百度 (BIDU) 2022 Q3 法說會逐字稿

Thank you for standing by, and welcome to the Baidu Inc. Third Quarter 2022 Earnings Conference Call.

(Operator Instructions) I would now like to hand the conference over to your host for today, Juan Lin, Baidu's Director of Investor Relations. Please go ahead.

Hello, everyone, and welcome to Baidu's Third Quarter 2022 Earnings Conference Call. Baidu's earnings release was distributed earlier today, and you can find a copy on our website as well as on Newswire services. On the call today, we have Robin Li, our Co-Founder and CEO; Rong Luo, our CFO; Dou Shen, our EVP in charge of Baidu AI Cloud Group; and Xin-Zhe Li, our SVP in charge of Baidu Intelligent Driving.

After our prepared remarks, we will hold a Q&A session.

Please note that the discussion today will contain forward-looking statements made under the safe harbor provisions of the U.S. Private Securities Litigation Reform Act of 1995. Forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from our current expectations. For detailed discussions of these risks and uncertainties, please refer to our latest annual report and other documents filed with the SEC and Hong Kong Stock Exchange. Baidu does not undertake any obligation to update any forward-looking statements, except as required under applicable law.

Our earnings press release and this call include discussions of certain unaudited non-GAAP financial measures. Our press release contains a reconciliation of the unaudited non-GAAP measures to the unaudited most directly comparable GAAP measures and is available on our IR website at ir.baidu.com. As a reminder, this conference is being recorded. In addition, a webcast of this conference call will be available on Baidu's IR website.

I will now turn the call over to our CEO, Robin.

Hello, everyone. Recapping the third quarter in broad terms, we delivered improved bottom line results despite a challenging macro environment, especially for the mobile ecosystem, operating profit resumed a positive year-over-year growth. And for AI Cloud, operating loss and margin improved meaningfully, both on a year-over-year and a quarter-over-quarter basis.

Businesses across the line from mobile ecosystem to AI cloud to intelligent driving have been negatively affected by the resurgence of COVID. Throughout Baidu's history, however, we have experienced many challenging environments. Periods of challenges have enabled us to emerge stronger given our relentless effort on building long-term growth. We're using this period to ready ourselves for current business conditions to improve.

Our key short-term tasks remain unchanged, which are efficiency optimization and continuous investments in the new AI businesses. Our new AI business such as AI Cloud and Intelligent Driving are well aligned with China's tech innovation and national initiatives. By doing so, we will further strengthen our leadership in the new AI business and reaccelerate business growth.

In Q3, Baidu Core ad revenue was down 4% year-over-year but improved from the second quarter's 10% year-over-year decline, as macro has improved gradually since June. Encouragingly, revenues from health care and retail recorded positive year-over-year growth in the quarter. Going forward, once COVID situation alleviates in major cities and revenues across different verticals such as travel, franchising and local services should rebound.

For our new AI business, I am proud to report some highlights. Revenues from AI Cloud increased by 24% year-over-year to RMB 4.5 billion in the quarter. AI Cloud has been a major growth driver for Baidu Core non-advertising revenue. In Q3, non-advertising accounted for 26% of Baidu Core total revenue, that's up from 21% a year ago.

Baidu Apollo auto solution continued to gain traction among leading automakers and the total projected accumulative sales reached RMB 11.4 billion recently, growing more than 50% year-over-year. Apollo Go completed 474,000 rides in the quarter, up 311% year-over-year. By the end of Q3, Apollo Go has completed 1.4 million rides on public roads. We believe we remain the largest autonomous ride-hailing service provider globally.

Now let's review the third quarter operational highlights. Revenues from AI Cloud increased by 24% year-over-year to RMB 4.5 billion in the quarter. In Q3, the operating loss margin for AI Cloud improved notably both year-over-year and quarter-over-quarter. Over the past quarters, we have made shifts away from some lower-margin businesses to build sustainable growth for AI Cloud. Meanwhile, we continue to take measures to standardize our solution to cut down deployment costs as we scale up. Our efforts have allowed us to gradually improve operating profit and margin.

I'll take ACE smart transportation as an example. In the past quarter, we continued to grow revenue and improve operating margin for smart transportation as we gain scale. Leveraging our strong AI capabilities and our insights into the industry, we have divested the solutions for typical use cases such as traffic management and V2X for urban road and highway.

Previously, we talked about how our projects helped improve traffic efficiency on city roads in Beijing, Guangzhou, Changsha, Chongqing and other cities. Recently, [Zhuzhou] city an important transportation hub in Hunan province, adopted our smart traffic management solutions. Our solutions covered more than 70% of the major intersections in the business district of Zhuzhou . After adopting our solutions, delays have been reduced by 22% during rush hour in the current region.

