DeepSeek Paused Its Second Fundraising Round After Leaked Comments — Here's What We Know

Prior to the pause, DeepSeek had just closed its first external fundraising round in June 2026 — over 50 billion yuan (~$7.4 billion) at a valuation above $50 billion . By mid-July, it was already in preliminary talks for a second round targeting a pre-money valuation of about $71 billion . DeepSeek might resume the deal at a later date, according to people familiar with the matter .

DeepSeek's unusual financing structure

The first round involved a founder-centric deal structure highly unusual for a startup of this scale :

• Investors did not buy DeepSeek shares directly. Instead, they invested in a limited partnership, with Liang Wenfeng retaining outsized control as managing partner .
• The structure is designed to let Liang prioritize artificial general intelligence (AGI) research over profit maximization and short-term monetization .
• Major participants in the first round included Tencent Holdings and CATL .
• Liang Wenfeng himself contributed about 20 billion yuan — the largest single investment in the round .

This arrangement has been described as "bizarre" and "aggressively founder-centric," rarely seen outside of founder-controlled tech giants . Investors accepted a five-year lockup period and no voting rights .

Liang Wenfeng's politically sensitive leaked investor meeting comments

A lightly edited transcript of a roughly four-hour closed-door investor meeting held on May 20, 2026, was published by Tencent Tech and quickly went viral in late July . Key claims attributed to Liang:

On Nvidia and China's hardware reliance:

• Liang argued that Nvidia's CUDA ecosystem is not irreplaceable and predicted "meaningful cracks could appear within a year" .
• He said domestic Chinese alternatives — specifically Huawei's chips — are expected to reach functional parity with Nvidia within a year, and that even a 1–2× price premium would be acceptable .
• He stated that "NVIDIA's CUDA dominance is crumbling" and that the monopoly power of Nvidia's AI chip ecosystem is a key vulnerability he expects to break .

On the US-China AI competitiveness gap:

• Liang's central thesis is that China's AI gap with the US is overwhelmingly a compute-access gap, not a talent or algorithmic gap .
• He estimated the gap at roughly one year but noted that China has used only about 1/20th of the compute that the US has — implying the gap is narrower than raw numbers suggest, provided China can secure adequate hardware .
• He argued DeepSeek can remain competitive by operating one or two model generations behind, using radically less compute and emphasizing efficiency .

These remarks are politically sensitive in China because they could be read as acknowledging a structural hardware dependency on US technology (Nvidia) and quantifying a concrete technology gap with the United States — topics the Chinese government typically downplays.

Future plans: Mainland China IPO

Despite the fundraising pause, DeepSeek is actively preparing for an initial public offering on mainland China's stock market:

• Target exchange: Shanghai's STAR Market (China's Nasdaq-style tech board) .
• Timeline: Filing as soon as late 2026 or early 2027, with a listing debut targeted for 2027, possibly as early as Q2 2027 .
• Target valuation: Approximately $71–74 billion (480–500 billion yuan) for the IPO-stage valuation .
• Advisors: DeepSeek is already working with accounting and banking advisors on IPO preparation .
• Motivation: The IPO is seen as necessary to access deeper capital pools onshore, given that China's domestic equity markets remain relatively shallow compared to US markets for AI companies .

Liang Wenfeng has indicated a preference for pursuing AGI over near-term monetization, which could create tension with public market investors who expect returns . The company is also still seeking a smaller private top-up (~$1.5 billion at ~$71 billion valuation) if the second round resumes .