Analysis of Differentiated Hong Kong IPO Paths and To B vs. To C Business Model Comparison Between Zhipu AI and MiniMax

Based on the latest market data and company announcements, I systematically analyze the differentiated Hong Kong IPO paths and the pros and cons of To B vs. To C business models between Zhipu AI and MiniMax.

---

Zhipu AI and MiniMax are advancing their Hong Kong IPO processes almost simultaneously, forming a direct competitive situation. Both companies obtained the overseas issuance and listing filing notice on December 22, 2025, and launched their IPO subscriptions on the same day, December 30 [1][2][3].

Item	Zhipu AI	MiniMax
Stock Code	2513	TBD
Expected Listing Date	January 8, 2026	January 9, 2026
Estimated Fundraising Amount	HK$4.3 billion	TBD
Issuance Valuation	Over HK$51.1 billion	HK$46.1-50.4 billion
Offer Price	HK$116.20 per share	HK$151-165 per share
Issuance Scale	37.4195 million H shares	25.3892 million shares

Zhipu AI is expected to take the lead in securing the title of ‘the world’s first large model stock’, which is of great significance for the flexibility of capital market operations [2][3]. A former middle manager of Zhipu stated that going public is not only for financing but also to provide an exit channel for external shareholders. The AI industry has a precedent—Megvii missed the optimal timing during its IPO process and was eventually overtaken by other enterprises [4].

Zhipu AI has completed 8 rounds of financing since its establishment, with a total financing scale exceeding 8.3 billion yuan, and has strong financial strength [2]. As of June 2025, the company’s cash and cash equivalents were 2.5 billion yuan, but its cash burn rate is fast: the net cash consumption from operating activities in 2024 was 2.245 billion yuan, with a monthly average of about 187 million yuan; the monthly average consumption rose to about 221 million yuan in the first half of 2025 [4].

MiniMax has more abundant cash flow on its books. As of September 30, 2025, its cash balance (including cash, equivalents, and wealth management products) reached 1.04 billion US dollars (about 7.334 billion yuan). The company expects a monthly cash consumption of 27.9 million US dollars (about 196 million yuan), which is slightly lower than Zhipu [4][5]. MiniMax has spent only 500 million US dollars since its establishment, achieving globally leading results in full multimodality with about 1% of OpenAI’s costs [5].

---

: Reduce usage thresholds via ‘open-source + API calls’, target developers and enterprise user markets, and form a closed loop from open-source ecosystem to commercial API paid conversion [1][6].

:

• Mainly serves enterprise clients and developers
• High dependence on large clients: Revenue from the largest client exceeds 10% of total revenue, and revenue from the top five clients is close to 50% [4]
• Overall gross margin remains around 50%, with relatively strong profitability [4]

:

• High Client Stability
  : Once a cooperative relationship is established with enterprise clients, it usually has a long contract cycle and continuous procurement demand
• High Average Customer Value
  : Enterprise-level projects usually involve large contracts and customized services
• Predictable Cash Flow
  : Contract-based revenue model makes financial planning more certain
• Strong Technical Barriers
  : Adapts to more than 40 domestic chips, with unique advantages against the backdrop of domestic substitution [2]

:

• High Policy Sensitivity
  : Demand from government and enterprise clients is greatly affected by policies
• Long Project Cycle
  : Enterprise-level sales processes are complex, and the payment collection cycle may be long
• Growth Ceiling
  : The enterprise market capacity is relatively limited, and expansion speed is restricted by the economic cycle
• Customization Pressure
  : Large clients often require customized development, increasing operational costs

: Focus on multimodal models to enter film and television content creation, directly reach C-end users via AI-native products, and form a diversified revenue model of ‘subscription services + in-app purchases + API calls’ [1][5][6].

:

• AI-native product revenue accounts for 71.4% of total revenue (2024), covering subscription and in-app recharge services
• Open platform and enterprise services account for 28.6% [6]
• Overseas revenue share rises sharply: 19.2% in 2023 → 69.8% in 2024 →73.1% in the first three quarters of 2025 [5][6]

:

• Average MAU of about 3.15 million in 2023
• More than six-fold growth to 19.11 million in 2024
• As of September 30, 2025, cumulative users reached 212 million, MAU reached 27.6 million, and paying users reached 1.77 million [5][6]

:

• Huge User Base
  : The ceiling of the C-end market is much higher than that of the enterprise market
• Strong Globalization Potential
  : Overseas revenue has become the main pillar, covering more than 100 countries and regions [5]
• Diversified Monetization Channels
  : Multiple monetization methods such as subscriptions, in-app purchases, advertising, and APIs
• Network Effect
  : User growth may bring social communication effects, reducing customer acquisition costs

