速递 | 智谱AI招股书大揭秘：中国AI公司的生存真相

Core Viewpoint - The article discusses the IPO prospectus of Zhipu AI, highlighting its unique business model and financial challenges, emphasizing the differences between Chinese and American AI companies in terms of market approach and operational difficulties [1][4][42]. Group 1: Financial Performance - Zhipu AI reported revenues of over 57 million in 2022, projected to exceed 1 billion in 2023, and over 3 billion in 2024, indicating a compound annual growth rate of 130% [8]. - Despite the revenue growth, Zhipu AI is expected to incur losses close to 2 billion in 2024, meaning for every 1 yuan earned, approximately 7 yuan is spent [9][10]. - The company’s significant losses are primarily attributed to costs in computing power and research and development [14]. Group 2: Business Model - Zhipu AI's revenue model is centered around "private deployment," where the entire model system is installed on the client's servers, contrasting with OpenAI's subscription-based model [19][20]. - The starting price for basic cloud private deployment is 250,000 yuan per year, with local private deployment costing more [21][22]. - The majority of Zhipu AI's revenue currently comes from these "heavy asset" private deployment projects, although the company aims to increase the proportion of API cloud service revenue to 50% [28]. Group 3: Market Dynamics - The urgency for Zhipu AI to go public is driven by a limited market window, as competitors like MiniMax are also pursuing IPOs [30][31]. - The AI industry is characterized by rapid changes, where companies must act quickly to avoid being left behind, similar to the competitive landscape seen in the past with the "group buying war" [38]. - Zhipu AI faces intense competition, particularly in the API subscription market, where pricing pressures from competitors like DeepSeek are significant [39]. Group 4: Industry Insights - The article highlights the fundamental differences between Chinese and American AI companies, noting that Chinese firms must balance research and profitability while dealing with price wars and data security concerns [42]. - Zhipu AI has a high proportion of R&D personnel at 74%, indicating a strong focus on development, but the company still faces challenges in producing competitive models at lower costs [43]. - The pursuit of an IPO is seen as a means for Zhipu AI to secure further capital, but the sustainability of its business model post-IPO remains a critical concern [45][46].