By the end of the quarter, Baidu ACE smart transportation solutions has been adopted by 63 cities, up from 24 cities a year ago and 51 cities a quarter ago based on contract amount of over RMB 10 billion -- RMB 10 million per city. As we demonstrated our capability to use AI to improve transportation efficiency in more cities, ACE smart transportation will further expand its market share.

We're trying to repeat our success in smart transportation in other key verticals, such as manufacturing, energy and utilities and the public sector. We will empower our customers with AI capabilities to help them increase productivity and cost control. With AI Cloud, our customers will be better positioned to take advantage of digital and Intelligent transformation.

We believe we remain well positioned in the early-stage, fast-growing market due to: Number one, our capabilities to establish end-to-end solutions based on our full stack AI capabilities ranging from chip design to deep learning frameworks to large language models to application-level software; and number two, our insight about customers' pain points and our growing know-how and capabilities to solve their problems.

While COVID caused delays in project implementation and complicated our sales team's efforts, hampering new contract wins, we believe the long-term trend of digital and intelligent transformation remains unchanged. Looking ahead, we will continue to focus on quality growth and aim to improve margins to achieve profitability for AI Cloud.

In Intelligent Driving, we marked several key highlights in the third quarter and our years of investments in intelligent driving have begun bearing fruit. Baidu Apollo auto solutions, our total projected accumulative sales exceeded CNY 11.4 billion recently. Based on our current pipeline, some of the major car models equipped with ANP and AVP are expected to be launched in the second half of next year. We expect meaningful revenue contribution from this business to begin in 2024, and profit to expand once material revenue kicks in. Such growth demonstrates the increasing demand for our auto solution.

In Q3, we extended our partnership with one of China's largest automotive and technology companies ANP, AVP and HD Maps for one of their popular models. As of today, we have announced a collaboration with many OEM partners. Serving various automakers has helped us standardize our auto solutions, making them compatible with more popular car model. As more cars equipped with our auto solutions get into consumer plans, we will continue to refine and update our solutions.

Considering the sizable development cost for high-level autonomous driving technology, many automakers seek partnerships with reliable suppliers who have strong brand and technology capabilities. We have differentiated ourselves with world-leading Level 4 autonomous driving, which has been established through years of investments. Now we are proactively building high-quality partnerships with an increased number of auto OEMs to accelerate our partners' progress in autonomous driving.

A few months ago, we integrated Baidu Map into our intelligent driving group to create synergies between the Baidu Map, also -- Baidu Map mobile app and the map solutions for auto and transportation industries. Since our map have already been widely adopted, we have obtained more insights in the transportation industry, allowing us to strengthen our solutions for the mobility sector.

On Apollo Go, we continue to scale up our operations, and we believe we remain the largest autonomous ride-hailing service provider worldwide. In the third quarter, the rides provided by Apollo Go reached 474,000, increasing 311% year-over-year and 65% quarter-over-quarter. A vast majority of the rides were for serving passengers during rush hours, covering places like subway stations, office buildings and shopping malls.

As our operation continues to expand, Apollo Go keeps learning and improving from scenarios that do not occur during the testing phase. Leveraging large-scale operations, we're making Apollo Go a professional AI driver while offering a safe and comfortable autonomous ride-hailing service.

Today, with an Apollo Go vehicle enters dense intersection, it is able to easily take a gap between pedestrians, bicycles and other moving objects, safety driving school while providing a comfortable experience to the passengers. Apollo Go continues to improve autonomous ride-hailing services under extreme weather.

Scalable operations have indeed reinforced our Level 4 autonomous driving capabilities, setting a strong foundation for further operation expansion. We believe this virtuous cycle will make Apollo Go the most experienced AI driver to handle various situations on the road on a large scale.

We're facing the expansion of Apollo Go based on a comprehensive financial model. In previous earnings calls, we talked about Apollo Go's achievement in the Yizhuang region of Beijing. In fact, Apollo Go has achieved significant progress in other Tier 1 cities as well.

In Q3, on average, each vehicle in Beijing, Shanghai and Guangzhou completed more than 15 rides per day. According to our knowledge, this number is quite close to the average daily rides for traditional ride-hailing services.

We believe that our strong and improving safety track record on public roads provide a strong endorsement for more cities to issue permits for fully driverless ride hailing. We believe fully driverless ride hailing will create more affordable urban transport and attract more consumers to the ride-hailing market. Baidu will continue to invest to capture this massive market opportunity.

Moving to the mobile ecosystem. Our operating profit improved on a year-over-year basis. We continue to generate strong cash flow in the quarter. Our mobile ecosystem continued to expand. In September, Baidu App's MAUs increased by 5% year-over-year to 634 million. In Q3, total mobile search queries increased by double digits year-over-year, and feed distributed through Baidu App increased by 23% year-over-year.

