Baidu, Inc. (BIDU) Earnings Call Transcript & Summary

Hello, and thank you for standing by for Baidu's Third Quarter 2021 Earnings Conference Call. [Operator Instructions] Today's conference is being recorded. If you have any objections, you may disconnect at this time. Now I'd like to turn the meeting over to your host for today's conference, Juan Lin, Baidu's Director of Investor Relations.

Hello, everyone, and welcome to Baidu's Third Quarter 2021 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 Mobile Ecosystem; and Herman Yu, our CFO. 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 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 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 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 also be available on Baidu's IR website. I will now turn the call over to our CEO, Robin.

Hello, everyone. We delivered another solid quarter in Q3 with Baidu Core revenue growing 15% year-over-year driven by AI cloud growing 73% year-over-year. Many years ago, we doubled down on AI, believing that it would expand our market opportunity and accelerate Baidu's long-term growth. Non-ad revenue accounted for 21% of Baidu Core revenue in Q3 and remains a strong growth engine for us. Through AI, we are bringing innovation across the consumer enterprise and the public sector on the backdrop of policy tailwinds and the rising tide of tech innovation in digital transformation, industrial internet, IoT and intelligent mobility. Baidu AI cloud is empowering enterprises from traditional industries and the public sector to better serve customers and move faster with greater efficiency. At Baidu World in August, which was broadcasted on CCTV, we showcased Baidu AI solutions, including smart quality assurance to automate the screening of electronic components of manufacturing assembly line, smart logistics to optimize fleet management and smart monitoring to ensure a steady energy supply from the utility sector. We are excited about Baidu AI cloud's continuing fast growth. Leading companies across industries are adopting Baidu AI cloud solution to improve their operations. And we are working hard to standardize such solutions for industry adoption. An example is our end-to-end AI solution composed of Iaas, PaaS and SaaS financial services sector. Our AI cloud customers are recognizing Baidu's value proposition with strong consumer Internet reach. For example, we are winning smart public parking project with our solution offering the convenience of collect public parking fees via Baidu Maps without the need for meter maids. Hospitals are choosing Baidu for cloud solutions that save patients' trips to the hospital with continued patient care for chronic diseases at home through DuerOS smart display. According to the World Health Organization, there are 1.35 million road traffic deaths every year. Apollo autonomous driving aims to reduce traffic accidents, accelerate mobility to EV and provide greater passenger safety especially after dark. Apollo Go has become the largest robotaxi service provider in the world based on our estimates. Our leadership position in autonomous driving will lend strength to ASD and Jidu Auto. DeepWay, a JV between Baidu and Lionbridge, a leading trucking company in China, recently unveiled its first-generation EV truck powered by Apollo to automate road freight, marking Apollo's entry into the multitrillion-dollar truck market. DeepWay leverages Apollo's leading autonomous driving capabilities and Lionbridge's market reach. Jidu is making great progress as it plans to release the first concept production car at the Beijing Auto Show early next year. Jidu will showcase Apollo's intelligent self-driving and in-cabin features with the aim to deliver mass production in 2023. For the third quarter, China's GDP grew 4.9%, much slower than the 18.3% and the 7.9% growth rate in the first 2 quarters. Our ad business has been impacted by sectors like education, real estate and home furnishing, travel and franchising as we expect this headwind to continue in the near term. As we look forward, there are also positives coming into play. For example, Baidu App is reaching over 600 million users each month. Our hosted marketing solution is becoming ever more popular with site owners. And the trend is for the Internet to become more open. Non-ad makes up over 1/5 of Baidu Core business and grew over 