On the evening of May 16, Baidu Group announced its financial report for the first quarter of 2024. In the first quarter, Baidu achieved a total revenue of 31.5 billion yuan and a core revenue of 23.8 billion yuan; The net profit attributable to Baidu (Non-GAAP) was 7.011 billion yuan, a year-on-year increase of 22%, and the net profit attributable to Baidu's core (Non-GAAP) was 6.628 billion yuan, a year-on-year increase of 26%.
Source: Baidu Group's 2024 Q1 performance report
Robin Li, founder, chairman and CEO of Baidu, said on the earnings call that Baidu's intelligent cloud revenue in the first quarter increased by 12% year-on-year to 4.7 billion yuan, of which AI accounted for 6.9%. Baidu's end-to-end optimization of the AI technology stack continued to drive intelligent cloud revenue growth in the quarter.
Robin Li also mentioned the key plans for the next AI search and autonomous driving business. "Search is most likely to be the 'killer app' in the AI era, and Baidu plans to accelerate the rollout and adoption of new product features, such as generating multimodal search results that enable multiple rounds of interaction." Robin Li said.
Luo Rong, Baidu's chief financial officer, said: "In the first quarter of 2024, Baidu's core operating margin (non-GAAP) reached 23.5%, an increase sequentially higher than the previous year. We have established a track record of continuous operational efficiency improvements and will enhance operational efficiency in the coming quarters to support the AI business and high-quality growth, as well as maintain healthy operating margins. ”
"The most important thing is AI"
In the first quarter, the online marketing revenue of Baidu's core business remained solid. According to the financial report, Baidu's core revenue in the first quarter was 23.8 billion yuan, a year-on-year increase of 4%. Among them, online marketing revenue was 17 billion yuan, a year-on-year increase of 3%; Non-online marketing revenue was 6.8 billion yuan, a year-on-year increase of 6%.
Robin Li said on the earnings call that Baidu's intelligent cloud revenue increased by 12% year-on-year in the first quarter. "The cloud performed well in this quarter, and the most important thing is because of AI, Baidu basically lowered the development threshold of AI applications this quarter. Our end-to-end optimization of our AI stack continued to drive intelligent cloud revenue growth in the quarter. Robin Li said.
In the first half of this year, Baidu launched three development tools, AgentBuilder, AppBuilder and ModelBuilder, and released three lightweight models with "price-effectiveness ratio" on the basis of versions 3.5 and 4.0 of the Wenxin large model flagship version.
According to Baidu, these tools help developers and enterprises create agents, develop AI-native applications and custom models, lowering barriers to entry and improving efficiency. As of mid-April, there were 13 million developers in the PaddlePaddle community.
In the first quarter, Baidu launched and improved MaaS platform tools for enterprise users. According to Baidu, the tools are designed to facilitate the development of AI-native applications, agents, and the creation of customized models on Baidu's MaaS platform.
"The daily call volume of the Wenxin large model has reached 200 million, and the inference cost has dropped to 1% a year ago. More and more enterprises are conducting model training, fine-tuning, and AI-native application development in Baidu's public cloud, and following Samsung and Honor, OPPO, vivo, Xiaomi, Lenovo Computer, and NIO all connected to the Wenxin Large Model API in the quarter. Robin Li said that with the improvement of the coverage of intelligent terminals, the user scale of the Wenxin model will also be further expanded.
Another growth driver for AI cloud is Baidu's cross-selling of CPU cloud services to GPU cloud customers. Robin Li said that with the high recognition of GPU cloud among existing and new customers, customers are increasingly switching their CPU cloud usage to Baidu public cloud.
Baidu has repeatedly emphasized that "closed-source large model + public cloud" has become the mainstream trend of the global AI market. Model inference will become a key driver of cloud business, and we will continue to optimize the cost of model inference and launch a variety of models and tools to lower the threshold for enterprises and individuals to develop applications, create agents, and customize fine-tuned models.
We plan to enrich the AI search function
Baidu will speed up the adjustment of AI to the company's basic business, Baidu Search.
According to the financial report, as of March 2024, the monthly active users of Baidu App reached 676 million, a year-on-year increase of 3%.
Robin Li said that Baidu is gradually improving AI-generated search results, and has added an "interactive" column to completely change to question-and-answer search, and currently 11% of Baidu's search results are generated by AI.
"On the consumer side, our goal is to build our proprietary AI-native applications that create new growth opportunities. Baidu Wenku's paying subscribers experienced double-digit year-over-year growth in the first quarter, but Baidu's search growth took longer than expected. Robin Li admits that this is because hundreds of millions of search users are very cost-sensitive.
"AI-generated search results have not yet been commercialized and will have an impact on ad revenue in the short term, but in the long term, it will continue to improve the user experience and increase search usage." Robin Li revealed that Baidu is optimizing its strategy and plans to accelerate the launch and adoption of new product features, such as generating multi-modal search results that enable multiple rounds of interaction. Another trend is the distribution of agents on Baidu Search, which may be the "mini program" of the AI era, which can improve the user experience and improve the efficiency of merchants in the short term. In the long run, the ecosystem formed by agents may make Baidu a super entrance.
Robin Li said that search is most likely to become a "killer app" in the AI era, and AI can generate more accurate and direct results, so that users can be satisfied in one step. Whether Baidu can gain a foothold in the AI era, how to combine AI with Baidu's basic disk search business, and truly enhance the user experience of search is crucial.
In the field of autonomous driving, Radish Express is an autonomous driving travel service platform under Baidu Apollo. Wuhan Radish Express is expected to fully enter the profitable period in 2025. According to the financial report, as of April 19, the cumulative number of service orders of Radish Express exceeded 6 million; In the first quarter, about 826,000 rides were provided, a year-on-year increase of 25%. The operating area of Carrot Express in Wuhan has more than doubled, with more than 55% of the total unmanned driving in the first quarter, and achieved all-weather operation in early March.
Baidu released the Apollo ADFM, a large autonomous driving model at Apollo Day on May 15, which can support L4 autonomous driving applications and is the core technical foundation for Baidu to achieve city-level autonomous driving.
Baidu said that through the application of the autonomous driving model Apollo ADFM, Baidu's pure vision city pilot assisted driving product ANP3 has been upgraded to ASD (Apollo Self-Dirving), which will be mass-produced and launched on all Jiyue cars, taking the lead in realizing intelligent driving nationwide. In addition, Baidu has also launched an unmanned vehicle equipped with Apollo's sixth-generation autonomous driving system solution, which will operate around the clock in Wuhan.
According to Baidu, by the end of 2024, Radish Express will break even in Wuhan and fully enter a profitable period in 2025. Carrot Express may become the world's first commercially profitable autonomous driving travel service platform.
"Basic models such as the Wenxin model will become infrastructure, and we are making the Wenxin model series more affordable and efficient. This will bring more opportunities for Baidu. Robin Li said.
Tianfeng Securities said that the focus of Baidu's cloud business additional revenue increment has transitioned from model training to model inference, and the significant reduction in inference costs is expected to drive the development of cloud business. With the further enrichment of the company's road test mileage data, the area and population covered by the company's radish run business are increasing, and the company's intelligent driving technology is expected to usher in a paradigm breakthrough moment.
Reviewer: Xiong Yonghong Editor: Li Ruoyu Jiao Yuanyuan Proofreader: Zhang Jing Signed: Peng Yong