We continue to introduce short videos in feed and search results. In September, short video distribution and time spend within Baidu Feed grew double digits year-over-year. For Baidu Search in September, 23% of the clicks on search result pages were short videos. And we expect the prevalence of video in search results to continue progressing rapidly.

We're using AI to produce more short videos for our mobile ecosystem. While it is still in a very early stage, we believe AI will allow us to generate short videos much faster and more cost effectively than human-generated content.

In e-commerce, users are coming to Baidu App for product search, which resulted in a continuous increase in product-related search queries. Quarterly GMV facilitated by Baidu Search grow by about 52% year-over-year in Q3. Today, users come to Baidu App not only to search for information and knowledge, but increasingly to look for services and merchandise. As we are able to make purchases and book services without leaving Baidu App, we have been deepening our understanding of user needs.

Meanwhile, more customers are leveraging our ecosystem infrastructure to build their marketing campaigns and even operate their businesses on our mobile ecosystem. This has allowed us to accumulate more customer insights and better serve the customers.

The improving insights into our users and customers have, in turn, allowed us to improve user experience and app conversion. In the future, we believe our mobile ecosystem will continue to generate strong profit and cash flow.

To close, during this uncertain time, we focused on making the long-term investments that will position us to be stronger coming out of this challenging period, including Baidu AI Cloud and Intelligent Driving. While the advertising industry has been impacted by COVID-19, some of our verticals could recover faster when there is an upturn in the economy, which will drive our ad revenue to resume growth.

Before I turn the call to Rong, I want to thank Baidu's employees for their diligent work in a challenging environment and the time and effort they put into making our company story of success in the long term.

With that, let me turn the call over to Rong to go through the financial highlights.

Thank you, Robin. Now let me walk through the details of our third quarter financial results. Total revenue was RMB 32.5 billion, increasing 2% year-over-year. Revenue from Baidu Core was RMB 25.2 billion, increasing 2% year-over-year. Baidu Core online marketing revenue was RMB 18.7 billion, decreasing 4% year-over-year, but improved 10% from the second quarter as macro has improved gradually since June. Baidu Core, non-online marketing revenue was RMB 6.5 billion, up 25% year-over-year, driven by cloud and other AI-powered businesses. In Q3, AI Cloud increased by 24% year-over-year to RMB 4.5 billion. Revenue from iQIYI was RMB 7.5 billion, decreasing 2% year-over-year.

Cost of the revenues was RMB 16.3 billion, increasing 1% year-over-year. Baidu Core, cost of revenue was RMB 10.7 billion, increasing 15%, 1-5 year-over-year, which was in line with the growth in sales of AI Cloud and other AI-powered businesses.

Operating expenses was RMB 11 billion, decreasing 19%, 1-9, year-over-year, primarily due to a decrease in channel spending, promotional marketing expenses and staff-related expenses.

Baidu Core selling, general and administrative expenses were RMB 4.2 billion, decreasing 31%, 3-1, year-over-year. SG&A accounting for 17%, 1-7, of Baidu Core revenue in the quarter and decreased from 25% in the same period last year. Baidu Core research and development expenses was RMB 5.3 billion, decreasing 4% year-over-year. R&D, accounting for 21% of Baidu Core revenues in the quarter and decreased from 22% in the same period last year.

Operating income was RMB 5.3 billion. Baidu Core operating income was RMB 5 billion, and Baidu Core operating margin was 20%. Non-GAAP operating income was RMB 7.2 billion. Non-GAAP Baidu Core operating income was RMB 6.7 billion. And non-GAAP Baidu Core operating margin was 26%.

Q3 present the first year-over-year growth in non-GAAP operating profit and margin perspective since Q2 2021.

Total other loss, net was RMB 4.8 billion, decreasing 78% year-over-year, which mainly including a fair value loss of RMB 3.1 billion and impairment of long-term investments of RMB 1.4 billion. In the third quarter last year, we recognized a fair value loss of RMB 18.9 billion. A significant portion of long-term investments, including but not limited to investments in equity securities of public and private companies, private equity funds and digital assets. This is subject to quarterly fair value adjustments, which may contribute to net income volatility in future periods.

Income tax expenses was RMB 908 million compared to an income tax benefit of RMB 1.8 billion in Q3 2021, primarily due to an increase in deferred tax benefit recognized on fair value loss of long-term investments and deduction on certain expenses that were previously considered non-deductible in the third quarter of 2021.

