Baidu executives interpret Q1 financial report: cloud industry price cuts have minimal impact on Baidu AI cloud business

Baidu (Nasdaq: BIDU; HKEX: 9888) today released its financial report for the first quarter of 2024 as of March 31: total revenue was 31.5 billion yuan, a year-on-year increase of 1%

Special topic: Baidu’s first quarter revenue was 31.5 billion yuan and adjusted net profit was 7 billion yuan. Baidu (Nasdaq: BIDU; HKEX: 9888) today released its financial report for the first quarter of 2024 as of March 31: total revenue was 31.5 billion Yuan. A year-on-year increase of 1%.
Net profit attributable to Baidu was 5.4 billion yuan.
Not according to U.S. GAAP, net profit attributable to Baidu was 7 billion yuan, a year-on-year increase of 22%.
, For details, see: Baidu’s first quarter revenue of 31.5 billion yuan, adjusted net profit of 7 billion yuan. See the picture: Baidu’s first quarter revenue of 31.5 billion yuan, Baidu App monthly active users of 676 million, after the financial report was released. Baidu Chairman and CEO Robin Li. CFO Luo Luo Shen Dou, president of Rong. Intelligent Cloud Business Group, and other senior executives attended the subsequent financial report conference call. They interpreted the key points of the financial report and answered questions from analysts.
The following analysis is the main content of the Q&A session: Citibank analyst Alicia Yap: Can you ask the management to talk about the help of artificial intelligence technology in improving business realization? What kind of feedback is there so far from advertisers who have adopted the system? Where do they see the greatest improvement? What areas can be further enhanced? Robin Li: As you know, Baidu is the first company in the industry to commercialize generative artificial intelligence technology. It creates hundreds of millions of yuan in incremental revenue for the company every quarter.
Since the second half of last year, we have been using Wen Xinyiyan to upgrade the commercial monetization system. Strengthening all aspects of the technology, including improving and adding functions, improving the option system, and formulating strategies for automated creative generation of ads by advertisers, etc., to improve Conversion rates and sales.
The move prompted advertisers to increase spending on Baidu’s platform.
Incremental advertising revenue related to artificial intelligence technology increased sequentially in the first quarter. We also expect this trend to continue.
These incremental revenues have alleviated the impact of macroeconomic weakness. They have also bought us time to rebuild our client products using Wen Xin Yi Yan.
As I emphasized in my opening remarks, we believe Wen Xinyiyan can provide significant long-term growth opportunities for the company\’s online marketing business.
Advertisers can use natural language multi-turn dialogue systems to establish online connections and interact with potential customers.
By using the Wenxin Agent Platform (Agent Builder), our advertisers can easily create customized Wenxin Yiyan Agents.
When advertisers express their intentions to the agent, the agent can more effectively achieve their goals. Whether it is helping potential customers understand their products or improving the quality of customer service.
Agents can also enrich our own content offerings and improve users\’ experience on Baidu.
Although the service is still in its early stages, even so, it has already helped some advertisers achieve better delivery results.
For example, we have a customer in the online education industry who used the Wenxinyiyan agent platform to create his own artificial agent, and added important content such as product introduction and professional knowledge to it.
By constantly providing feedback and making improvements, the agent has been able to provide high-quality consultation around the clock, significantly improving the quality of the company\’s online customer service. The adoption of the agent drove a 20% increase in the online education company\’s ad conversion rate .
I think this is just the beginning. We believe that agents will become the main content and service form of the new artificial intelligence era. Baidu will continue to improve the functions of Wenxinyiyan agents.
Intelligent agents will not only be used to improve user experience and improve advertisers\’ conversion rates and effects. Over time, it will also promote an increase in platform transaction volume. And this shift should also help us shift from traditional pay-per-click (CPC) ) model to a more efficient Charge Based on Sales (CPS) model.
Morgan Stanley analyst Gary Yu: I have a question about Baidu’s artificial intelligence cloud business.
What impact will the price cuts of some companies in the industry have on Baidu\’s artificial intelligence cloud business? As competition intensifies, how should we view the profitability of the company\’s cloud business? How do you view the prospects for sustainable growth of the cloud business in the future? Shen Dou: As we have already mentioned, generative artificial intelligence and basic models are transforming the cloud industry from general computing to artificial intelligence computing.