70% year-over-year. We stand to benefit from China's plan to leverage technology to grow the enterprise and the public sector part of the economy. ESG is at the heart of Baidu as we leverage our strong Internet foundation to grow our AI business at scale. Apollo's mission is aligned with lowering carbon emission from making traffic in large cities smoother to accelerating the adoption of EV through Jidu, software for EVs and EV-powered ride-hailing. In June, we announced our goal to become carbon neutral by year 2030. Our Yangquan's Data Center was awarded the carbon-neutral data center leader certification, and MSCI named Baidu as 1 of the 2 companies in China with transparent reporting on greenhouse gas emissions in its Net-Zero Tracker. Turning to Q3 operational highlights. Baidu AI cloud continues to see strong growth, leveraging our world-class AI technologies and our capabilities to provide one-stop shop solution to our customers. Our AI solutions are being increasingly adopted across different industries and for different scenarios. We enriched our cloud offering with the release of an end-to-end AI cloud solution powered by Kunlun AI chip and PaddlePaddle deep learning framework to help financial services firms digitize and automate their operational processes, enlisting leading customers like China Life and Bank of Jiangsu. Baidu AI cloud is expanding into industrial Internet by partnering with industrial parks and municipalities. Our deployment to serve a large pool of enterprises is shortening. Last year, we spent 7 months implementing our AI Paas in Guiyang Economic and Technology Development Zone, serving approximately 100 enterprises. Our deployment of AI PaaS to enterprises in Tongxiang, including those in their economic development zone, took only 3 months. Tongxiang hosts 40,000 enterprises, mostly manufacturing-based. And we believe our AI solutions can help these companies improve their operational capabilities. Baidu AI cloud is moving into smart city. Lijiang, a UNESCO heritage site with 800-year-old bridges and waterways in Southwest China is using Baidu AI solutions to keep the city state and claim for visitors. Leveraging the digitization of tourist areas, Baidu AI cloud help local authorities timely detect and address infractions such as illegal parking and public littering. Our success in Lijiang is leading to projects with Chengdu, Suzhou and other cities. Moving to ACE smart transportation. Baidu ACE has been adopted in 24 cities, up threefold from last year based on contract size over RMB 10 million. About half of ACE projects signed in Q3 came from repeat customers. In addition to supporting policy tailwinds, ACE is benefiting from the value we deliver to the public sector. For example, the ACE smart transportation solution that was deployed in Guangzhou earlier this year helped improve congested traffic by over 30%. Smart parking has made it easier for local residents to find available parking spaces, shortening travel time. Turning to autonomous driving. Apollo Level 4 testing has accumulated over 10 million miles or 16 million kilometers, setting a new milestone. We continue to receive green light in our application for autonomous driving testing as we demonstrate to local authorities Apollo's progress in technology and operation. Apollo has received 411 autonomous driving permits, an increase of 237 permits from a year ago. Our progress in autonomous driving technology, operations and continuous cost reduction positions us for ridesharing monetization in a few years as paid ride scale. In July, Baidu opened its third Apollo Park in Shanghai, a 10,000 square meter facility housing Apollo Cloud and data operations. Shanghai marks Apollo Go's 5th ride-hailing service opened to the public following the cities of Beijing, Guangzhou, Changsha and Cangzhou. In August, we introduced the Chinese brand for Apollo Go, Luobo Kuaipao, a fanatic interpretation of robotaxi in Chinese. Apollo Go is resonating with ride-hailing passengers with total rides doubling quarter-on-quarter to reach 115,000 in the third quarter. Apollo Go is off to a good start as we set an ambitious goal to expand operations into 65 cities by year 2025 and 100 cities by year 2030. Turning to ASD and DuerOS in-vehicle infotainment software. WM Motor, a Chinese EV OEM, signed with Baidu to install Apollo navigation pilot in its new W6 SUV, taking Apollo partner network