Net loss attributable to Baidu was RMB 146 million, and diluted loss per ADS was RMB 0.87. Net income attributable to Baidu Core was RMB 25 million. Non-GAAP net income attributable to Baidu was RMB 5.9 billion. Non-GAAP diluted earnings per ADS was RMB 16.87. Non-GAAP net income attributable to Baidu Core was RMB 5.8 billion, and non-GAAP net margin for Baidu Core was 23%.

Adjusted EBITDA was RMB 8.9 billion, and adjusted EBITDA margin was 27%. Adjusted EBITDA for Baidu Core was RMB 8.2 billion, and adjusted EBITDA margin for Baidu Core was 33%.

As of September 30, 2022, cash, cash equivalents, restricted cash and short-term investments were RMB 184.5 billion and cash, cash equivalents, restricted cash and short-term investments, excluding iQIYI were RMB 179.5 billion. Free cash flow was RMB 6.6 billion, and free cash flow, excluding iQIYI, was RMB 6.4 billion.

Baidu Core had approximately 36,500 employees as of September 30, 2022. On a separate note, iQIYI generate positive operating profits on both GAAP and non-GAAP basis in this quarter. In addition iQIYI also generated positive free cash flow in this quarter.

With that, operator, let's now open the call to questions.

(Operator Instructions) The first question today comes from Alicia Yap with Citigroup.

I have 2 questions. First, can management comment how should we look at the advertising demand recovery in the near term and also in the next 2 quarters?

And then secondly, can management comment on how you think about the competitive landscape in the China advertising market will evolve going forward? And what would Baidu long-term market share be?

Alicia, this is Robin. Let me answer your questions. Yes, our revenues are very sensitive to COVID-control measures. Baidu Core's ad revenues returned to positive year-over-year growth in August. But in September, it decreased again because of a new round of COVID resurgence. Overall, Q3 was much better than Q2 because there was a recovery from the second quarter, in which there were a lot of disruptions caused by COVID.

When we entered the fourth quarter, the situation improved in October but since early November, it got a bit cloudy because another round of COVID impacted some regions, like Guangzhou and Beijing.

We're closely watching how the situation will develop. I think the short term will probably still be quite volatile, but the economy should improve in the midterm and beyond. China has been fighting against this COVID for almost 3 years and the country has been gaining experience.

Well, we're certainly going to see some ongoing disruptions and uncertainties. The overall situation should improve -- should move in a positive direction over the next few quarters. As you know, many of our ad verticals were affected by COVID and macro. So once COVID and macro situations improve, our ad revenues from different verticals such as travel, franchising or local services should rebound.

Again, we will closely watch how COVID situation develops, and we will work very hard to bring our ad revenue back to a positive year-over-year growth as soon as possible.

In addition, AI Cloud and Intelligent Driving, our new businesses, also negatively impacted by the COVID disruption. So if COVID impact subsides, this non-advertising businesses should also see improvement.

And regarding to your question on competition in the online advertising market, first of all, I think Baidu App is one of the very few super apps in China's mobile Internet industry, well recognized and well established. Search ads enjoying the best ROI among various types of performance ad because users explicitly express their intent in our search box. And search ads connect users' intentions with the most relevant product and service offerings. This is why a lot of advertisers consider Baidu Search as the most important and most efficient channel to reach their targeted audience.

And secondly, China's mobile Internet is maturing. User growth is gradually slowing down and the competitive landscape is much less volatile than before.

With this backdrop, Baidu App still managed to grow its user base nicely over the past many quarters. On top of that, with the unique positioning of Search, we believe we should be able to sustain our market share in the online advertising market over the long run.

And thirdly, unlike our peers, most of our advertisers are SMEs, right, being in the real economy. Many of them run businesses in the service sector such as local services, business services and travel. This businesses work hurt the most by the pandemic. So when COVID impact subside, ad revenue should rebound quickly and Baidu stand to gain share in the overall advertising industry.

The next question comes from Eddie Leung with Bank of America.

Pretty solid quarter despite all the macro headwinds. I have a couple of questions on cloud services. So we have seen -- it seems like a slowdown in the industry. So could you talk a little bit about the reasons behind the slowdown? And how you guys are thinking about the growth rates in the upcoming, let's say, your quarters or in the medium term?

And then secondly, could you also talk a little bit about the competitive landscape. We have seen a media reporting market share gains by cloud companies outside the Internet sector in the past 1 or 2 years, right? So any color on the competitive landscape, will be useful?

And then finally, I remember, Robin, you mentioned about the margin improvement of cloud. So could you also talk about your thoughts on the role to profitability? Any time line would be helpful.

Thanks a lot for your questions. This is Dou. I'll try to answer your questions. So for the revenue growth part, right, I think it's slowing down mainly due to the COVID-19 impact, such -- for example, because of the travel restrictions, we could not even implement our projects on time, and the bidding for the new contracts was also affected.

