TL;DR: Chinese AI startup DeepSeek is in discussions to raise $1.5 billion at a $71 billion pre-money valuation while preparing for a 2027 initial public offering in mainland China or Hong Kong. The round follows a recent $7 billion funding closed in mid-2025 and positions the company's valuation at a sevenfold increase from its April $10 billion level. The startup continues to gain enterprise market share by running its models on domestic Huawei hardware, bypassing US semiconductor export controls.

How is DeepSeek reaching a $71 billion valuation within two years of founding?

DeepSeek is leveraging high developer adoption and a capital-efficient open-source strategy to justify its rapid valuation climb from $10 billion in April to $71 billion. According to developer platform Vercel, DeepSeek processed nearly 23% of all enterprise AI gateway tokens in June, trailing only Anthropic's 32% share. This growth rests on the company's V3 and R1 reasoning models, which deliver performance comparable to frontier US systems at a fraction of the operating cost. The startup's rapid market penetration is driven by its ability to offer high-performance AI reasoning at lower prices than its Western competitors. It built this user base in a highly competitive market that includes established US developers and domestic rivals like Moonshot AI.

Unlike US rivals that build proprietary, closed-source ecosystems, the Hangzhou-based startup distributes open-source weights. This approach drives rapid enterprise integration, while the company generates revenue through API calls and model licensing. However, financial reports indicate the firm's current revenue is still far from the levels typically required to sustain a $71 billion valuation, making its commercial monetization strategy a primary point of scrutiny for incoming private and public investors.

Rapid capital accumulation and Liang Wenfeng's ownership

DeepSeek closed its first outside funding round in mid-2025, raising $7 billion from major Chinese technology and industrial firms. This investor group included Tencent, CATL, JD.com, NetEase, IDG Capital, and China's National Artificial Intelligence Industry Investment Fund. Founder Liang Wenfeng personally injected $3 billion into that round, retaining a 78% stake in the company. The latest $71 billion valuation pushes Liang’s net worth to approximately $36 billion, positioning him as one of the wealthiest AI founders globally. Outside of the state-backed National Artificial Intelligence Industry Investment Fund, which secured direct voting rights, other outside investors hold capital in a limited partnership controlled by Liang, subject to a five-year lock-up period.

DeepSeek targets a 2027 IPO to bypass the private funding constraints of US rivals

DeepSeek has initiated structural preparations for an initial public offering to secure permanent public capital before its American competitors. The startup is working with accounting firms and banking advisors to finalise its financial documentation and complete audit reports by December 2026. This timeline positions the company to file for an IPO as early as late 2026, with a targeted exchange debut in 2027.

This rapid transition to public markets contrasts with US-based OpenAI, which recently pushed its own projected IPO timeline to 2027 while seeking a $1 trillion valuation. By listing earlier, DeepSeek plans to transition from private funding rounds to liquid public equities. This move gives its early institutional backers a clear exit path and establishes a continuous capital pipeline. The rapid pace of this fundraising cycle—climbing from a $10 billion valuation in April to $71 billion in July—raises questions among analysts about whether private market pricing is outrunning actual business fundamentals. Public market investors will evaluate these metrics closely when the company files its official registration statement.

The choice between mainland China and Hong Kong exchanges

While DeepSeek has not finalised its listing venue, financial advisors expect the firm to choose either mainland China's A-share market or the Hong Kong Stock Exchange. Listing in Hong Kong allows the company to attract global capital while complying with Beijing’s regulatory frameworks. Recent listings of domestic competitors like Zhipu AI and MiniMax provide a precedent; Zhipu shares surged nearly 1,600% after its January debut. The chosen venue will indicate how DeepSeek intends to manage geopolitical risks and balance international investor access with domestic security oversight.

Can DeepSeek sustain its AI development under US semiconductor export controls?

DeepSeek maintains its competitive model performance by running its cloud infrastructure on domestic chips from Huawei Technologies, demonstrating that Chinese AI firms can operate without advanced Nvidia hardware. US export restrictions prevent Chinese entities from buying Nvidia's top-tier Hopper and Blackwell chips. DeepSeek sidesteps this bottleneck by building its cluster architecture around domestic alternatives, validating the viability of China's domestic AI hardware stack.

To further insulate itself from supply chain disruptions, DeepSeek is also designing its own proprietary AI chips. This parallel development track aims to reduce reliance on external suppliers like Huawei. Developing custom silicon is expected to improve the company's unit economics, lowering the infrastructure costs associated with running massive reasoning models. This technical independence challenges the prevailing assumption that American export curbs would permanently slow Chinese AI development. Instead, the firm's progress indicates that software optimization and alternative hardware architectures can compensate for hardware limitations.

Global market dynamics and the open-source model trajectory

The demand for cost-effective AI solutions aligns with global market growth projections. Precedence Research estimates the global large language model market at $7.77 billion in 2025, expanding to $149.89 billion by 2035 at a 34.44% compound annual growth rate. In this expanding market, DeepSeek's open-source distribution model challenges the premium pricing structures of closed-source US labs like OpenAI and Anthropic. If DeepSeek's low-cost API strategy continues to capture developer market share, it could force a downward pricing shift across the global AI industry, even as the company works to scale its own revenue.

Key Takeaways

  • Rapid Valuation Climb: DeepSeek is seeking $1.5 billion in new funding at a $71 billion pre-money valuation, marking a 37% increase from its previous valuation just weeks prior and a sevenfold increase since April.
  • Accelerated IPO Timeline: The company is preparing for an IPO targeted for 2027 in either mainland China or Hong Kong, working with advisors to complete the necessary financial documentation by December 2026.
  • Geopolitical Independence: By utilizing Huawei hardware and developing proprietary silicon, DeepSeek is bypassing US chip export limits while capturing 23% of enterprise AI gateway tokens processed on Vercel.