to 31 makes, including those from automakers like GM, Ford, Toyota, Honda and Great Wall. Turning to Jidu Auto. Jidu has completed its first vehicle wind tunnel testing. It is a full-sized oil and clay model. We have identified the model design schemes and engineering development and design optimization are underway. Jidu will showcase the latest ASD, demonstrating the appeal of smart EVs with advanced autonomous driving and smart in-cabin features. We hope the success of Jidu will increase appeal of Apollo solutions to other automakers. On DuerOS. Xiaodu again was ranked the #1 in smart display globally and #1 in smart speaker in China based on second quarter shipment according to Strategy Analytics, Canalys and IDC. A key growth driver for Xiaodu's smart display is the shift of user time spend at home to large screens for services like karaoke, short and long videos, online games, education services, video conferencing and other visually oriented activities. Popular mobile apps are appearing on DuerOS skills, like Douyin, Kuaishou, Red, Dianping, 58.com, HEYTEA and Kentucky Fried Chicken, to name a few. In addition, Xiaodu completed its Series B financing at a market valuation of $5.1 billion in August, 9 months after Series A financing at $2.9 billion. Baidu remains a super majority shareholder after the Series B round. The continued strong performance of our new AI business will not have occurred without our relentless investment in technology and our pursuit of better products and more customer adoption from our technology leadership. Baidu released PLATO-XL, the world's first 11 billion parameter pretrained dialogue generation model, achieving breakthroughs in Chinese and English conversations. PLATO allows for different kinds of conversations with users, including chit-chats, knowledge-based dialogues and conversational Q&A, which will be incorporated into Xiaodu in the future. Turning to mobile ecosystem. Daily log-in users in Baidu App reached another all-time high at 79% and in-app search queries was up 11% year-over-year, reflecting a better search experience from Baidu app. We continue to focus on deepening the service offering of our key verticals. For example, Baidu Health has built a strong community of medical experts that allow users to frictionlessly move from search to telehealth consultation. We recently enabled doctors in our network to write prescriptions for online consultation and home delivery. Baidu app now offers instant replies to search queries from over 30,000 industry participants, spanning 19 industry verticals. Instant replies were up fivefold from a year ago and 40% are paid services. For example, users who search for tips on job interviews can readily connect to career advisers to get help on reviewing their resumes and selecting career training programs. Instant replies have enriched Baidu's search experience, further driving the vibrancy of our in-app search. Revenues from Managed Page reached 43% of Baidu Core advertising as we make doing business online for merchants simpler through Baidu's hosted marketing cloud. For example, Meinian Onehealth, a physical checkup clinic chain in Shanghai, opened a Managed Page storefront, allowing users to browse its services, chat with customer service, make payments, leave comments and share reviews. Monthly orders increased fourfold 3 months after using Managed Page promotion. Search by nature is open and we are benefiting from the trend to make Internet more open in China. Our e-commerce feature is seeing good progress. In September, the number of third-party SKUs from China's top e-commerce sites, searchable on Baidu, reached almost 1 billion. And e-commerce GMV on Baidu, though small, grew 90% sequentially. The open nature of our mobile ecosystem is leading smart speaker -- smartphone makers to select Baidu's Smart Mini Program as the landing page for their browser search. MAU from such collaboration has reached 24 million. And we believe such trend will enhance user experience and monetization or union search. Last but not least, I want to welcome Rong Luo, who has joined us recently as our new CFO. Rong brings a wealth of financial management and capital market experience. I also want to thank Herman for his 4 years of service as Baidu's CFO and we expect more great things from him as CSO. With that, let me turn the call over to Rong to go through the financial highlights.