:

• High Customer Acquisition Cost
  : Fierce competition in the C-end market requires continuous investment in marketing and promotion
• Low User Stickiness
  : Consumers have many choices and low switching costs
• Volatile Gross Margin
  : Gross margin of C-end business is only 4.7% (about 50% after excluding Xingye), and profitability is affected by product mix [4][6]
• Fierce Competition
  : Needs to compete directly with giants like ByteDance, Alibaba, and Tencent [7]

Indicator	Zhipu AI (To B)	MiniMax (To C)
2025 Revenue	190 million yuan (first half)	376 million yuan (first 9 months)
Gross Margin	~50%	23.3% (overall), 4.7% (C-end native)
Large Client Dependence	Top 5 clients account for 50% of revenue	2,500 paying clients on open platform
Overseas Revenue Share	Not disclosed	73.1%
User Scale	2.7 million developers + enterprise clients	212 million users, 27.6 million MAU
Cash Burn Rate	Monthly average of 221 million yuan	Monthly average of about 196 million yuan

---

Both companies are in the stage of explosive revenue growth, but at the same time face huge losses:

:

• Revenue in the first half of 2025 was 190 million yuan, exceeding the full-year level of 2023
• Net loss of 2.358 billion yuan in the same period [7]
• Revenue has doubled for three consecutive years [2]

:

• Revenue in the first three quarters of 2025 was 53.44 million US dollars (about 376 million yuan), a year-on-year surge of 175% [5]
• Revenue in 2023, 2024, and the first three quarters of 2025 was 3.46 million US dollars, 30.523 million US dollars, and 53.437 million US dollars respectively [6]
• Net loss of 512 million US dollars (about 3.6 billion yuan) [7]

The losses of both companies are mainly due to sustained R&D investments. Both Zhipu and MiniMax have self-developed large models, which require large-scale cash burn. Zhipu clearly stated that it will continue to invest heavily in R&D [4].

Although MiniMax has a larger revenue scale (376 million yuan in the first 9 months of 2025 vs. Zhipu’s 190 million yuan in the first half), Zhipu’s IPO valuation is higher (HK$51.1 billion vs. MiniMax’s HK$46.1-50.4 billion). This difference may stem from:

1. Technical Positioning
   : Zhipu is the largest independent large model vendor in China by revenue in 2024, ranking first in the industry [3]
2. Market Position
   : Zhipu has more than 2.7 million clients and developers, establishing a broader ecosystem [2]
3. Domestic Substitution Dividend
   : Adapts to more than 40 domestic chips, with strategic value against the backdrop of information technology application innovation (ITA) [2]
4. Profitability
   : Zhipu’s 50% gross margin is higher than MiniMax’s, showing stronger commercialization capabilities [4]

Huaan Securities pointed out that Zhipu reduces usage thresholds via ‘open-source + API calls’ and targets developers and enterprise user markets to form a conversion from open-source ecosystem to commercial API; MiniMax focuses on multimodal models to enter film and television content creation, and the two have differences in segmented markets [1].

---

Bai Wenxi, vice chairman of the China Enterprise Capital Alliance, believes that under the impact of DeepSeek, the differentiation route of the ‘Six Little Dragons’ has become clear [1]:

: Zhipu and MiniMax rush to Hong Kong first

• Zhipu bets on enterprise privatization + API
• MiniMax focuses on public cloud calls of coding/voice models

: Dark Side of the Moon, Jieyue Xingchen

• Backed by Alibaba and Tencent + state-owned assets respectively
• Model indicators are still in the first camp, but commercialization has just started
• Expected to launch IPOs from the second half of 2025 to 2026 [1]

: Mianbi Intelligence, 01.Wanwu

• Basically cut or reduced pre-training, turning to industry agents and scene-based packaged applications
• Valuation logic changes from ‘foundation model’ to ‘solution’
• Subsequent independent IPOs are more difficult, and mergers and acquisitions are more likely [1]

:

1. Cash Flow Certainty
   : Enterprise clients provide stable contract revenue
2. Technical Barriers
   : Deeply customized services establish competitive barriers
3. Policy Dividends
   : Strong demand for digital transformation from governments and enterprises
4. Brand Endorsement
   : Serving top clients brings market credibility

:

1. Market Scale
   : The ceiling of the consumer market is much higher than that of the enterprise market
2. Growth Potential
   : User growth may show exponential爆发 (e.g., ChatGPT)
3. Network Effect
   : User data feedback promotes model iteration
4. Globalization
   : C-end products are easier to break through geographical restrictions
5. Valuation Logic
   : User scale supports higher valuation imagination space

It is worth noting that the two companies are not completely separated. MiniMax has formed a ‘B+C dual-drive’ model, with C-end引流 and B-end monetization, and its open platform has penetrated enterprise clients in more than 100 countries and regions [5]. Zhipu is also exploring the developer ecosystem and establishing a broader user base through open-source strategies.

