Baidu (Nasdaq: BIDU; HKEX: 9888) today released its financial report for the third quarter of 2024 ending September 30: total revenue was 33.6 billion yuan, down 3% year on year. Net profit attributable to Baidu was 7.6 billion yuan. Net profit attributable to Baidu without US GAAP was 5.9 billion yuan.

After the financial report was released, Baidu Chairman and CEO Li Yanhong, Mobile Ecosystem Business Group President Luo Rong, Smart Cloud Business Group President Shen Zhi, and Acting CFO He Junjie attended the subsequent financial conference call to interpret the highlights of the financial report and answer questions from analysts.

The following is the main content of the analysis question and answer session:

Alicia Yap, Citibank Analyst: My questions relate to artificial intelligence and the search business. How does management view the accelerated increase in the penetration rate of generative artificial intelligence search results? What is the optimal level of permeability? How long will the transition plan for search products last? Can management also elaborate on recent developments in artificial intelligence search in terms of commercial monetization? When does management expect to start commercializing AI search results?

Luo Rong: We have been using artificial intelligence technology to continue to transform Baidu's core search business. As Li Yanhong just mentioned, now more than 20% of pages and more than 50% of monthly active users have participated in or combined with artificial intelligence content. However, I think this number only represents the initial progress we have made in one aspect of AI-driven search transformation. It should not be viewed as a comprehensive indicator of our AI-generated content, nor should it be used as the sole benchmark for measuring our progress; what really matters is far more than that.

In fact, we aim to transform search and transform every aspect of the search experience through simple words. No matter when users need it, we can provide them with unrestricted and customized personalized content by combining multiple formats, use Wenxin's words to deeply understand the user intentions behind increasingly complex queries, generate customized content for users to better meet their needs, and continuously improve the quality of content generated by artificial intelligence. We're also working to diversify content formats, such as artificial intelligence summaries, images, videos, smart assistants, posts, and even digital people that you can see on the homepage.

All of these different formats can be dynamically combined to create personalized social experiences through simple words. We can customize generic content and presentation formats according to what users want to see and how they prefer to consume information, which allows us to optimize the user experience and increase user engagement.

The AI-driven search transformation has already achieved some initial progress in certain user metrics, which gives us confidence to continue to advance related work. Through continuous exploration, we're opening up new fields of possibility, unlocking the unprecedented potential of artificial intelligence, and every quarter, we'll continue to see new results in every aspect of AI-driven search transformation.

Specifically, when it comes to your question about monetizing your business, we've seen encouraging early results in some vertical sectors, including law, education, and B2B services.

For example, our Wenxin Smart Assistant for advertisers has now achieved the growth of effective sales leads, which not only enhances value for advertisers, but also enhances the overall user experience. As our early foundation models continue to evolve, we expect these intelligent assistants to unlock greater revenue opportunities in the future.

Even so, we're still in the early stages of exploring commercial monetization opportunities for AI search, and we're taking a prudent approach at our own pace. In the short term, we will prioritize improving the user experience rather than rushing to monetize the business. In fact, our product transformation journey strongly reflects our strategic direction, which is our deliberate choice to prioritize long-term value creation over short-term benefits.

Admittedly, we're facing some short-term pressures, but as we work towards a long-term vision beyond search, these trade-offs are something we're willing to accept. In addition, we have built a comprehensive mobile cooperation system, covering multiple consumer-oriented products, such as information flow, encyclopedias, and health products, to serve a huge user base. We are also using Wenxin's words to transform the entire mobile system product line, improve content production, optimize distribution mechanisms, and enhance monetization capabilities. We are still fully committed to these transformational efforts, focusing mainly on improving user experience and user engagement. These encouraging early results have strengthened our confidence in our strategic direction, and we believe that after this period of adjustment, we expect our advertising business to begin to improve next year.

Alex Yao, Analyst at J.P. Morgan Chase: My question is about Wenxin's application programming interface (API). The number of API calls has grown very rapidly in recent months. Could management please share the key drivers behind it? Furthermore, are there potentially explosive applications in the B2B or B2C market? Finally, what is the management's outlook on the medium- to long-term adoption of the word internal culture and mentality?