Thank you, Robin. Hello, everyone. I'm very excited to have joined Baidu recently and participate on my first earnings call at Baidu. I look forward to meeting with everyone in the coming months. Now let me walk you through the details of our Third quarter 2021 financial results. All monetary amounts used in my discussion are in RMB unless stated otherwise. Baidu's Q3 revenue was RMB 31.9 billion or USD 5 billion, up 13%, 1-3, year-over-year. Baidu Core's Q3 revenue reached RMB 24.7 billion or USD 3.8 billion, up 15%, 1-5, year-over-year. Non-advertising for Baidu Core reached RMB 5.2 billion or 21% of Baidu cost revenue. AI cloud revenue was RMB 3.8 billion, up 73% year-over-year. Our AI cloud growth is benefiting from the demand for cloud services by customers from the Internet media, financial services, energy, manufacturing and the public sectors. ACE smart transportation is another growth driver for AI cloud revenue. We are finding that automakers who subscribe to our infotainment solutions are also interested in Apollo ASD. Thus we are reclassifying ASD revenue into AI cloud. ASD revenue is quite small now and adding approximately 1% growth year-over-year to AI cloud as a result of the retroactive adjustment. On ID and OGI, Xiaodu continues to make great progress. Demand for higher ASP products keeps growing as customers come to appreciate the value of DuerOS smart assistant and service revenue continues to grow. Baidu Core ad revenue was RMB 19.5 billion, up 6% year-over-year. In-app search advertising was solid, partially offset by the weakness in unit revenue. iQIYI revenue was RMB 7.6 billion, up 6% year-over-year. iQIYI subscribers reached 104 million in September. And the membership revenue was up 8% year-over-year mainly due to the refined membership strategy and improved monetization capabilities. Cost of revenues was RMB 16.1 billion, up 26% year-over-year primarily resulting from an increase in tech content courses and the cost of sales associated with new AI business. Operating expenses were RMB 13.5 billion, up 46% year-over-year. During the third quarter, we incurred a contingent loss of RMB 976 million, pertaining to the legal proceeding involving former advertising agency. Excluding such contingent loss, operating expense was up 35% primarily due to an increase in channel spending, promotional marketing personnel-related expenses. Non-GAAP operating income was RMB 4.6 billion or USD 731 million, and non-GAAP operating margin was 15%, 1-5. Non-GAAP operating income for Baidu Core was RMB 5.8 billion or USD 904 million. And non-GAAP operating margin for Baidu Core was 24%. Adjusted EBITDA was RMB 6 billion or USD 925 million. And adjusted EBITDA margin was 19%, 1-9. Adjusted EBITDA for Baidu Core was RMB 7 billion or USD 1.1 billion. And adjusted EBITDA margin for Baidu Core was 28%. Cash and short-term investments for Baidu Core as of September 30, 2021, was RMB 183.6 billion or USD 28.5 billion. Free cash flow for Baidu excluding iQIYI was RMB 2.9 billion or USD 449 million. Baidu Core has approximately 39,000 employees as of September 30, 2021. As Robin mentioned, China's GDP growth has slowed over the last 3 quarters. And sectors in our app businesses, such as education, real estate and home furnishing, travel and franchising have been negatively impacted. With the resurgence of COVID-19 in many cities recently, local governments are putting in place preventive measures, including quarantines, travel suspension and mass testing. Consequently, our own advertising business may remain soft in the coming quarters before normalizing. Turning to Q4 guidance. For the fourth quarter of 2021, Baidu expects the revenue to be between RMB 31 billion, which is USD 4.81 billion, and RMB 34 billion, which is USD 5.27 billion, representing a growth rate of 2% to 12% year-over-year, which assumes that Baidu Core revenue will grow between 5% and 16%, 1-6, year-over-year. The above forecast takes into consideration that the current COVID-19 situation in China, which is still evolving, and business visibility is limited. The above forecast reflects our current and preliminary view, which is subject to substantial uncertainty. Before I turn the call to operator, let me recap this quarter. China is adjusting its economic growth drivers with the introduction of a new 5-year plan in March this year. Historically, the Chinese economy has been very resistant to such adjustments and new drivers allow the GDP growth to come back. We are optimistic about China's future especially when the current COVID-19 situation comes under control. It's quite clear that China will leverage technology to grow the enterprise and public sector portion of the economy. Leveraging AI, our cloud services, our cloud business continued to outperform the market, growing 73% year-over-year in the third quarter. Apollo is making great progress from L4 autonomous driving testing surpassing 10 million test miles to robotaxi ride-sharing doubling rides sequentially to ASD signing on new partners to Jidu finishing the wind tunnel testing of its first car model 8 months after its CEO joins to build a team. We are moving at China speed. China is doling out new policies to support great energy. We hope that Baidu will benefit from the promotion of decarbonization as we leverage AI to minimize traffic congestions in dozens of cities across China and help accelerate the switch to EV with Jidu, ASD and Apollo Go robotaxi. On mobile ecosystem, Baidu MAU reached 607 million, up 12% and daily login reached 79%. Our app business is susceptible to macro environment. And we expect our ad growth rate to peak back up when GDP growth reaccelerates. Operator, with that, let's now open the call to questions.

