On September 10 2026 the AI story that has defined China’s technology sector for the past year took its next logical step. DeepSeek — the Hangzhou-based AI company whose open-weight models sent the global AI industry into a reassessment of what was possible and at what cost — has appointed CITIC Securities to prepare a listing on the Shanghai Stock Exchange’s STAR Market. The company aims to begin the IPO process this year, according to Reuters, while simultaneously completing a follow-on funding round that would value it at roughly 500 billion yuan — approximately $75 billion at current exchange rates.
A $75 billion valuation for a company founded in 2023 that has never charged for its flagship models places DeepSeek in a category of its own in Asian startup history. Follow every Asia tech and startup story at BestStartup Asia.
DeepSeek’s STAR Market IPO — What September 10 2026 Means
The Shanghai Stock Exchange’s STAR Market — the Science and Technology Innovation Board launched in 2019 — is China’s designated listing venue for high-tech and innovation companies. It operates under a registration-based IPO system rather than the approval-based system that governs the main boards, which means faster listing timelines and more flexible profitability requirements. Previous STAR Market listings include SMIC, the semiconductor manufacturer, and numerous AI and biotechnology companies that needed capital before reaching conventional profitability thresholds.
DeepSeek’s appointment of CITIC Securities as a lead underwriter alongside three other underwriters in the frame — per South China Morning Post’s reporting — signals that the process is substantive rather than exploratory. Companies do not appoint major investment banks and begin exchange preparation unless they are serious about completing the listing.
The $75 billion valuation at which the parallel follow-on funding round is being discussed would make DeepSeek the most valuable Chinese AI startup ever to list publicly. For context, the $75 billion figure is larger than the market capitalisation of most traditional Chinese internet companies and represents a premium that could only be justified by the scale of commercial demand for DeepSeek’s models — a demand that has proven real, global, and accelerating since the V3 and R1 model releases in late 2024 and early 2025. Read more China AI startup stories at BestStartup Asia.
DeepSeek’s Story — From Quant Fund to Global AI Disruptor
DeepSeek was founded in 2023 as a side project of High-Flyer Capital Management, a Hangzhou-based quantitative hedge fund. Its founder Liang Wenfeng decided to explore large language models as a research direction, initially without commercial ambitions. The company’s first public model releases attracted attention within China’s AI research community. Then V3 and R1 changed everything.
DeepSeek R1, released in January 2025, matched or exceeded the performance of OpenAI’s best models on most standard benchmarks — at a fraction of the reported training cost. The efficiency gap raised immediate questions about the assumptions underlying the multi-hundred-billion-dollar AI infrastructure buildout in the United States. If you could train frontier models for tens of millions of dollars rather than hundreds of millions, what did that mean for the companies spending billions on compute? The question ran through global AI markets in January 2025 and its reverberations have not stopped since.
By September 2026, DeepSeek’s models are among the most widely deployed open-weight AI models globally. Enterprise customers across Asia, Europe, and increasingly the United States are using DeepSeek models for applications ranging from customer service automation to scientific research to software development. The open-weight release strategy — making model weights freely available for deployment — created distribution that no sales force could have achieved at the same speed.
What DeepSeek’s $75 Billion Valuation Means for Asian AI
The $75 billion valuation does not reflect DeepSeek’s current revenue. The company has not publicly disclosed its monetisation model or commercial revenue figures. The valuation reflects the market’s assessment of what DeepSeek’s models, distribution, and engineering talent are worth as a strategic asset in the global AI race — and the answer, apparently, is $75 billion.
For the broader Asian AI ecosystem, the DeepSeek STAR Market IPO announcement carries three significant implications. First, it signals that China’s domestic capital markets are ready to absorb frontier AI company listings at global scale valuations — a development that reduces Chinese AI companies’ dependency on US public markets that have become increasingly difficult to access for geopolitical reasons.
Second, it validates the open-weight model strategy as commercially viable at scale. DeepSeek’s decision to release model weights openly rather than keeping them proprietary has been controversial in the AI industry — some argue it gives away too much competitive advantage. The $75 billion valuation suggests the market disagrees. Distribution, trust, and adoption at scale are worth more than secrecy.
Third, it intensifies competition for AI talent and capital across Asia. A DeepSeek IPO at $75 billion will generate significant liquidity for early employees and investors, who will recycle capital and talent into new AI ventures across China, Southeast Asia, and the broader region. The secondary effects of a successful AI IPO at this scale are often more important for ecosystem development than the IPO itself. Follow every Asia startup story at BestStartup Asia.
Taiwan Foundries, Washington’s Chip Policy, and the Geopolitical Context
The DeepSeek IPO announcement landed on a day when Asia’s technology sector was already processing multiple simultaneous pressures. Taiwan’s foundries — led by TSMC — posted another record month of production in August 2026. Washington has continued telling third countries that the semiconductor race no longer permits spectators, using export control expansion and bilateral technology agreements to draw clearer lines around which countries can access which chips.
DeepSeek’s efficiency-focused approach to model training was partly driven by restricted access to the most advanced NVIDIA chips. The company built world-class models using chips it could actually obtain — a constraint that paradoxically produced innovation. The STAR Market IPO will give DeepSeek access to capital that allows it to invest in whatever compute it can legally acquire, potentially accelerating the gap between its open-weight models and the competition.
Key Takeaways — DeepSeek STAR Market IPO September 2026
DeepSeek has appointed CITIC Securities and three other underwriters to prepare a STAR Market IPO in Shanghai as of September 10 2026. A parallel follow-on funding round is being discussed at approximately 500 billion yuan — roughly $75 billion. DeepSeek was founded in 2023 as a side project of High-Flyer Capital Management in Hangzhou. The company’s R1 model, released January 2025, matched frontier US model performance at a fraction of reported training cost. DeepSeek’s open-weight model strategy has created global distribution across Asia, Europe, and the United States. A successful STAR Market listing at $75 billion would be the most valuable Chinese AI company to list publicly. Taiwan foundries posted another record production month in August 2026. US chip export controls continue to reshape which Asian companies can access which semiconductor technology.
Frequently Asked Questions
What is DeepSeek and when was it founded?
DeepSeek is a Hangzhou-based AI company founded in 2023 as a side project of High-Flyer Capital Management, a Chinese quantitative hedge fund. It became globally known when its R1 model, released in January 2025, matched the performance of frontier US models at a fraction of reported training cost. Its open-weight models are now among the most widely deployed AI models globally.
What is DeepSeek’s IPO valuation in September 2026?
DeepSeek is pursuing a follow-on funding round discussed at approximately 500 billion yuan — roughly $75 billion at current exchange rates — alongside its STAR Market IPO preparation announced September 10 2026. CITIC Securities has been appointed as a lead underwriter for the listing.
What is the STAR Market?
The STAR Market is the Shanghai Stock Exchange’s Science and Technology Innovation Board, launched in 2019. It operates under a registration-based IPO system with more flexible profitability requirements than China’s main boards, making it the preferred listing venue for high-tech and AI companies that need public capital before reaching conventional profitability.
Why did DeepSeek release its models as open-weight?
DeepSeek’s open-weight model strategy — releasing model weights freely for deployment — created global distribution that no sales force could have achieved at the same speed. The $75 billion valuation at which it is pursuing its IPO and follow-on round suggests the market believes distribution and trust at scale are worth more than proprietary secrecy.
Where can I follow Asia AI and startup news?
Follow every Asia startup funding story, AI company profile, and technology investment trend at BestStartup Asia — updated every week across every country in the region.