Li Yanhong: Yes, we have also seen a significant increase in API calls. I think there are a few factors driving this growth. First, Wenxin is getting smarter every day, and the limitations have been drastically reduced, so more and more apps can use it to add value to users. Second, inference costs are getting lower and more customers can afford them. Third, we provide our customers with a toolchain that allows them to easily customize models according to their specific needs in each application scenario. The large volume of Wenxin's API calls is driven by both internal and external demand.

We began our product transformation in the second quarter of last year. We first upgraded our commercial monetization system, achieved a higher return on investment, and generated hundreds of millions of dollars in incremental revenue every quarter. Since then, we have extended the transformation of artificial intelligence to all major consumer-oriented products, including products with monthly active users exceeding 0.1 billion, such as Baidu Search, Baidu Library, Baidu Netdisk, Baidu Input Method, and Baidu Maps. We are encouraged to see that artificial intelligence can completely transform and reshape these products, which have a huge user base and have been around for a long time, and this year we have accelerated the innovation of our search business, showing more and more AI-generated content on search results pages.

We believe the new Baidu search has potential and is ready to become a popular app in the era of generative artificial intelligence. This is because search is inherently deeply rooted in language and text understanding, which fits perfectly with the capabilities of the Big Language Model (LLM). Along this line of thought, millions of smart assistants are being built to answer questions from users and customers, which also contributed to the growth of early API calls.

Currently, most API calls come from innovations in our consumer-facing (B2C) products, and artificial intelligence has been highly recognized for helping enterprises effectively meet challenges. The number of external API calls grew quite fast. In the last week of the third quarter, compared with the previous quarter, it increased by about 240% month-on-month. We've seen significant adoption in various industries such as online education, social media, food service, healthcare, legal advice, and recruitment. This widespread adoption shows that companies recognize the value of our strong model and are willing to invest in it.

We are satisfied with the progress made and are committed to transforming from an Internet-centered business to an AI-first business and accelerating the application of artificial intelligence in all product lines. As generative artificial intelligence becomes the core of our entire product line, we are confident to unlock new revenue channels and solidify our market leadership position.

Morgan Stanley Analyst Gary Yu: I have a macro-related question. Can management discuss changes in advertising demand trends, and have you seen any improvement in ad spending intentions after the latest stimulus policies were introduced? Also, what is the management's view on the macro situation in 2025?

Luo Rong: I think our advertising business is highly related to the macro environment, especially to offline SMEs in many industries. They occupy most of our advertising space. Since these SMEs are closely linked to domestic consumption and are highly sensitive to changes in the macro situation, their vitality and recovery speed are more relevant indicators for our business. In the third quarter, we saw continued weakness in vertical sectors such as real estate, chain operations, and healthcare. As of the fourth quarter, we had not observed a significant improvement in advertisers' spending patterns, and consumer spending remained sluggish.

Having said that, we are very encouraged by the strength and timeliness of recent stimulus policies, and at the same time, we have also seen some constructive initiatives in the market. However, we believe that it will take some time for these measures to spread to offline SMEs and boost their confidence in advertising spending, so we are cautiously optimistic about the future recovery process. We believe that once the macro environment improves and domestic consumption picks up, the confidence of small and medium-sized enterprises will quickly rebound.

Baidu has been the platform of choice for these advertisers for a long time, and has achieved remarkable results in driving customer acquisition and business growth. We expect them to renew advertising spending, which will help drive a marked recovery in our advertising business. With our extensive coverage of offline SMEs and our unique value proposition, we believe our advertising business will rebound strongly once market confidence is restored.

Goldman Sachs Analyst Lincoln Kong: Given the ongoing pressure on the advertising business and our plans to accelerate and expand the transformation of artificial intelligence, can management elaborate on our current business priorities, priorities in terms of investments, such as spending and capital allocation, and how should we view profit margin trends in the fourth quarter?

He Junjie: Yes, we will continue to adhere to a strategy that focuses on artificial intelligence, making it an immediate priority and a long-term strategic priority. Given the ongoing transformation of the search business driven by artificial intelligence and the fact that we will not be commercializing search results generated by artificial intelligence on a large scale in the short term, we expect the online marketing business to remain under pressure in the short term.