[Operator Instructions] Question comes from the line of Alicia Yap from Citigroup.

Congratulations, Rong, on your new role. I actually try to fit in two, if I may. So for the fourth quarter core -- Baidu Core revenue guidance, you provided a wider range, which is 5% to 16%. Can management elaborate the scenario and the situation that you bake into the low end of this 5%? What would be the implied core ad revenue growth versus the AI cloud revenue growth if we end up in the low end? And then similarly, what would be the scenario if we end up in the high end of the revenue guidance of 16%? Would that come from better macro or more cloud project that will be closing? And then secondly, just in this overall broader regulation backdrop. So have Baidu started to have any discussions with other major Internet peers as related to this potentially opening up of their social network content to the search engine? If this were to go ahead, what is management view on the potential benefit and upside to Baidu fundamental going forward? And also any discussion you have with some of the e-commerce leading player to allow Baidu Core into the e-commerce content item on their digital forefront? So any color you can provide would be appreciated?

Alicia, I will have Herman answer your first question, and Dou will answer your second one.

So with regards to our guidance, we normally guide -- internally assume that we will hit the midpoint. You asked what do we factor in and what would it be if it goes to the low end. I think obviously, as Robin has mentioned, 21% of Baidu Core is non-advertising and almost 80% is advertising. And as we're in the process of trying to control COVID-19, that obviously is the highest risk right now. In our model, we assume that COVID-19 will be under control for the most part in China by the beginning of December. We have seen since Guangzhou's situation that normally 60 days after that happens, that it gets under control. But this time, it spreads to several cities. But given the pattern that we've seen, given the -- for example, in Beijing, we've seen the cases slow down. We think that's a doable situation. Secondly is that the other side of our business is mainly cloud, right? So when you think about the cloud, trying to recognize revenue, trying to make sure that the software is installed, it's a more complicated process than advertising. So as you're getting into year-end, the risk there is that we deliver the software, but it's not up and running so we cannot recognize revenue. And that's always the case when you have 2B businesses. So I think that -- there's a risk there with our cloud business. There's a risk there with smart transportation. So I think those are the key risk factors, I think, I would call out. But as we said, normally, when we go out with the guidance, given the information that we have today, we think we're more likely to be somewhere in the middle.

Alicia, I'll take the second one. As you know, search by nature is open. And we believe being able to share content across apps provides a better user experience. And also, we believe this will be a going-forward trend. So actually, we already see things are happening around this direction. For example, some developers are more willing to open up their ecosystem nowadays because it's also good for their own business. So as Robin already mentioned, on our e-commerce effort, almost 1 billion SKUs from the top e-commerce sites were now searchable on Baidu. And the GMV from Baidu grows 90% sequentially. So actually another great example I want to share, which Robin already mentioned, is that the major smartphone makers in China are adopting our open-source Smart Mini Program framework for their in-house browsers. Now with this adoption, users will have better experience and the developers are significant -- can significantly extend their reach through the major browsers in China. So you're right. We believe we'll benefit from the openness of the whole Internet. And we are seeing things are happening that way.

Next question comes from the line of Piyush Mubayi from Goldman Sachs.

When I look at the guidance that you talked through, Herman, it appears to be generally slowing down. And we realize it's a very high base for the cloud business in the fourth quarter of last year. So as we look at that pace of growth and we go through the core, which looks like it's 10% to 15% -- sorry, 5% to 16% range, and we try to estimate where are the -- in the advertising vertical, the weaknesses and whether it's COVID related or macro related or third-party related, as you exited third quarter, what was the pace of decline that -- with the commencement of the fourth quarter? If you could take us through that pace, so we can understand where we might end up to a certain degree? And also, I know it's very early to talk about it, but broadly speaking, into 2022, with the opening up of platforms, with so much that's going on from a business perspective on the AI front as well as on the cloud front, what is the sort of pace of growth we can expect on the core?

Piyush, can you repeat that last sentence? I missed that.

As we look at -- so we have seen a slowing down into the third quarter, which is probably one of the reasons why you're giving a very wide range for 4Q for the core. But there's so much that's going on that is -- that leads to a level of optimism for what 2022 would look like, including the opening up of the systems, opening up of platforms, how much search you can originate in the e-commerce vertical -- and that's just the e-commerce space. I don't know how many other verticals you can tap into. So as we think through 2022 and the core pace of growth, any feelers around how that would be, would be great.

