高盛：中国生成式AI应用的商业变现之路（英文）

Core Insights - The report by Goldman Sachs focuses on the monetization of generative AI applications in China, discussing key debates around monetization, competitive advantages, user cases, and costs [1][6][16]. Monetization Progress - Monthly active users (MAU) for single consumer-facing AI applications have surpassed 10 million, with a paying ratio between 3% and 13% [9][22]. - The target for enterprise clients for single AI software is approximately 1,000 this year, although revenue contributions from AI software remain low, ranging from single digits to high teens [9][15]. Pricing Strategies - Consumer-facing AI tools are priced between $20 and $200 per user annually, while enterprise-level applications charge between $80 and $1,000 per user per year, or $3,000 to $20,000 annually [2][10]. - To attract users, consumer vendors typically offer trial periods of 7 to 30 days, while enterprise vendors provide pre-training series for large-scale clients [10][16]. User Cases - AI applications are categorized into four main segments: AI creation (image, video, audio generation), AI productivity (data analysis, scheduling, coding), AI industry tools (education, healthcare, architecture), and AI enterprise services (finance, HR, supply chain, sales) [2][10][30]. - The emergence of multiple foundational models allows AI applications to generate content through deep reasoning, with AI agents further optimizing processes [10][30]. Competitive Advantages - Software vendors' competitive edge lies in offering differentiated features that users are willing to pay for, especially when general AI assistants are available for free [2][10]. - Enterprises require AI tools that can perform complex tasks and deliver higher ROI, leading to a gradual adoption starting from senior management to all employees [17][18]. Cost Dynamics - The cost of AI deployment has decreased, with the API token fee for the DeepSeek AI model dropping to $0.6 per million tokens, compared to OpenAI's pricing of $1.6 to $8.0 per million tokens [3][16]. - The expansion of local AI infrastructure is expected to further reduce costs, making AI technology more accessible to a broader user base [1][6].