However, as we see tremendous long-term value, we will continue to move in this strategic direction. According to this strategy, we are in the stage of continuing to advance various businesses. We will further strengthen product innovation, with a particular focus on artificial intelligence search, and will continue to invest to improve our technical capabilities in artificial intelligence while enhancing our artificial intelligence products to maintain healthy profit margins. In addition, we will also expand our autonomous driving projects, so we can expect to achieve profits.

We believe these initiatives are critical to maintaining our position as a leading technology innovator in China. As a result, profit margins will be adjusted in the near future. Looking ahead to 2025, we will focus on optimally allocating resources to high growth opportunities while being consistent with our long-term strategy.

Mizuho Securities Analyst James Lee: I have a few questions about the cloud business that I'd like to follow up on. As far as revenue growth is concerned, the growth of this business has slowed compared to previous quarters. Can you ask management to discuss the performance of personal cloud and enterprise cloud? Also, can you give me an update on the competitive landscape of the market? Can you look ahead to the cloud business, particularly the contribution of generative artificial intelligence, and how should we view long-term profit margins?

Shen Shao: Since last year, driven by the development of generative artificial intelligence and basic models, we have observed that China's cloud industry is accelerating its transformation to artificial intelligence computing, and more and more customers are choosing Baidu's artificial intelligence cloud infrastructure and basic models. Revenue related to generative artificial intelligence reflects its growing appeal. When we first reported in the fourth quarter of last year, it accounted for about 5% of our total AI cloud revenue, but by this quarter it had more than doubled to 11%. So we are confident that this upward trend will continue.

As you know, our AI cloud business consists of two parts: personal clouds and cloud services for enterprises and the public sector. Total AI cloud revenue growth slowed slightly this quarter, mainly due to short-term business adjustments that had a temporary impact on personal cloud revenue. However, we believe the ongoing personal cloud innovation will help mitigate this short-term impact and prepare it for longer-term growth opportunities.

At the same time, cloud services for enterprises and the public sector, which account for the majority of our AI cloud revenue, maintained a strong development momentum this quarter, and their growth rate continued to exceed the growth rate of the overall AI cloud business. The growth in enterprise cloud revenue is driven by strong demand for model training and inference in various industries such as the Internet, education, and finance. This reflects our customers' high recognition of our artificial intelligence infrastructure and large-scale platform capabilities. In particular, major customers in the Internet, technology, and automotive industries have increased their spending on our GPU (graphics processor) public cloud. Additionally, we are also seeing growth in the number and expenses of midsize enterprises, especially in growing industries in today's cloud market, such as marketing software.

In today's market, basic model capabilities are becoming increasingly critical. Given the high requirements for strong artificial intelligence infrastructure, professional artificial intelligence expertise, and significant capital investment, whether in China or overseas, only a basic model with strong strength can survive, and we will definitely be the best and in a leading position.

We are proud that Wenxin has demonstrated its market leadership position through extensive application and advanced technology. These advantages enable Wenxin to stay ahead, attract more cloud customers, and consolidate our market leadership position. In terms of profitability, our artificial intelligence cloud business is healthier than ever before, and operating margins that are not calculated according to US GAAP (non-GAAP) are increasing year by year. This improvement is driven by the continued increase in profit margins from generative artificial intelligence revenue and our commitment to achieving high-quality revenue growth. Looking ahead, we are confident that Baidu's AI cloud business will maintain a strong revenue growth momentum for a long time and continue to achieve healthy operating profits.

Jefferies Analyst Thomas Chong: My question relates to the Apollo Go (Apollo Go) autonomous travel service. Given the recent listing of some autonomous driving-related companies with partnerships with major automakers and related industry trends, how does management think about the future competitive landscape? Looking forward to the future, how do you view Radish Express's free travel service and business strategy development? Can you share detailed plans for its future expansion and development?

Li Yanhong: Starting an autonomous driving business has a high threshold. It requires top professional technology, excellent talents, and firm long-term investment and strong capital reserves. We have built an unrivaled foundation in all of these areas, which has given us a strong competitive advantage and set a high standard for other players. As one of the earliest players in the market, we have continued to invest for more than 10 years and have always unswervingly demonstrated our determination. This long-term investment has built our technical advantage and made us a global leader in autonomous driving technology.