Okay. Let me answer that. So we talked about Q3, right? So when you look at advertising in Q3, we've been gradually slowing down in terms of our growth. And in Q3, what we've seen, sectors that have impacted us. Some of these are pretty obvious, like education, like real estate and home furnishing, like travel and franchising. These would probably be more related to COVID-19. So these kind of situation are going to extend into Q4. And we suspect that this might be a multi-quarter impact. So based on what we have seen right now, we expect that advertising for Baidu Core, which was 6% year-over-year in Q3, will probably go into even a slower growth rate in Q4, assuming that COVID-19 gets under control by the beginning of December. So a lot of uncertainty. So if that continues to not get -- if that continues to have new cases every day, then our advertising could even perform worse than that. So I think a lot of it has to do with COVID-19 but there are other factors that's impacting that, such as regulations and so forth. I think with regards to the AI businesses, I think COVID-19 does impact that a lot. As we've talked about before that, our AI cloud is based on solutions. And we have to customize for certain customers and so forth. So if we are unable to travel or we have -- we're not able to be as fluid as before, that will have some impact. Others are just, end of the year, trying to complete these enterprise solution. And sometimes given the difficulties that we have right now traveling about and so forth, might not make it. So I think we're just putting that uncertainty in there. With regards to our range, actually, we've been pretty consistent. If you look at it over the last year, it's -- at the Baidu consolidated level, it is always a plus or minus 5%. I don't think we've made any changes in the recent quarter. So going into 2022, it's kind of hard to look out so many quarters into '22. But what we have seen for advertising, as I mentioned, we think that this could be a multi-quarter impact because, for example, COVID-19 comes and goes and then the regulations and so forth. And then when we look at -- talk to our peers in the market, there seems to be a consensus that this could be beyond 1 quarter. So I think that's where we are with advertising. Sure enough, there's also that upside with the opening up with different apps and so forth. But it's hard to bake that in until we actually see more development. So we'll keep you guys apprised as that comes. The other side of this is that we're pretty optimistic with our AI businesses. Robin has [indiscernible] year-over-year. If you look at the last few quarters, we've been always growing above the market speed. We continue to believe that we can grow above the market. If you look at our NDR, for example, it's been gradually improving or increasing, if you -- so we're coming -- feeling pretty good about that. One of the things is that if you look at our cloud business, SaaS business is the fastest piece right now. We were ranked by IDC as one with the largest solutions in China. And this is a piece that's the fastest of our cloud solution or our cloud [indiscernible]. So I think we're pretty -- feel like we're in a good spot right now because of the policy trends supporting technology, supporting -- leveraging AI for enterprises and also for the public sector.

Yes. Let me just add about the opening up of e-commerce platforms that will enable users to do transactions on the Baidu Mobile Ecosystem, especially Baidu apps more easily. Given that the e-commerce infrastructure in China has matured, it's very easy for people to pay using their mobile phone and it's very quick to get something delivered to our home. And now that we can index a lot of SKUs on Baidu, we expect a very quick increase in terms of GMV from a low base. The revenue contribution directly from transactions will not be large. But once users get used to do transactions on the Baidu platform, I think the overall conversion rate for our advertisers will also improve. So I think that the benefit will be well beyond simple transactions on the Baidu platform.

Your next question comes from Alex Yao from JPMorgan.

Rong, welcome back to the Internet investment community. So I have 2 questions. Number one is we have seen a lot of regulatory and operating environment challenges recently, for example, Apple's IDFA policy change; for example, the implementation of PIPL on the 1st of November. Can you guys talk us through the operational and the financial impact from those changes, particularly how should we think about the union business into the next couple of quarters? And the second question is regarding the general corporate strategy. Well, the advertising revenue outlook is weak for the next couple of quarters. How do you guys think about the general corporate strategy? For example, are you still pushing the investments on headcounts and user acquisition costs? And I think in the past, you guys had a very tough cost control when the top line was under pressure. Are you thinking very differently this time? Any thoughts on the general corporate strategy in the current advertising revenue outlook would be helpful.

First one, so on the IDFA side, right, because the majority of Baidu Core ad business does not rely on user data, it's not like some of the U.S. companies do. So the impact on that side is less pronounced actually. So in general, as we all strive to be a global technical -- technology leader. And issues such as this, type of data privacy and security, has been taken into consideration for a long while. So many years ago, before this topic even caught the industry's attention, we saw the need to establish internal committees to define strict policies and procedures to address data privacy, data management and data security. Also, we are working with the leading companies globally and domestic to create ideas and derive best practices. We also share our experience with authorities to help the industry address these complicated issues. So given all the efforts we have done, I think the impact on this side is not that pronounced.