Apart from that, I also want to highlight that we are focused on healthier and more sustainable growth by cutting some low-margin businesses. So we believe this approach is crucial for the long-term development.

Actually, looking beyond the current quarter, as Robin has just mentioned, the trend for China's traditional industries and the public sector to use AI and move their business onto the cloud remains unchanged. So I think post pandemic, the companies should gain more confidence in their future growth so that they are more willing to spend more money on digital and intelligent upgrades.

Baidu has already demonstrated that we are very capable of using AI to improve the efficiency in transportation. As you can already see, our ACE smart transportation revenues have been growing rapidly, and we continue to gain market share accordingly.

So in the meanwhile, we are reducing the operating loss for small transportation because we continue to expand scale and continue to increase operating leverage. So in our next steps, so we are excited about repeating our success in transportation in other traditional industries like manufacturing and utilities.

So actually, in the -- just the past quarter, so our cloud revenues from the manufacturing and utilities industries both grow solidly. And it's largely because of our continuous efforts in improving the end-to-end AI capabilities and our understanding of the fundamental needs in these industries.

So to answer your questions, we believe this -- our sustainable competitive advantages to compete against other players. In addition, the market is expanding and there are plenty of opportunities for us and our peer companies to grow over the long term.

So talking about the profitability. So Baidu AI Cloud reduced operating loss and continue to improve operating loss margin this quarter. To expand on that, our personal cloud kept generating decent operating profit and margin. And our enterprise part grow faster than personal cloud, and we're very happy to see the trend of loss reduction because of the efforts I just mentioned earlier. So our strategy to standardize solutions in the user -- in the use cases for key industries started to bear fruit. So we have worked very hard to replicate more solutions from one user case to another to grow business scale and improve margins.

So it's already proven in smart transportation as we have mentioned. And in the future, we look forward to expanding this approach at a large scale in more key industries. So AI Cloud is very important new business for Baidu, and we will continue to grow the business and improve margins going forward.

The next question comes from Alex Yao with JPMorgan.

I have a couple of questions on Baidu Core margins. In the past several quarters, you guys did a very good job in terms of cost control and the margin improvement. As large-scale BU and headcount streamlining has been already down, how much more room should we think about for further cost cutting in Baidu Core in the coming quarters and 2023? And the related question is how should we think about the Baidu Core operating margin going forward?

Lastly, would Baidu slow down the cloud and autonomous driving investments if the macro situation further deteriorates?

Thank you, Alex, for your questions. Let me try to figure your questions. This is Julius. I think Robin has said at the very beginning of the prepared remarks, I think for Baidu Core, we continue to optimize the cost expenses and we continue to improve our operational efficiency, which is one of our key short-term tasks. At the same time, we kept investing in the new AI businesses for our future growth.

This strategy actually has not changed in the past few quarters. And this strategy allows us to weather through the challenging market environment, and we believe that it will prepare us for accelerated growth again once the macro downgrade is over.

In Q3, let me recast some of the numbers. Baidu costs non-GAAP operating profit margin expanding to 26% from 24% in the same period last year. I think it has represented the first year-over-year growth in non-GAAP operating profit and margin perspective since Q2 2021. And we made a lot of efforts to achieve such results. You'll probably see that from our earning release. Our SG&A has decreased once again in Q3, this quarter from the same quarter last year. We have been very disciplined with the channel spending and promotional activities. And just to add that, it was already less consecutive quarter that SG&A achieved a year-over-year decline. And going forward, we will continue to control the variable costs and expenses as same as what we did in the past.

If we look into our each part of our business, for mobile ecosystem, our operating profit actually is going up on a year-over-year basis, even though we know the revenue was negatively impacted by COVID-19. And we have also already mentioned earlier and which is just repeated by Dou, we made a lot of efforts for the healthy growth of AI Cloud. This quarter, the operating loss margins for AI Cloud improved significantly, both on a year-over-year basis and quarter-over-quarter basis.

If we separate two-part, for enterprise and public status, we're allocating our resources to high-margin business and reduce the low-margin business. Also, we continue to standardize our end-to-end solutions, including the IaaS, PaaS, SaaS for key user cases (inaudible) and as much expectations started one of the examples, we continue to grow fast and at the same time, improve the margins quite notably. And in the personal cloud, same as well in the past, we continue to generate a decent profit this quarter.

For our robotaxi business, Apollo Go, we also chose to grow Apollo Go at a well-measured pace. We have very comprehensive financial models to optimize and forecast the cost and expenses in this business, especially the labor costs and the vehicle hardware costs. We have been very careful to measure and estimate the cash flow for this business.

