中国软件：生成式AI应用货币化——差异化功能、AI代理及定制知识中心推动付费率

The Top debates on AI software are centered around 1) Monetization, 2) Moat & IP,3) User cases, and 4) Cost. We have talked to 20+ companies and illustratedhow-to-use (exhibit 10-19), and addressedkey FAQs1) What is thein this report:progress on AI Monetization?Software vendors have started charging AI software,while looking for a balance between commercialization and new user expansion;2)What is AI pricing strategy?Annual subscription fee, trade-off betweencommercialization and user expansion;3) What are the top user cases?AI search,AI video creativity/ editing, AI productivity favored by ToC users, and Finance/ HR/Knowledge base/ Procurement are earlier in ToB AI applications;4) What are thesoftware vendors’ moats?Platform to streamline the working process byintegrating AI and software features, and accumulated industry know-how. Theemergence of multiple AI models in China has lowered training/ inferencing cost andcomputing power requirement, while we still see room for cost to further go downwith expansion of local AI infrastructure, making AI available to more users.In this note, we provide a summary snapshot of the monetization status of differentapplications and current capabilities.Read-across to China Software:We view Kingsoft Office (AI Productivity), Kingdee/Yonyou (AI Enterprise Services), Arcsoft (AI Photography), iFlytek (AI education/Healthcare), Glodon (AI BIM) early beneficiaries of AI monetization.1) What is the progress on AI Monetization:The MAU of single ToC AInapplication has reached over10m units3%~13%, with paying ratio at(Exhibit 2);The number of single AI software’s enterprise clients target atthis~1,000 unitsyear (Exhibit 3). Overall, the revenue contribution of AI software are still low thisyear, ranging from single digit to high teens, while we see software vendorsview AI as key growth drivers in next few years, with API token fee of AIfoundation model continuing to go down.2) What is the AI pricing strategy:Per the companies, ToC AI tools are priced atnUS$20~US$200/ user per year(Exhibit 1), and AI creation tools are pricedAllen Chang+852-2978-2930 |allen.k.chang@gs.comGoldman Sachs (Asia) L.L.C.Verena Jeng+852-2978-1681verena.jeng@gs.com|Goldman Sachs (Asia) L.L.C.Ting Song+852-2978-6466ting.song@gs.com|Goldman Sachs (Asia) L.L.C.China Software: Gen-AI apps monetization: Differentiated features, AIagents, and customized knowledge hubs to drive paying ratioGoldman Sachs does and seeks to do business with companies covered in its research reports. As a result,investors should be aware that the firm may have a conflict of interest that could affect the objectivity of thisreport. Investors should consider this report as only a single factor in making their investment decision. For Reg ACcertification and other important disclosures, see the Disclosure Appendix, or go towww.gs.com/research/hedge.html. Analysts employed by non-US affiliates are not registered/qualified as researchanalysts with FINRA in the U.S.China Software: Gen-Al apps monetization: Differentiated features,Alagents, and customized knowledge hubs to drive paying ratioChina Software</Verena JenThe Top debates on Al software are centered around 1) Monetization, 2) Moat & IPTing Son3) User cases, and 4) Cost. We have talked to 20+ companies andillustratedhow-to-use (exhibit 10-19), and adclressed key FAQs in this report: 1) What is theprogress on Al Mlonetization? Software vendors have started charging Al software,while looking for a balance between commercialization and new user expansion; 2)What is Al pricing strategy? Annual subscription fee, trade-off betweencommercialization and user expansion; 3) what are the top user cases? Al search,Al video creativity/ editing, Al productivity favored by ToC users, and Finance/ HR/Knowledge base/ Procurement are earlier in ToB Al applications; 4)What are thesoftware vendors' moats? Platform to streamline the working process byintegrating Al and sofware features, and accumulated industry know-how. Theemergence of multiple Al models in China has lowered training/ inferencing cost andcomputing power requirement, while we still see room tor cost to turther go downwith expansion of local Al infrastructure, making Al available to more users.In this note, we provide a summary snapshot of the monetization status of differentapplications and current capabilities.Read-across to China Software: We view Kingsoft Office (Al Productivity), Kingdee/Yonyou (Al Enterprise Services), Arcsoft (AI Photography), iFlytek (Al education/Healthcare), Glodon (Al BIMI) early beneficiaries of Al monetization.1)What is the progress on Al Monetization: The MIAU of single ToC Alapplication has reached over 10m units, with paying ratio at 3%~13% (Exhibit 2);The number of single Al software's enterprise clients target at ~1,000 units thisyear (Exhibit 3). Overall, the revenue contribution of Al software are sill low thisyear, ranging from single digit to high teens, while we see software vendorsview Al as key