Customer Concentration Risk Assessment Report for Zhipu AI

Based on the prospectus and market data, I will conduct a systematic customer concentration risk assessment for you.

---

Based on data disclosed in Zhipu AI’s post-hearing materials for Hong Kong Stock Exchange listing[1], customer concentration exhibits the following characteristics:

Indicator	2024 Data	H1 2025 Data	Risk Rating
Revenue Share of Largest Customer	19%	Over 10%	⚠️ Medium-High Risk
Revenue Share of Top 5 Customers	45.5%	Consistently Over 40%	🔴 High Risk
Total Number of Institutional Customers	-	3,156 (65x growth from 2022)	📈 Improving
Customer Concentration Ratio (CR5)	45.5%	>40%	Above Industry Average

---

[1][2]:

Customer Type          2022    2023    2024    H1 2025
─────────────────────────────────────────────────────────
Internet & Technology      55.2%     67.6%     49.6%     38.3%
Public Services          Declining Trend  Continuous Growth  ~30%   ~30%
Telecom Industry          Stable      Stable      Stable      13.6%
Others              -         -         -         -

:

1. Key Customer Dependency Risk
   : The top 5 customers contribute 45.5% of revenue, meaning
   the loss of any major customer could cause severe revenue volatility
   [1]

2. High Customer Turnover
   : The prospectus shows that
   there is almost no overlap in the list of top 5 customers
   , reflecting the issue with one-time delivery models - revenue is unsustainable, requiring constant acquisition of new customers[1]

3. Customer Structure Transformation Pressure
   : The revenue share of internet and technology customers has continued to decline (from 67.6% to 38.3%), and these customers have the strongest willingness and most mature technical capabilities to develop their own models[2]

---

Zhipu AI adopts the

, which is specifically divided into[3]:

Delivery Model	Revenue Share (H1 2025)	Customer Characteristics	Impact on Concentration
On-Premises Deployment	~85%	Large Enterprises, Government Agencies	High Customer Value, High Customization
Cloud Deployment	15.2%	SMEs, Developers	Low Customer Value, High Frequency

:

• On-premises deployment has a long expansion cycle and high degree of customization
• Difficult to achieve explosive growth in the short term[1]
• Gross margin of cloud deployment business has turned from positive to negative (-0.4%), with ongoing price war pressure[3]

---

Comparison with major players in the large AI model industry:

Company	Customer Concentration Characteristics	Business Model	Risk Differences
Zhipu AI	Top 5 Customers: 45%+	B2B Focused (MaaS)	High Concentration
MiniMax	Relatively Dispersed	C2C Focused (Subscription)	Dispersed
Baidu/Aliyun	Relatively Dispersed	Platform-Based	Low

: In 2024, the size of China’s large language model market reached RMB 5.3 billion, with institutional customers contributing RMB 4.7 billion (89%) and individual customers only RMB 0.6 billion[3]. Zhipu’s choice of a B2B strategy aligns with the market structure, but it faces the inherent customer concentration challenges of the B2B sector.

---

Positive Factors	Analysis
Rapid Growth in Customer Base	Number of institutional customers increased from 48 in 2022 to 3,156[2], with concentration dispersion improving
Optimization of Customer Structure	Increased revenue share from public services and telecom customers, diversifying customer sources
Overseas Business Expansion	Overseas revenue accounted for 11.6% in H1 2025[1], reducing reliance on a single market
MaaS Platform Advantages	API platform has over 2.9 million developer users, forming a long-tail effect[3]

---

Risk Dimension	Rating	Explanation
Concentration Risk	🔴 High	Top 5 Customers: >40%, single dependency on largest customer
Customer Churn Risk	🔴 High	High annual customer turnover rate, one-time delivery model
Bargaining Power Risk	🟡 Medium	Major customers have strong bargaining power
Decentralization Progress	🟢 Improving	Rapid growth in customer base, ongoing structure optimization

: ⚠️

Zhipu AI’s customer concentration risk requires continuous attention. Although the number of customers is growing rapidly (65x growth in the previous two years), revenue remains highly concentrated among a small number of major customers. If key customers churn or reduce their budgets, it will have a significant impact on the company’s revenue.

---

1. Key Monitoring Indicators
   : Changes in top 5 customers, number of new/churned customers and their revenue contributions, customer retention rate
2. Focus on Inflection Points
   : When CR5 drops below 30% and annual customer overlap rate increases, the risk will decrease significantly
3. Risk Hedging
   : Pay attention to the company’s cloud business expansion and overseas market growth to reduce reliance on B2B concentration

---

[1] Yicai Global - “Going Overseas, Going Public: China’s First Batch of Large Models Finally Break Through” (https://m.cbndata.com/information/294851)

[2] 36Kr - “At the IPO Door, China’s Large Models Finally Settle the Books” (https://m.36kr.com/p/3617797772461057)

[3] QbitAI - “Urgent Analysis of Zhipu AI’s Prospectus: Annual Revenue of RMB 300 Million with 130% Growth Rate” (https://www.qbitai.com/2025/12/362256.html)

[4] Caixin - “Zhipu AI Switches to Hong Kong Stock Exchange Listing and Passes Hearing; H1 Revenue RMB 191 Million, Net Loss RMB 2.3 Billion” (https://companies.caixin.com/2025-12-20/102395298.html)