So Alex, you mentioned about our corporate strategy. So where we are at this time is that we have our expenses, you can look at mainly 2 buckets, right? One bucket of spending is related to our growth with our new business, whether that's in cloud, whether that's in intelligent driving and so forth. And you've got to invest ahead of time to make sure that your sales force is there, to make sure that people who are doing presales is there, to support your future growth. And as we mentioned earlier, that piece of the business is still growing very fast. When you look at non-advertising, we're growing over 7%. So you've got to build a funnel. You've got to get the people in and so forth. I think that's going to continue. Secondly is a new AI business has a higher cost of goods sold, whether you're talking about selling smart devices, whether you're talking about delivery for -- to these solutions or smart transportation, okay? So those things, we have to factor in for this new business. That's just part of it. And you look at the gross margin very carefully, but you still have to fund the OpEx, the salespeople, the development people to support their business growth. And we're leading in those businesses. So we feel comfortable continuing to invest. The second pool of investment that we have is with the Mobile Ecosystem. And what you see there mainly is supporting our Baidu ABP. So a big part of that spending is channel spending, marketing spending for the Baidu ABP. This year, we've been pushing our Baidu ABP light, which would allow us to tap into users from lower-tier cities and that has been working. And in addition to that, we've been beefing up some development in that area to help improve our search experience. So I think we manage that side of the expense, especially channel spending with ROI. So as long as we're seeing positive ROI, we will continue to spend in the channel and marketing areas.

Yes. I would just like to emphasize that our investment is pretty much packed with the growth potential. We see a lot of growth potential in the AI-enabled new businesses. So we will continue to aggressively invest in that area. Mobile Ecosystem in general, not just for Baidu, I think the overall mobile Internet market in China is maturing. So we will use discipline to balance our investment and growth.

Our next question comes from James Lee from Mizuho.

My question is on intelligent driving, autonomous driving. And obviously, there's a lot of moving parts in this business right now. I think you guys already lay out your plans of robotaxi. Can you also lay out the road map and how Apollo can achieve a meaningful presence in connected infrastructure and OEM licensing your technology? So what I mean is that does your footprint of your smart transportation influence OEM's decision to choose their autonomous driving solutions?

James, Apollo is a platform for both smart transportation and autonomous driving. It has gained a lot of partners, customers, and so on. We've been a leader in many of the fronts, including robotaxi, as I mentioned. We are probably the largest robotaxi service provider in the world by a number of rides we provide. And it is open to the public in 5 cities in China. And for many of those areas, we also have smart transportation projects going on. Government pay us to install roadside units and also software systems to better manage the traffic and help autonomous driving become safer and greener. We also work with quite a number of OEMs to provide infotainment system as well as other pilots or navigation systems, what we call ASD. And we have gained quite a number of OEM customers. So all in all, as a platform, Apollo connects smart transportation OEMs, ride-hailing services altogether. And we are quite confident that we will continue to lead this market and even become dominant in the future.

Your next question comes from Jiong Shao from Barclays.

First off, big congrats to both Herman and Rong on new roles. My question is related to your Apollo autonomous driving and robotaxi as well. Just following up on the previous question, could you elaborate a bit on the top line for Jidu in terms of when the assembly line manufacturing facility will be ready? What kind of scale we are talking about when you talked about mass production in 2023? And you talked about your partnership with Lionbridge for semi-truck. I understand Jidu just announced their ambition for the EV semi-truck business a couple of days ago. I think I was wondering, since they are your big partner, do you have any plan to work with them for their semi-struck EV initiative? And then lastly on robotaxi, I recall -- I may have read somewhere, you may have commercial operations up and running for robotaxi sometime next year. Is there anything there you can share with us?

