The company's recent situation
On July 6, 2024, the company held a forum at the 2024 World Artificial Intelligence Conference, showcased the self-developed large model product matrix and released product updates around the theme of “New AI, New Application, and New Ecology”. At the same time, the company shared investors' concerns about the big model technology path, commercial application, and future iteration prospects. In addition, the company recently released GMV performance data for the full cycle of the 618 shopping festival (May 20 to June 30). Pan-shelf GMV increased 75% year over year, search GMV increased 110% year over year, and short video trailer GMV increased 69% year over year.
reviews
A new version of the Keling Big Model was released, and Kuaishou's self-developed Big Model product matrix was unveiled. The company's self-developed Happy Italian language model, the “graphic” literary chart model, and the “smart” video generation model were unveiled at the conference. Among them, the Keling Master Model announced a version update: 1) the first basic model upgrade after launch to achieve higher video generation quality; 2) Improved video control capabilities. The new version supports the first and last frame control of video generation, and supports lens scheduling control; 3) Enhances ease of use. The Keling Master Model is online on the web and a free application for a limited time, and the integrated AI image, AI video and video editing functions on the web provide users with one-stop AI video editing capabilities. Ketu Big Model also showcased the world's leading AI image generation capabilities at the conference, and it was announced that it will be officially open source starting today (July 6).
The multiple dimensions of content, community ecology, and commercialization are expected to benefit from large model capabilities. In terms of content and community, the company said that the first AIGC short drama “Shanhai Mirrors: Splitting the Waves” based on the Tu and Ling Da models will be launched within the month; at the interactive level, the company has launched an “AI Xiaokuai” intelligent review robot.
In terms of commercial applications, the company showcased three commercial products based on large model capabilities: 1) Magnetic Innovation:
Used for one-stop material editing to improve the productivity of advertising materials and reduce output costs; 2) Nu Ewa Digital Human: We believe that this digital human product has achieved the ability to express and interact commercially; 3) Pi Digital Staff: AI-driven intelligent customer service, which we believe can now better handle basic Q&A and lead form collection functions. We believe that Kuaishou has data and scenario advantages for AI applications. With the iteration of self-developed large-scale model product capabilities, Kuaishou's AI-driven commercial products are also continuously improving usability and commercial effectiveness.
Pan-shelf performance was outstanding during the “618” period. Focus on GMV's progress in the second quarter and the whole year. According to Kuaishou e-commerce marketing observations, we believe that the overall performance of the “618” GMV market is lackluster, judging that the company's GMV growth rate may be slightly lower than previously anticipated. Looking at the whole year, we still expect core advertising and e-commerce business revenue to achieve a year-on-year growth rate of more than 20%.
Profit forecasting and valuation
Considering the company's performance during the “618” period, we lowered the 2Q24 and 2024 GMV forecast year-on-year growth rate from 25%/25% to 20%/22%; however, considering the company's commercial efficiency advantages, we maintained the 2Q24 and 2024/2025 profit forecasts. Maintaining an outperforming industry rating, considering the lackluster overall consumption and GMV expectations, the target price was lowered by 12% to HK$66, corresponding 14/13 times the 2024/2025 non-IFRSP/E, with 47.5% upside compared to the current stock price.
risks
Competition in the industry intensifies, large-scale model iterations fall short of expectations, commercial growth falls short of expectations, risk of losing users, higher investment in new businesses than expected, and content and regulatory risks.