DeepSeek Officially Initiates STAR Market IPO Preparations

Is China’s Richest Man About to Be Replaced?

On September 9, Reuters reported that DeepSeek has selected CITIC Securities to handle its STAR Market listing preparations, with plans to formally submit its application within the year. CITIC Securities has already begun due diligence.

If successful, DeepSeek would become the first cutting-edge general-purpose large model company to list on the STAR Market.

From launching its first external funding round in April, securing funds in June, seeing its valuation soar to 500 billion yuan in July, to now finalizing its IPO process, the entire capital campaign took just five months.

DeepSeek’s founder, Liang Wenfeng, maintains firm control over the company. If the listing succeeds, his net worth could reach the trillion-yuan level, potentially marking a historic shift in who holds the title of China’s richest man.

Two Rounds of Multi-Billion Financing: Liang Wenfeng Strengthens Control

The capital market is fiercely chasing DeepSeek. According to Tianyancha data, DeepSeek launched its first external funding round in April this year and completed the transaction in June, raising a total of 51 billion yuan, with a post-investment valuation nearing 400 billion yuan—setting a record for the highest single-round financing in China’s AI sector.

In this funding round, founder Liang Wenfeng personally contributed about 20 billion yuan, becoming the largest single investor. Tencent invested approximately 10 billion yuan, while the CATL ecosystem contributed around 5 billion yuan. NetEase, JD.com, Monolith Capital, and IDG Capital each invested about 3 billion yuan, with Zhengxingu Investment and Shixiang Technology contributing around 1.5 billion yuan each. The National Artificial Intelligence Industry Investment Fund also strategically invested about 980 million yuan.

Following this, in August, DeepSeek initiated a second funding round, raising its target valuation to approximately 500 billion yuan, with a planned fundraising scale of around 50 billion yuan. Combined with the 51 billion yuan raised in the first round, the total planned fundraising across both rounds exceeds 100 billion yuan.

While raising funds, Liang Wenfeng has maintained tight control over the company.

Before the first funding round, Liang increased his direct shareholding to 34% through industrial and commercial capital increases. Combined with indirect shareholdings, his overall control exceeds 84%, with voting rights nearing 100%.

Moreover, except for state-owned funds, all other external investors hold shares indirectly through limited partnership platforms managed by Liang. These external shareholders lack voting rights and face a five-year lock-up period for their shares.

Despite these stringent conditions, capital continues to fiercely compete for stakes. Even special purpose vehicles (SPVs) are trading shares at a premium in the primary market, with some investors willing to pay 5 million yuan just for a chance to meet and communicate with Liang.

This highly centralized equity structure, which isolates external capital interference, not only solidifies DeepSeek’s operational decision-making but also lays a crucial foundation for the founder’s wealth explosion post-listing.

Under STAR Market rules, companies with a total share capital exceeding 400 million yuan after issuance only need a minimum public offering ratio of 10%. DeepSeek’s two funding rounds have not involved external offshore funds like U.S. capital, significantly simplifying the STAR Market review process.

Based on its current pre-investment valuation of 500 billion yuan, DeepSeek’s post-listing market capitalization could reach 1.5 to 2.5 trillion yuan. Liang currently holds about 84% of the shares directly and indirectly. If valued at a median of 2 trillion yuan, his shareholding value would surpass the trillion-yuan threshold.

Compared to Bloomberg’s current richest list, where Zhang Yiming holds 92.8 billion USD (about 622.3 billion yuan), if DeepSeek successfully lists, Liang could achieve a wealth leap, potentially dethroning Zhang as China’s richest man.

Behind this grand wealth creation narrative lies DeepSeek’s deeper motivations for rushing to list at this moment.

Why Is Liang Wenfeng Actively Raising Funds Now?

From a lone wolf averse to external financing to now actively pursuing the capital market, DeepSeek’s strategic shift is driven by dual pressures from policy incentives and industry competition.

On one hand, the STAR Market has opened a policy window. On June 17, 2026, the Shanghai Stock Exchange released the "Guidelines for Reviewing General AI Large Model Companies Under the Fifth Listing Standard on the STAR Market," expanding the fifth listing standard to include general AI large model firms. The new rules no longer strictly require proven profitability, focusing instead on product deployment, large-scale user adoption, and technical capabilities.

For DeepSeek, this timing is impeccable. The company has already amassed a massive user base. QuestMobile data shows that as of June 2026, DeepSeek had approximately 130 million monthly active users, ranking among the top three AI-native applications in China. Its monetization is also gradually adjusting, shifting from free and low-priced APIs to refined tiered pricing. DeepSeek’s consecutive API price adjustments in August signal its commercialization trials.

