DeepSeek Eyes 2027 IPO, Targets RMB 10B Pre-Listing Raise in China AI Race
DeepSeek is preparing for a domestic initial public offering targeting a 2027 listing, while simultaneously pursuing a private fundraising round of at least RMB 10 billion (US$1.39 billion) at a pre-money valuation of approximately RMB 481.4 billion (US$71 billion) — a move that signals the company is shifting from stealth-mode research lab to a capital-market-ready institution, even as its founder fights to retain near-total operational control.
The disclosure, first reported by Bloomberg citing people familiar with the matter, marks the most concrete timeline yet for a DeepSeek public listing. According to those sources, the company is already working with accounting firms to complete audited financial statements by end-December 2026 — a prerequisite for filing with China's securities regulators. An IPO application is expected no earlier than late 2026, with the formal listing targeted for 2027. DeepSeek did not respond to requests for comment before publication.
The fundraising cadence is striking in its velocity. Just weeks after closing a landmark first institutional round exceeding RMB 50 billion (US$6.94 billion) in early June 2026 — the largest Series A equivalent in Chinese tech startup history — the company is already in early-stage discussions with new investors for a follow-on raise. The Financial Times reported that the new round's pre-money valuation of US$71 billion represents a 37% premium over the post-money valuation implied by the first round, suggesting investor appetite has not cooled despite the compressed timeline.
Liang Wenfeng's Wealth Surge Reframes the Founder-Control Equation
The capital activity has had a dramatic effect on the personal balance sheet of DeepSeek's founder and CEO, Liang Wenfeng. According to the Bloomberg Billionaires Index, his net worth has more than doubled to US$36 billion (approximately RMB 244 billion), overtaking Anthropic co-founder and CEO Dario Amodei and OpenAI co-founder and President Greg Brockman to become the wealthiest AI model founder globally.
Yet the wealth figure obscures a structural tension that will define DeepSeek's path to public markets. Bloomberg estimates that Liang's equity stake was diluted to approximately 78% following the first institutional round. While that remains an extraordinarily high founder concentration by any standard, the ownership architecture is unconventional: major investors including Tencent (HK: 0700), CATL (SZ: 300750), JD.com (NASDAQ: JD), and NetEase (NASDAQ: NTES) channeled capital not directly into DeepSeek's equity, but into a limited partnership of which Liang serves as general partner. That entity then injects capital into DeepSeek.
The practical implication is consequential: those institutional backers hold economic interests — profit-sharing rights — but carry no direct equity, no board seats, and no voting rights. Their shares are subject to a five-year lock-up. Liang reportedly required full LP look-through verification to prevent equity from flowing to unidentified entities. In the private market, this structure gives him near-100% voting control. On the public market, it creates a disclosure and structural cleanup problem that regulators at the China Securities Regulatory Commission (CSRC) and the Shanghai Stock Exchange (上交所) will require to be fully unwound and documented.
Shanghai's New "Fast Lane" Rules Effectively Tailored for DeepSeek
The regulatory backdrop has shifted materially in DeepSeek's favor. On June 17, 2026, the Shanghai Stock Exchange published Issuance and Listing Review Rules Guidance No. 10, a framework specifically designed for AI large-model companies seeking listings on the STAR Market under the so-called "Fifth Set of Standards." The rules do not require profitability or even substantial revenue — only demonstrable technological leadership.
Key qualifying criteria include: primary business focused on autonomous R&D of AI large models or model services; evidence of breakthroughs in core technologies or participation in national AI priority programs; at least one large-model product commercially deployed at scale; and verifiable metrics including user counts, API call volumes, and benchmark rankings on mainstream domestic and international evaluations.
DeepSeek satisfies each criterion by a considerable margin. The guidance effectively creates a regulatory green channel for companies with DeepSeek's profile — technically dominant, pre-profit, and strategically significant to national AI objectives. Notably, the National Artificial Intelligence Industry Investment Fund has already taken a direct RMB 1 billion (US$139 million) stake in DeepSeek exempt from lock-up restrictions and carrying voting rights — an unusual arrangement that signals state-level strategic interest in the company's capital structure ahead of any listing.
Compute Infrastructure, Not Commercialization, Drives the Capital Raise
DeepSeek's senior management has been explicit with prospective investors: the capital is earmarked for frontier AI research and infrastructure, not near-term monetization. Liang has pledged in at least one investor meeting to continue developing open-source AI models while pursuing AGI as the company's primary long-term objective.
The infrastructure ambition is substantial. In June 2026, DeepSeek published a recruitment plan for an IDC (Internet Data Center) team, with job descriptions specifying the construction of gigawatt-scale compute campuses — a build-out that DeepSeek intends to own and operate internally rather than procure from third-party cloud providers.