This shift is reshaping the competitive landscape of the cloud industry and provides us with a valuable opportunity to establish our leading position in the field of artificial intelligence cloud.
We provide the domestic market with the most efficient artificial intelligence infrastructure and the most advanced large-scale model training and inference platform.
Therefore, we see more and more enterprises choosing to conduct model training, fine-tuning and development of artificial intelligence native applications on our public cloud.
In fact, this growing demand has significantly boosted the company\’s artificial intelligence cloud revenue since the second half of last year.
Baidu\’s artificial intelligence cloud revenue has grown from a year-on-year decline in the third quarter of last year to an 11% increase in the fourth quarter of last year. In the first quarter of this year, revenue growth further accelerated to 12%.
There are two main factors behind the revenue growth: first, the incremental revenue directly generated by generative artificial intelligence and basic models. The second is the new opportunities brought by our traditional cloud business.
As Robin Li just mentioned, in the first quarter of this year, revenue from generative artificial intelligence and basic models accounted for 6.9% of our artificial intelligence cloud revenue. And our traditional CPU cloud business is leveraging generative artificial intelligence and basic models to bring Opportunity.
Both factors are important growth drivers for our AI cloud business.
In addition, although the intelligent transportation business is still tepid, its impact on the overall artificial intelligence cloud business in the first quarter was much smaller than in previous quarters.
So. Overall, we expect the AI ​​cloud business to continue to benefit from this trend and maintain strong revenue growth momentum in the coming quarters.
In terms of profit, under non-GAAP, as in previous quarters, the artificial intelligence cloud business continued to generate operating profits for the company in the first quarter. We are also committed to achieving sustainable and healthy revenue growth.
This quarter, we have been focused on delivering high-quality revenue growth by cutting back on low-margin businesses.
The business of generative artificial intelligence and basic models is still in a very early stage. In the long term, our focus is still on popularizing relevant knowledge, introducing our products to more enterprises, and increasing product adoption rates.
We also expect that the normal profit margins of generative artificial intelligence and basic models and related businesses will further improve and be higher than our traditional cloud business.
Regarding the changes in pricing policies of some competitors you just mentioned. In fact, it is very common for cloud service providers to adjust the pricing of certain products. This is also a trend we have observed many times in the past. And our cloud services It has expanded from traditional CPU cloud to high-value artificial intelligence products and services. Pricing changes in the CPU cloud industry have minimal impact on the development of Baidu\’s artificial intelligence cloud business.
Utilizing the company\’s unique four-layer artificial intelligence architecture and powerful end-to-end optimization capabilities, Baidu\’s cloud platform service has now reduced Wen Xinyiyan\’s inference cost to 1% of the version in March last year.
Just this month. Wen Xinyiyan can already handle about 200 million API call requests or about 250 billion tokens every day.
We believe that Wenxinyiyan\’s increasing adoption rate will continue to improve its performance, improve efficiency, and then further reduce costs.
It is also important that we evaluate the price/performance of different models under different workloads and not just focus on surface price.
Therefore, we believe that Baidu’s most advanced artificial intelligence infrastructure and large-scale platform can provide customers with the best value.
Looking to the future, our powerful artificial intelligence capabilities will continue to attract new customers and encourage existing customers to increase their spending on Baidu Artificial Intelligence Cloud. Our goal also includes continuing to create positive operating profits under non-GAAP accounting principles.
JP Morgan analyst Alex Yao: In view of the domestic chip shortage, how can Baidu enhance its leading edge in China\’s large language model technology while fulfilling its commitments and providing differentiated value? Robin Li: In this regard, our ideas are indeed different.
We adopt an application-driven approach to promote the development of artificial intelligence.
For example, using the most advanced large language model to solve high school mathematics problems may not be the focus of our current work. However, being able to convincingly provide correct suggestions for users to purchase products is more important for many applications.
With this in mind, Baidu is leveraging its unique four-layer artificial intelligence technology stack to optimize and improve the cost and performance of the Wenxinyiyan model.
And we ensure that customers and developers can use tools such as Wenxin agent platform Agent Builder, Qianfan light application production tool App Builder and Qianfan large model platform Model Builder to easily build applications.