Our strategy is to strengthen our leading position in robotaxi all over the world, acquire sizable market shares in the ride-hailing sector, especially in the key cities in China and in the future, and ultimately try to generate more profits in these key areas. We are not expanding our operations without thinking about building a sustainable and profitable business models for Apollo Go.

For auto solutions, since we mentioned just now, we already have a projected cumulative sales of around RMB 11.4 billion, while revenue contribution is still very small at this stage, and we expect it to see more revenue kicking in from the second half of next year or maybe early in 2024, as small cars with our solutions can become available in the market. And once the meaningful revenue kicks in, this business should start to generate profit.

I would like to also make it clear the RMB 11.4 billion is our estimate including the contract side and nomination letters we received from OEMs. In auto industry, our nomination letter means a supplier is affected for certain projects, and then the OEMs will sign up contracts with us first. Our estimate is based on our assumptions of the timing of launch, the pricing, the future volumes. I hope this gives you a better understanding of how we have made these calculations.

Looking forward, we will continue to be very disciplined with the cost and expenses. And at the same time, we'll continue to invest in the AI Cloud and Intelligent Driving for our long-term growth despite of the challenging environment in the short term. And in the future, we believe our mobile ecosystem will continue to generate decent profits and cash flow to support our investments in the new AI business. And we'll also continue to work very hard to narrow our loss in AI Cloud.

The next question comes from Gary Yu with Morgan Stanley.

Congrats on the expanded partner network and growing backlogs for ADS business. I have a question on your auto solutions. Just wondering, when should we expect meaningful revenue to start kicking in? And how do you differentiate the Apollo's auto solution from your peers? And could you help us understand the underlying market space for Baidu Apollo's auto solutions? And a related question to that is, have you noticed a significant change in the attitude of the OEMs towards self-developing for intelligent driving solutions?

Gary, this is Robin. We think huge opportunities in the auto solutions market, in the first 10 months of this year, EV sales increased by more than 100% in China. We're now the largest EV market in the world, accounting for more than half of the global EV sales.

A clear trend for the auto industry is vehicle intelligence. Baidu is benefiting from this trend. Our years of investments in autonomous driving have begun to bear fruit. Baidu Apollo's auto solutions derived from our L4 technology. A lot of AI models we build for robotaxi can be used for ASD.

Another point to note is that as the market leader, we have been investing in autonomous driving for about 10 years. And thanks to these investments, we not only have completed tens of millions of testing miles on the public roads, but also accumulated a very valuable experience by running the largest robotaxi fleet on urban roads on a daily basis. This has provided us great insight, and we have used these insights to develop auto solutions that best meet our customers' needs.

So for example, ANP 3.0. We believe it is the most advanced intelligent driving solution for city roads on the market. Because ANP 3.0 is derived from our core driving technology, we were able to use a very limited amount of R&D personnel to make this advanced solution available to the market within a short time. Jidu's first car, ROBO-01 will be the first one to use ANP 3.0. We believe that with more and more automakers adopting our auto solutions, we will become more experienced in making our solutions compatible with more popular car models.

On the other hand, when more and more cars with our auto solutions are on the road, we will be able to grow experience from everyday users and we'll continue to improve our auto solutions.

As I mentioned earlier, our total projected cumulative sales for Apollo's auto solutions reached CNY 11.4 billion, which is more than 50% from last year.

Based on our current pipeline, some of the major car models that are equipped with ANP, AVP should be launched starting in the second half of next year. So meaningful revenue contributions for this business should begin in 2024, and profit will follow.

On the differentiation, if you look at the cost side, as we mentioned many times before, developing advanced intelligent solutions require massive investments in technology and talent. Frankly speaking, most companies cannot afford this kind of investment.

Automakers trust our brand and technology. So a lot of them have chosen to partner with Baidu Intelligent Solutions. For them, we are a trustworthy and reliable supplier because we have made years of investment in intelligent driving technology, and we have a strong cash position, and our advertising business continue to generate strong cash flow to support our investments.

Going forward, this will continue to support the collaboration with many automakers at large scale. That's why more and more leading automakers have recognized us and have broadened partnership with Baidu. For example, in the quarter, we expanded our partnership with one of China's largest auto companies, and we plan to use our ASD solutions for yet another of their popular car model. And the ASD solution used include ANP, AVP and HD Maps. Going forward, we are looking forward to more partnerships with OEMs, both domestically and internationally.

The next question comes from Kenneth Fong with Credit Suisse.

I have a follow-up question about Apollo Go. On a fully autonomous driving, what kind of progress will be seen in the near term? Could management also talk about a key milestone that Apollo Go will achieve in 2023 and even in the longer term? And to achieve this milestone, how should we think about the impact on your P&L as well as cash flow? And when should we expect Apollo Go business to achieve breakeven?