As we mentioned earlier, Radish Express has reached a major milestone, that is, it has completed a total of 8 million trips, and we have become the world's top autonomous online car-hailing service provider. In addition to having cutting-edge technology and strong operational capabilities, we have also achieved significant cost efficiency in terms of hardware. The sixth-generation unmanned vehicle Apollo RT6 has a very competitive cost of mass production, with a price of less than $0.03 million, making it the most competitive choice on the market. All of these benefits are driving us forward and paving the way to validate our business model.

The entire market is still in its infancy, so competition can help accelerate market growth and create a regulatory environment more conducive to innovation. We believe that such a regulatory environment is critical to the healthy development of this industry, and we are ready to expand at the right time. At the same time, we are actively seeking new opportunities for international expansion. We see that in cities where fully driverless operations can be carried out on a large scale, such as Wuhan and Chongqing, there is still huge potential for business development. Additionally, as the business grows, we are happy to explore different business model options and focus on asset-light strategies to maintain the flexibility and efficiency of our operations.

UBS Securities Analyst Wei Xiong: I have a question about funding allocation. Can management provide updates on shareholders' capital return plans, such as share repurchases and dividends?

He Junjie: Since becoming a listed company in 2005, as a leading technology company in China, we have been relentlessly creating long-term value for shareholders through our own growth, including implementing stock repurchase plans. Over the past few years, we have maintained a steady pace of share repurchases, with an average annual repurchase amount of approximately $1 billion.

I'm happy to tell you that through our ongoing share repurchase efforts, our total number of outstanding shares has declined this year. Although the pace of share repurchases may vary from quarter to quarter according to our established execution pace, we remain committed to making share repurchase programs a key measure to create value for shareholders. We are currently focused on implementing a sustainable and regular share repurchase plan, and we are open to evaluating various options to give back value to shareholders.

We have always believed Baidu has long-term and sustainable growth potential and will ensure that shareholders are rewarded for their trust in us. The fundamental way to serve shareholder interests is to have a solid business foundation and take full advantage of growth opportunities. We are in the midst of a paradigm shift in generative artificial intelligence and basic model technology transformation. At Baidu, we have artificial intelligence infrastructure, expertise, and financial and human resources to implement our plans, so we hope to invest flexibly as needed to drive new businesses forward, thereby bringing long-term value to shareholders.

Bank of America Merrill Lynch analyst Miranda Zhuang: Can management outline Wenxin's technology roadmap for future development? What milestones are coming to the company's artificial intelligence model? When are these milestones expected to be achieved?

Li Yanhong: We launched Wenxin in March of last year. Since then, we have continued to improve our basic model capabilities, especially our flagship model. In October of last year, we launched Wenxin 4.0, which is China's first GPT-4-like model with world-leading capabilities. Then, in June of this year, we launched Wenxin 4.0 Turbo, taking its performance to the next level.

Based on these milestones, we are committed to continuously improving our flagship models to provide better performance, greater accuracy, and broader support for diverse user needs. We expect to launch a new version of Ernie early next year to further solidify our leadership position in the basic model field.

Compared to overseas companies, we clearly have an application-oriented approach. We believe that basic models have real value only if they support a wide range of applications that meet user needs. Over the past 18 months, our model development has revolved around solving real problems according to market needs. Through this approach, we have made significant progress in solving the “illusion” of the model — generating content that does not match the facts — and improving accuracy. We've also launched a range of customized models to make the Wenxin model easier to use and less expensive. We've improved performance, reduced inference costs, and increased response speed. Our strength is that we know our strengths and the way forward, and prioritize the technologies that are most closely integrated with our business and create the most value.

I mentioned earlier this year that we have expanded our vision-based model capabilities to the field of autonomous driving in order to further solidify our leading position in this field. Now, we are also actively exploring multimodal capabilities and applications based on the Wenxin Language Model to create more synergetic value and open up new possibilities in various fields. We are very careful in allocating resources and optimizing our underlying model in the direction of maximizing the impact on our business, which helps us maintain our market leadership position. (end)