took about two months to reach 100 million users). It became a phenomenon-level product among domestic large models.

Common questions

Why did Liang Wenfeng personally invest 20 billion RMB instead of bringing in more external capital?

According to reports, in this fundraising structure, Liang Wenfeng’s personal contribution of 20 billion RMB is the largest single investor, while external institutions (Tencent, CATL, etc.) combined about 30 billion RMB. Large personal investment helps maintain the founder’s control over the company and sends the outside world a strong signal of confidence in DeepSeek’s long-term value.

What does the National AI Industry Investment Fund’s holding of voting rights mean?

According to reports, the National AI Industry Investment Fund took a stake of 1 billion RMB and holds voting rights. This is a special arrangement that differs from ordinary financial investments. It means the fund has a certain standing in the company’s major decisions, typically reflecting the intent of policy-driven capital to guide specific strategic directions.

What were DeepSeek’s original “three no theorems,” and where were they announced?

According to reports, about a year earlier (around 2025), Liang Wenfeng announced DeepSeek’s three no theorems in a public setting: no fundraising, no IPO, and no commercialization. This principle was previously viewed as DeepSeek’s core positioning distinct from other AI unicorns, and this 50 billion RMB funding round indicates that the principle has now officially ended.

Disclaimer: The information on this page may come from third-party sources and is for reference only. It does not represent the views or opinions of Gate and does not constitute any financial, investment, or legal advice. Virtual asset trading involves high risk. Please do not rely solely on the information on this page when making decisions. For details, see the Disclaimer.
Comment
0/400
No comments