Simultaneously, Reuters reported on July 7 that DeepSeek initiated in-house AI chip development approximately one year ago and is now engaging external partners, including chip design firms, foundries, and memory manufacturers. The economics justify the investment: a high-performance ASIC for large-model inference using 3nm process technology carries a one-time R&D cost of approximately RMB 3.5–7 billion (US$486–972 million); even a domestically produced 7nm equivalent runs close to RMB 3 billion (US$417 million). When downstream costs — data center construction, chip volume production — are factored in, DeepSeek's total AI chip expenditure could exceed RMB 10 billion.
This capital intensity explains why the company is raising again so quickly after its record first round. The IPO, if executed, would provide a permanent and scalable capital base for what is effectively a multi-year infrastructure and research program.
China's Broader AI Capital Blitz Accelerates Sector-Wide
DeepSeek's financing activity is not occurring in isolation. It is the most visible node in a synchronized capital mobilization across China's AI sector that is reshaping valuations, competitive dynamics, and the global compute supply chain.
Zhipu AI, listed on the Hong Kong Stock Exchange (HK: 2390), moved immediately after its six-month IPO lock-up expired on July 8, 2026. The following morning, it announced a placement of up to 19.78 million new H shares at HK$1,588 per share — a 13% discount to the prior close of HK$1,825. China International Capital Corporation (CICC) acted as sole bookrunner. The placement raised HK$31.4 billion (approximately US$4.0 billion) in gross proceeds, the largest single equity raise by a Chinese AI company post-IPO and the largest tech-sector placement on the Hong Kong market in 2026. The transaction settled in full on July 13. Zhipu's stock rose 14% on the day of the announcement. The company plans to deploy 55% of proceeds into core R&D including continued iteration of its GLM foundation model, 15% into commercial expansion including automotive cabin AI and autonomous driving, and 30% into debt repayment and working capital.
Moonshot AI, operator of the Kimi AI assistant, has executed the most aggressive valuation re-rating in the cohort. From a US$4.3 billion post-money valuation at its December 2025 Series C, the company raised at an US$8 billion valuation in February 2026, then closed a US$2 billion Series D in May at a post-money valuation exceeding US$20 billion, and has since launched a new round targeting a US$30 billion valuation. The six-month valuation increase exceeds 6x.
ByteDance, the privately held parent of Douyin and Doubao, is pursuing a different financing route. According to foreign media reports, the company is in discussions with Wall Street banks — including Citigroup, JPMorgan Chase, and Goldman Sachs — for an offshore syndicated loan of approximately US$20 billion, with a base tenor of three years and an extension option to five years. If completed, the facility would set a record for offshore borrowing by a Chinese corporate. The driver: ByteDance has revised its 2026 AI infrastructure capital expenditure budget upward by 25% to RMB 200 billion (US$27.8 billion), after Doubao's user base and token call volumes surged beyond initial projections and hardware costs — particularly for HBM and DRAM memory chips — came in materially above plan.
Governance Overhang Remains the Critical Variable for DeepSeek's IPO Path
The competitive logic is clear: in a race toward AGI, capital starvation means falling behind, and falling behind is irreversible. DeepSeek's hiring notice states explicitly: "Humanity stands on the eve of AGI. Join DeepSeek to witness the development of AGI firsthand." Following the June first-round close, the company announced its largest-ever recruitment drive on June 25, pledging to at least double the headcount of every department.
But the IPO path is not without friction. The limited-partnership ownership structure that has served Liang's control objectives in the private market will require full restructuring and regulatory disclosure before any A-share listing can proceed. How the economic interests of Tencent, CATL, JD.com, and NetEase — each of which committed tens of billions of RMB — translate into tradeable public equity is a question that the CSRC will demand answered in granular detail.
The second round of private fundraising, sources indicate, may also be designed partly to bring additional state-affiliated funds onto the cap table — a move that would smooth regulatory review and align DeepSeek's ownership structure with the strategic priorities embedded in the SSE's June 2026 guidance. If that reading is correct, the pre-IPO raise is as much a regulatory positioning exercise as it is a capital event.
The timeline remains contingent. Execution depends on market conditions, financial audit completion, and DeepSeek's operating performance through year-end 2026. But the direction of travel is unambiguous: China's most technically advanced AI company is preparing to enter public markets, on its founder's terms, and at a pace that the global AI industry has not previously seen.
Related Coverage:
DeepSeek Seeks Second Funding Round at $71B Valuation as Infrastructure Push Accelerates