Thanks for your question. This is [James]. We are making very good progress in the package of providing fully driverless ride-hailing service for the public. We believe Apollo Go is the largest autonomous ride-hailing service provider in the world, because, number one, when we look at the numbers back on [coverage] rate. Apollo Go has provided 1.4 million rides for the public. In Beijing, Shanghai and Guangzhou each half complete 15 rides per day. We are also providing fully driverless ride-hailing services to public in Wuhan and Chongqing. And the number of ride providers by fully driverless [AI] is growing very fast.

And number two, Apollo Go is already in more than 10 cities in China. It is also cities have a population of more than 10 million, as Rong Luo mentioned in his script, building up operation of us is to improve technology and safety. He gave examples about technical improvement. Let me share some examples about how our operation help us to improve user experience.

In operations, we got a lot of feedback from passengers. For example, passengers, one half, a, operated to moderate; b, to fill them up faster and to better plan or route than before. We then refined and improved our auto quality. Our efforts have helped us gain more recognition from passengers. Today, in Chongqing, Beijing, Apollo Go is already importantly for AI computing.

After trying Apollo Go, people come to realize that it is a reliable and fixed service. It drives just like a professional driver. Some people may wonder why is Apollo Go can be better than human drivers? This operation is always be in compliance with traffic regulation and the AI driver never get tired or distracted like human drivers, all of those allow Apollo Go to provide very stable drive to public.

Our strong track record of around [150] has helped us to entrust, both from our user and also from our regulator. We mentioned before, set in Wuhan and Chongqing. We will allow to provide fully driverless ride-hailing service this August. The operation in both cities is well on track.

Going forward, we plan to continue to expand this operation -- and this operation and as more fully driverless vehicles. In 2023 and beyond, we will continue to fulfill our operation. In particular, fully driverless ride-hailing operation in more region and reduce hardware and the vehicle cost. We believe robotaxi will eventually be profitable and cheaper than the current ride-hailing service. We will continue to invest in robotaxi to capture a huge market opportunity.

Kenneth, this is Julius. Regarding your question about the impact on our P&L and cash flow. We believe the overall impact is actually manageable. As Robin said in the script, actually our team today has built a very comprehensive financial world for Apollo Go. This model actually helped us to understand what should we do, where we should improve and adjust to generate profit in these models in the future.

For example, today, we are working very hard towards 2 goals. Number one is we aim to remove safety officers in the cars, as labor cost is where we should reduce and make a kind of new ways. And number two, we continue to reduce the hardware costs. During the past quarters, we continued to improve our L4 autonomous driving technology and just now, both Xin-Zhe and Robin have given some examples of how we do that, how we leverage the large-scale operation to improve technology. These efforts help us to earn trust and build track records.

Today, we are providing fully driverless ride-hailing services in Wuhan and Chongqing, meaning no safety officers in the car at all. In Beijing, we're also making good progress as the safety officers today are now allowed to -- not to see behind the steering wheel and maybe even the front seats, which will help us to remove the safety officers in more cities and reduce the labor costs in the future.

And on the other hand, for the hardware perspective, most of the newly added car in the coming 12 months will be RT5, the model of Apollo Go. This means there will be some investments in hardware next year. Once a sizable amount of RT6 which we have just launched a few months ago, put in operation in the year 2024, our unit economy will specifically improve because RT6 has much lower production costs than any previous generations of robotaxis.

The unit economy in our operations in key cities is also improving because we continue to scale out and improve the order cost -- the cost efficiencies. Robin has mentioned, our rides continue to grow in the past few quarters and also each car now provide 15 rides every day in some Tier 1 cities.

Apollo Go today is becoming -- recognized as a reliable and efficient way for daily commute. Overall, our Apollo Go measures -- we aim to achieve the unit economy in key cities before we expand to more cities. I think just a summary, all of this impact to our cash flow and the P&L is manageable, and we have the patience.

The next question comes from Lincoln Kong with Goldman Sachs.

I want to ask about our sales of our advanced chips to China because we have seen recently a lot of press report around U.S. chip restriction against China. So we wonder how you are seeing these restrictions impacting your ability to grow the business by AI Cloud, the autonomous driving and lastly, wider our AI business? So could you also remind us our business that's the most depending on the advanced large AI chips and if there's any way, we'll be able to manage this with the master chip capability?

Thank you, Lincoln. You're right. This is a hard topic to reason actually. So the short answer to your question is that we think the impact is quite limited in the near future, and here are the reasons. So first of all, as of today, a large portion of our AI Cloud business and even wider AI business does not rely too much on the highly advanced chips. And secondly, for the part of our businesses that need advanced chips, we have already started enough in hand actually to support our business in the near term.

