Baidu (09888.HK, NASDAQ: BIDU) has released its financial report for the second quarter of 2024, showing that the company achieved a total revenue of 33.9 billion yuan for the quarter, with Baidu's core revenue at 26.7 billion yuan; the net profit attributable to Baidu was 5.5 billion yuan, and the net profit attributable to Baidu's core was also 5.5 billion yuan, with a net profit margin for Baidu's core of 20%.
Baidu stated that in the quarter, the continuous acceleration of its intelligent cloud business offset the ongoing macroeconomic headwinds faced by online marketing revenue, bringing a slight positive growth to Baidu's core revenue.
The financial report indicates that out of Baidu's core revenue of 26.7 billion yuan, online marketing revenue accounted for 19.2 billion yuan; non-online marketing revenue was 7.5 billion yuan, a year-on-year increase of 10%, mainly driven by the intelligent cloud business.
Furthermore, in the second quarter of 2024, Baidu's intelligent cloud business revenue reached 5.1 billion yuan, a year-on-year increase of 14%. The growth is primarily due to the increased demand for AI-native applications and Baidu's leading position in AI infrastructure and model training platforms.
Advertisement
During the earnings call, the group's founder, chairman, and CEO, Robin Li, also disclosed the latest data for the Wenxin Yidian large model, which currently has an average daily usage of over 600 million calls and processes about 1 trillion Tokens of text per day, both of which are the highest in China. Compared to the average daily usage of 50 million calls for the Wenxin large model announced in Q4 of 2023, it can be seen that this number has grown more than 10 times in half a year.
A research report issued by Tianfeng Securities stated that Baidu is actively promoting the application of large models in different scenarios, fully exploring the potential capabilities of large language models, and driving incremental demand through enterprise applications. At the same time, the progress of applications can drive innovation in the basic models, complementing each other to promote model iteration, with the potential for further development in the future.
AI Drives Accelerated Growth of Baidu's Intelligent Cloud Business
Baidu's intelligent cloud revenue continues to accelerate its growth this year, with a 12% increase in the first quarter. The financial report shows that in the second quarter of 2024, Baidu's intelligent cloud revenue increased by 14% year-on-year.
"The main source of Baidu's intelligent cloud revenue is the rapidly growing revenue related to generative AI and foundational large models," said Baidu's management during the financial report call, indicating that this part of the revenue has grown from about 5% of the total intelligent cloud revenue last quarter to nearly 9% in this quarter. To meet the diverse needs of customers, Baidu offers a variety of products and solutions. For customers requiring model training and hosting services, Baidu also provides the most cost-effective and powerful AI infrastructure.Financial reports indicate that Baidu released the Wenxin Large Model 4.0 Turbo in June to expand the richness of the Wenxin series models. Compared to the Wenxin Large Model 4.0, the Turbo is designed to operate faster and at a lower cost. Recently, Baidu also reduced the prices of its two flagship models, Wenxin Large Model 3.5 and 4.0, and made ERNIE Speed, ERNIE Lite, and ERNIE Tiny models freely available.
Baidu's management believes that the generative AI and large model market are still in their early stages. Baidu is educating the market to make AI more accessible. Therefore, Baidu not only focuses on the capabilities and ease of use of the models but also on whether the prices are affordable for users. To encourage usage, Baidu has introduced many preferential policies, leading to a rapid increase in ERNIE API calls, from about 200 million per day in mid-May 2024 to over 600 million per day now, which translates to generating more than 1 trillion tokens (units of AI-generated content) per day.
An article recently published by IDC shows that the market size for China's large model platforms and related applications in 2023 reached 1.765 billion yuan, with Baidu's Intelligent Cloud reaching a market size of 350 million yuan in 2023, ranking first in the market with a market share of 19.9%.
Baidu stated that its Intelligent Cloud business is healthier than ever, has been steadily generating non-GAAP operating profits, and is improving profit margins. While continuously enhancing the capabilities of the ERNIE model, Baidu is also reducing inference costs, leveraging scale effects as a market leader, and believes it can continue to expand market share. "Ultimately, we believe that the profit margins of generative AI-related cloud services should exceed those of traditional cloud services."
Baidu also revealed that it will continue to launch upgraded flagship foundational large models, expanding competitive advantages in an application-driven manner to differentiate Baidu from other companies.
Intelligent entities demonstrate commercial value. For internal business, Baidu has been promoting the AI transformation of its existing business lines. For example, Baidu Search is becoming the largest entry point for intelligent entity distribution. Intelligent entities built on the Wenxin Intelligent Entity Platform AgentBuilder can be distributed within Baidu Search and other ecosystems to enrich the content ecosystem and improve customer efficiency.
"Starting from the second quarter of 2024, we accelerated the use of ERNIE models to improve search speed. The main goal is to enhance user experience, provide more features, and increase user engagement on Baidu." Baidu's management revealed that currently, 18% of search results on Baidu Search are provided by generative AI, and these results are more comprehensive and relevant than traditional searches, offering information that was previously unattainable or scattered in different places. Firstly, generative AI can provide unlimited content on demand; secondly, when the generated AI search results have met the user's needs, Baidu will recommend other content or services (similar to information streams) to increase the time users spend on Baidu and their stickiness; thirdly, the new interactive features of Baidu Search allow users to refine questions and follow up through multiple rounds of dialogue, making it more natural and efficient for users to express their needs.
It is reported that an increasing number of users are now posing complex questions on Baidu, seeking comprehensive and in-depth answers. In addition, some users have begun to use Baidu to create content, such as writing and editing text, images, etc. "We are still in an exciting but not fully developed early stage," said Baidu's management.Additionally, Baidu is accelerating the distribution of intelligent agents in search results to provide users with intelligent assistants. Currently, the distribution volume of intelligent agents within the Baidu ecosystem is rapidly increasing, with an average daily distribution exceeding 8 million times in July, double that of May. The most commonly used intelligent agents include types such as content creation, personality tests, and schedule planning. Users, developers, service providers, and merchants are all important participants in the intelligent agent ecosystem. Merchants developing intelligent agents on Baidu have reached 16,000, covering industries such as education, law, and B2B.
For example, recently a lawyer created and published an intelligent agent on the platform. A user who had a conflict with law enforcement officers sought advice from this intelligent agent, resulting in 288 rounds of dialogue that lasted for 5 hours.
"It is hard to imagine a real lawyer having the time and patience to engage in such a long conversation with a potential client," said Baidu's management. An increasing number of advertisers, especially in the legal, education, and B2B sectors, are beginning to use intelligent agents. These fields typically require a longer decision-making time and personalized pre-sales consultation, and the interactive functions of intelligent agents are increasingly meeting these needs. "As intelligent agents become smarter, they have the potential to drive direct sales, bringing more value to advertisers, thus allowing us to transition from a cost-per-click (CPC) model to a more effective cost-per-sale (CPS) model."
To date, Baidu's marketing platform has over 16,000 active merchant intelligent agents providing services daily, with more than 2.5 million conversations per day, covering over 30 industries. After using merchant intelligent agents, the customer conversion rate has increased by 26.8%. Wang Fengyang, Vice President of Baidu Group and Head of the Mobile Ecosystem Business System, stated that market recognition has made Baidu's marketing even more confident that intelligent agents will become the new generation of business operation positions.
Robin Li stated that new features such as intelligent agents have not yet been commercialized, but they are a transformative force in search, enabling Baidu to seize substantial long-term growth opportunities, change the way search is used, and consolidate Baidu's leadership in the AI search field. He emphasized that intelligent agents will become the "new website ecosystem" of the AI era.
Leave A Reply