At the artificial intelligence infrastructure layer, our excellent capabilities in GPU cluster management are one of the key factors to improve the efficiency of model training and inference.
We have recently achieved some breakthroughs. The ability to integrate GPUs from different vendors into a large-scale unified computing cluster allows us to use less advanced chips for efficient model training and inference.
Our deep learning framework PaddlePaddle continues to help reduce the cost of model training and inference through continuous innovation and enhancement.
PaddlePaddle is compatible with more than 50 different chips. Many of them are domestically designed. Its developer community has grown to 13 million people. And Wenxinyiyan versions 3.5 and 4.0 remain the flagship models for complex tasks.
By launching lightweight large language models, introducing model development and application development toolkits, and applying hybrid expert models (MoE) in model inference, we have made accessing Wenxin Yiyan simpler, lower cost, and higher in performance.
Adhering to the purpose of driving applications. We have used Wenxinyiyan extensively to innovate our own products. We have also gained experience and industry insights in training and using this tool to develop artificial intelligence native applications.
We are making all of these capabilities available to customers and developers.
Through all these efforts, we strive to cultivate a vibrant and healthy ecosystem around the Wenxinyiyan system.
Baidu is taking a holistic approach to developing general artificial intelligence and large language models. This is very different from some of our competitors.
Our chip reserves, as well as the chips available from the market, should be enough to support millions of artificial intelligence applications in the future.
In the long run, I think China will form its own ecosystem. The chip performance may not be that powerful, but it will have the most efficient local software stack.
There is ample room for innovation at the application layer, model layer and framework layer. We have a self-developed four-layer artificial intelligence infrastructure and a strong R&D team. Our focus on artificial intelligence and application-driven methods are building The ecosystem will be based on the Wenxinyiyan system. In the long run, we are confident that Baidu can become the leader of China\’s artificial intelligence ecosystem.
Goldman Sachs analyst Lincoln Kong: My question is about the advertising business.
If compared with our peers, what is the most important obstacle to the growth of Baidu\’s advertising business? What trends does management see in terms of advertising budgets and advertisers\’ willingness to invest, especially heading into the second quarter? How should we view the normal growth of advertising business throughout this year? Robin Li: Our online marketing revenue in the first quarter increased by 3% year-on-year. Traditional search is maturing.
We are working hard to use artificial intelligence technology to achieve user experience innovation. Currently, about 11% of search results are calculated using generative artificial intelligence technology.
These results provide more accurate, organized, and direct answers to questions asked by users. In some cases, users can do things they could not do before.
We have not yet begun to commercially monetize the results of this part of generative artificial intelligence. Therefore, the corresponding growth in related revenue will have to wait for some time.
The weakness of the macro economy has also led to the weakness of our advertising business. Baidu\’s advertisers come from all walks of life. Most of them are small and medium-sized enterprises. Therefore, our advertising revenue is highly sensitive to the macro environment. It is especially sensitive to the offline economy.
In the first quarter, advertiser spending in certain vertical areas such as real estate and franchising remained weak.
Specifically, in the real estate industry, not only the expansion of developers and agents is relatively moderate, but this impact also extends to upstream and downstream industries, such as upstream energy, chemicals, machinery, and building materials, as well as downstream home decoration and furniture industries. .
Resulting in these industries being restricted in their advertising spend on our platform.
In addition, many small and medium-sized enterprises in offline industries have faced many difficulties in the past few years and need more time to recover.
Entering the second quarter, advertiser sentiment has not improved.
Due to the limited visibility of the market outlook, it is unlikely that advertiser investment sentiment will improve. Coupled with the high year-on-year base in the second quarter, Baidu\’s online marketing business revenue should remain basically stable. But from a growth perspective, in the next few years It is likely to remain weak in the next quarter.
If short-term challenges are not considered, we expect that online marketing will still be Baidu\’s main business in the future.
Search is one of the most popular applications in the Internet era. Baidu is still China\’s largest search engine, with nearly 700 million monthly active users. Search may also become a killer application in the artificial intelligence era.
Technological innovation will allow us to better interact with users, developers and merchants. Directly link user intent with the most relevant products and services in a more natural way.
(Continuously updated.
.
.
)

Like (0)
Previous May 16, 2024 11:24 pm
Next May 16, 2024 11:24 pm

Related posts