Across the industry, Zhipu and MiniMax have already listed in Hong Kong, while the A-share market lacks a pure general-purpose large model company. By listing first on the STAR Market, DeepSeek can not only enjoy higher liquidity premiums in the A-share market but also seize the "pricing power" for A-share large model companies, becoming a key benchmark for subsequent large model firms returning to A-shares and valuation references.

On the other hand, the competition for top industry talent is intensifying. As the large model race enters deep waters, relying solely on technical ideals makes it difficult to retain core R&D personnel.

In recent times, talent mobility within the large model industry has been frequent, with some of DeepSeek’s core technical leaders being poached by well-funded competitors. Reuters noted that Liang hopes to use IPO proceeds, combined with post-listing liquidity-driven equity incentives, to offer more attractive incentive packages to the core team, stabilize the R&D workforce, and prevent further talent drain.

Policy windows, vacant A-share tracks, and talent incentive needs collectively push DeepSeek to accelerate its IPO. However, the primary pressure sustaining such high market expectations comes from the massive capital consumption driven by model iterations and computing power expansion.

Rapid Model Iterations: Net Losses Exceed 700 Million Yuan in 7 Months

AI R&D has always been a cash-burning endeavor.

According to The Information, DeepSeek generated approximately 475 million yuan in revenue in the first seven months of 2026, nearly ten times its full-year revenue in 2025. Its comprehensive gross margin reached 44.6%, with the API business achieving a staggering 82.9% gross margin.

However, DeepSeek also reported a net loss of 715 million yuan over the same period. The root cause lies in the explosive growth of infrastructure and labor costs.

In the first seven months of this year, DeepSeek spent 11 billion yuan on AI infrastructure and computing power equipment, compared to just 1.2 billion yuan for the entire year in 2025—a geometric increase. Additionally, the company is advancing data center construction in Ulanqab, Inner Mongolia, and accelerating compatibility with domestic chips.

Parallel to computing power expansion is aggressive team expansion. In June, DeepSeek announced plans to at least double the size of all departments. In early September, the Harness team leader released approximately 150 engineering positions, covering areas like Agent elastic computing, server-side development, and data engineering, despite DeepSeek’s total headcount previously hovering around 300.

DeepSeek is also maintaining competitiveness through frequent model releases.

After significantly adjusting API pricing for its flagship V4 Pro model in mid-August, DeepSeek officially released its new 552B-parameter MoE model, DeepSeek V4.1 Flash, on September 10.

The model adopts a new Causal-Encoder-Decoder architecture, with only 8B input activations and 16B output activations. It natively supports multimodal visual understanding, drastically reduces KV Cache usage, and outperforms the previous V4 Pro in performance, cost, and speed across all metrics.

Alongside the new model launch, DeepSeek simultaneously reduced API prices, cutting cache-hit input costs by up to 60%. It also plans to route V4 Pro requests to V4.1 Flash, further solidifying its ecological moat through ultimate cost-effectiveness.

Continuous technological iteration and commercialization adjustments are not only DeepSeek’s core means of maintaining industry competitiveness but also the foundation for its IPO and scalable development.

From its early days quietly evolving in labs funded by High-Flyer Quantitative to achieving fame with the R1 model, and now making a push for the STAR Market with intense R&D investment, computing power deployment, and rapid model iterations, DeepSeek is shedding its early "small but elite" lab label and entering the engineering and capitalization phase that all tech giants must undergo.

However, while an IPO can resolve financing challenges, it cannot erase the scrutiny of business models. After listing on the secondary market, investors will adopt more pragmatic evaluation standards, focusing less on technological highlights and more on sustainable operational strength.

DeepSeek’s API users face low migration costs, meaning competitor price or performance changes can directly impact revenue growth. Meanwhile, computing power, R&D, and labor costs remain high, with losses likely to continue expanding. If revenue growth stagnates and losses persist, even with technological advantages, DeepSeek’s high valuation will face reevaluation.

This IPO spectacle, involving hundreds of billions in capital and targeting a trillion-yuan valuation, has only just begun.

References:

"China’s Richest Man Is About to Change," Investment Horizon;

"DeepSeek Prepares for STAR Market IPO: Sources Say CITIC Securities Has Commenced Due Diligence," 21st Century Business Herald;

"DeepSeek Eyes STAR Market IPO," Zhidx;

"The Cash-Rich DeepSeek Rushes to List?" China Newsweek;

"Liang Wenfeng’s Phone Is Overwhelmed with Investor Calls," Rongzhong Finance;

"DeepSeek V4.1 Flash: Stronger, Faster, More Inclusive," DeepSeek Official;

"Just Now, DeepSeek Released a New Model—Flash Outperforms Pro," APPSO.

THE END

Special Disclaimer: This article is for reference only and does not constitute investment advice. Unauthorized reproduction is prohibited.