Huaan Securities pointed out that Hong Kong IPOs may guide AI large model vendors to shift their narrative logic from ‘telling technical stories’ to ‘realizing commercial value’, and will also provide a basis for subsequent financing and valuation of large model enterprises [1].

---

According to the UNCTAD report, the global AI market is expected to soar from 189 billion US dollars in 2023 to 4.8 trillion US dollars in 2033, growing 25 times in just ten years [5]. The global large model market will grow rapidly from 10.7 billion US dollars in 2024 to 206.5 billion US dollars in 2029, with a compound annual growth rate (CAGR) of 80.7% [6].

This huge market space provides sufficient development opportunities for Zhipu and MiniMax, but also means more fierce competition.

:

• Limited market scale, growth restricted by enterprise IT budgets
• Marginal costs are difficult to reduce due to customization needs
• Risk of reduced bargaining power due to large client dependence
• Revenue stability affected by policy cycle fluctuations

:

• Continuous rise in customer acquisition costs, diminishing marginal benefits of user growth
• Time required to cultivate willingness to pay, and user habits not yet fully formed
• Giants enter the C-end market, uncertain competitive landscape
• Increased content compliance and regulatory risks

:

• Scarcity of ‘the world’s first large model stock’
• Domestic substitution policy dividends
• High gross margin and client stickiness
• Tsinghua University technical team background

:

• Global business layout, overseas revenue share exceeds 70%
• Rapid user scale growth, 212 million user base
• Leading multimodal technology, outstanding commercialization of video generation models
• Lower valuation multiple

:

• Sustained losses and fast cash flow consumption
• Rapid technological iteration, possible rapid disruption
• Competitive pressure from giants (OpenAI, ByteDance, Alibaba, etc.)
• Regulatory policy uncertainty

---

Zhipu AI and MiniMax have chosen two completely different commercialization paths: the former focuses on To B, emphasizing enterprise services and API ecosystems; the latter focuses on To C, prioritizing consumer applications and global expansion. Both models have their pros and cons: the To B model provides more stable cash flow and higher gross margins, but growth space is limited; the To C model has greater market potential and global space, but faces fierce competition and uncertain profitability.

From the current financial data, Zhipu has stronger profitability (50% gross margin), while MiniMax has faster growth (212 million users). The listing of the two companies will bring the first batch of AI large model targets to the Hong Kong market and provide an important reference for subsequent financing and valuation of AI enterprises.

Regardless of which path is chosen, the final competition will return to the comprehensive contest of technical strength, product experience, and commercialization efficiency. As MiniMax founder Yan Junjie said: ‘If the technology is done well, commercialization will come naturally.’ In the field of AI large models with rapid technological iteration, sustained technical investment and product innovation are the fundamental sources of enterprises’ long-term competitiveness.

---

[1] Sina Finance - ‘Going Overseas, Going Public: China’s First Batch of Large Models Have Survived’ (https://finance.sina.com.cn/stock/companyt/2026-01-03/doc-inheyxtw8478200.shtml)

[2] Beijing Municipal Government - ‘Zhipu Passes Hong Kong Stock Exchange Listing Hearing, Beijing Will Produce ‘the First Large Model Stock’’ (https://www.beijing.gov.cn/fuwu/lqfw/gggs/202512/t20251222_4355761.html)

[3] Securities Times - ‘Zhipu Rushes to Hong Kong IPO, Industry Leaders Compete for ‘World’s First Large Model Stock’’ (https://www.stcn.com/article/detail/3559175.html)

[4] Sina Finance - ‘To B Zhipu and To C MiniMax: Large Model Businesses Are Both Hard to Do’ (https://finance.sina.com.cn/roll/2025-12-24/doc-inhcxizc9649528.shtml)

[5] Jiefang Daily - ‘Top Long-Term Capital Heavyweights Invest in the World’s Youngest AI Team: MiniMax Will List in Hong Kong on January 9’ (https://www.jfdaily.com/news/detail?id=1044536)

[6] Huxiu - ‘Going Overseas, Going Public: China’s First Batch of Large Models Have Survived’ (https://www.huxiu.com/article/4822987.html)

[7] 36Kr - ‘Two ‘Chinese OpenAIs’ Queue for IPO’ (https://m.36kr.com/p/3606581768668417)