Thirdly, there were some alternative to the restricted chips and we have the technologies to use these alternatives to achieve almost the same effectiveness and efficiency in our AI Cloud and the wider AI businesses. Last, but not least, automotive chips are not on the prohibited list. So this means that in the near future, in-vehicle computing is not affected.

So when we look at the mid- to longer term, we actually have our own developed AI chip, so named Kunlun. Actually, we already started to use Kunlun chip to support some large-scale AI computing tasks internally. We also use Kunlun to serve external customers already. So because we have full stack AI capabilities from chips to AI frameworks to foundation models and then to application software, so we can achieve much higher efficiency as we optimize AI tasks from end to end.

So let me give you some examples. By using our Kunlun chips (inaudible) in large language models, the efficiency to perform text and image recognition tasks on our AI platform has been improved by 40% and the total cost has been reduced by 20% to 30%.

There are more cases like this in a quality inspection in our smart manufacturing projects and image recognition in the smart city products and so forth. So as our business grows, we believe our end-to-end capabilities will gain us even stronger competitive advantages.

For the core chips, so we expect to see more auto parts, including core chips to be manufactured in China in the future, so as China's intelligent driving market continues to develop rapidly. So this means that supply chain in auto industry may become more and more independent and [relies really] on the imports.

So once again, I want to emphasize that Baidu has built a very strong R&D team. We keep introducing cutting-edge technologies to the market, and we help users and customers to use these technologies to improve efficiency. So that's why we believe we can leverage our strong AI capabilities and AI algorithms to support the top level computing, even though we face some turbulence in the semiconductor supply chain.

So to sum up, chip sales restriction should have a limited impact on our business operation in the near term. Instead, we think increased some good market opportunities for the Chinese chip companies and our equipment AI chips and our AI business will eventually benefit from these opportunities.

The last question today comes from James Lee with Mizuho.

I have 2 here. First, can you guys provide some more color on your progress of e-commerce search and short-video opportunities? Can you guys talk about maybe from a consumer, merchant's and creator's point of view, what kind of pain points are you trying to resolve here?

And second, can you talk about the traction that you have made in open mobile ecosystem and potential monetization opportunities here?

Yes, this is Robin. I mentioned earlier that we have been building closed-loop experience for our users and advertisers in our mobile ecosystem, especially in Baidu App. For e-commerce, because of the efforts we made in the past, users now come to Baidu not only for information and knowledge, but also for services and merchandise.

Since the beginning of this year, merchandise-related search queries on Baidu have grown much faster than last year. You may be wondering why? It is because users increasingly find out that on our platform, they can now find detailed product information, product reviews and even buy the product without leaving our app.

Now we have a huge number of SKUs available for search on our platform. We have also deepened the partnership with leading e-commerce platform, so that users can easily buy products they need here on Baidu App.

Just like I said in my prepared remarks, in Q3, GMV facilitated by the Search continue to grow very fast. I also mentioned earlier that retail has been an outperforming vertical for our online ad business for quite a few quarters. Even very recently, revenues from the retail vertical during the Double 11, e-commerce promotional season, grow by double digits from last year. And this all prove that our efforts gradually bearing fruit.

As for short video, we're making short videos increasingly available in our Feed and Search services. For feed, short video distribution and time spend continue to grow steadily. Right now, about 85% of the feed distributed by Baidu App are short videos. And for Search, more than 20% of the clicks on the search result pages were short videos. This number increased by more than 80% year-on-year during the month of September. So we believe the popularity and the adoption of video in our search results will ramp up quickly.

When we talk about monetization, short videos, especially short videos that provide fully immersive video experience, can be much better monetized than text and images. For example, in feed, ECM -- eCPM for fully immersive video ads are much higher than text and images. This is one key reason for feed revenue to show positive growth year-over-year in this quarter, even though the macro environment was unfavorable. This progress in feed has made us quite confident that short videos will also benefit the search revenue later on.

Now we're also trying to use AI-generated content to enrich our short-video portfolio. We're still at the very early stage for them. We believe with AI, we will generate a large number of short videos in a faster and more cost-effective way.

To sum up, e-commerce and short videos are 2 areas that could drive incremental revenue growth for us, and we will continue to work hard on that. Meanwhile, we would like to stress that profit growth remains a top priority for our mobile ecosystem.

As for interoperability, we believe it is a long-term trend. As we talked about it before, being a search engine, we benefit from more content becoming searchable on the Internet.

Looking into the longer term, we believe the government's willingness to build an open mobile Internet industry remains unchanged because it benefits the users and the SMEs. And for us, by the way, is getting ourselves ready for that trend.

This concludes our question-and-answer session and concludes the conference call. Thank you for attending today's presentation. You may now disconnect.