On Jidu, like I mentioned during the prepared remarks, we are planning to launch the concept production car early next year at the Beijing Auto Show. So that car will be the final shift in the architecture for when it is available for sale in the latter part of 2023. So by mass production, we mean everyone, every consumer can place an order or to buy the Jidu car sometime in the second half of 2023. Of course, it's too early for us to predict how many units we can sell at that time. But we are all very excited, the features that we're putting in and the value we can offer to our consumers when it's available. In terms of autonomous trucking, it's a huge market. We already have a partnership with Lionbridge. And Jidu, as you know, is a strategic partner of Baidu too. We are open to all kinds of collaborations with them. And again, these kind of collaborations are not exclusive. Apollo is an open platform. We can work with everyone. And we would like to help a lot of companies to become successful. And on the commercial operation of robotaxi, we can charge right now in a couple of the areas in Beijing and Hangzhou, and we are applying such licenses in other cities. So it's a very quick ramping-up process. We are already providing -- we already provided 115,000 rides in 1 quarter in Q3. And in Q4, I think that number will continue to increase. And this is probably much, much larger than the reported number you can hear anywhere else in the world.

Our next question comes from Natalie Wu from Haitong International.

Congratulations, Herman and Rong for your new role. My question is regarding your investment plan for your new AI business. So firstly, if we take -- we just look at the development history of AI cloud, when the revenue scale approached CNY 20 billion annually, EBITDA loss ratio actually [ narrowed ] significantly. So should we expect the margin profile of cloud business to improve next year after AI cloud revenue scales, which is also about RMB 20 billion scale? Or how should we see the margin trend of your AI cloud business in the next 1 to 2 years? Also for the intelligent driving, what kind of investment scale should we expect for the next 1 to 2 years? And how could that impact your margin?

Yes. So on the AI cloud, I think -- yes. I think when you talk about other cloud players, they are talking about scaling with regards to IaaS right? As we talked about before, IaaS is a smaller piece of our business. Where we're more excited about, where we're differentiating in the market is our SaaS. And as I mentioned earlier, SaaS is the fastest-growing piece of our business. And when you look at SaaS, as we talked about at Baidu World, there's several factors with regards to margin improvements, right? As we come out with new solutions, working with new customers, the first project -- first few projects, we might have to spend that extra time in order to integrate with their legacy system, in order to build tools and so forth for it to go up and running. When we're going into new industries, that would also -- there appears that ramp-up cost to be able to adapt our current solutions to that new industry. So you're seeing with our fast growth, as Robin mentioned, 73% year-over-year growth is much faster than market right now. And we could not have done that without going to new industries with new solutions, without consistently getting new customers. So at the early stage where we're just expanding market, expanding customers, you're not going to see good margins. But when you look at a particular solution over time, when you're looking at a particular customer over time, you're seeing margin improvements. And we went through that in very -- detail doing the Baidu World presentation. And our IR can go through that with you again on those things. But this is a proven business. If you look at our counterparts in the U.S. who are in the SaaS market, you see the same trends. So we're pretty confident that as we continue to grow our business this year, in the beginning, you're not going to have a good margin because you have to customize. But as the product line becomes more standardized, as we become more standardized solution by industry and so forth, you're going to see that margin pick up. That's number one. Number two is the proportion of SaaS is getting higher versus IaaS. And IaaS, as we know, because it's a commodity cloud, is just going to have lower margins. So you have several trends that are playing in our favor. So I can see it a few years out, you're going to see margins improving, okay? Intelligent driving, I think there are 3 pieces as we talked about before. Autonomous driving is a technology. And then out of that, there's 3 ways to monetize, right? There's Infotainment business, right? There's the -- what we call ASD -- the autonomous driving navigation pilot, so forth. So that is a business where you have to invest upfront with your R&D and so forth. But when we sell it on a gross margin level, that gross margin is usually pretty healthy compared to other AI businesses, okay? There's -- the other piece for Apollo is smart transportation. Smart transformation is mainly with the government sector and so forth. It's usually a very complicated solution and so forth. Those margins are pretty healthy especially when you compare to a business like IaaS. So I think when you're going into smart transportation, the factors that impact the margin, very similar to what I described earlier, coming out with new solutions. When you're working with new cities, you've got to make that investment with new solutions. But as we're working with the same city, we understand their legacy system. We build out the tools and so forth. Over time, with a particular customer, we're seeing the margin improvement, right? So I'm pretty confident because when you look at us, we're growing on the one hand, as we mentioned -- Robin mentioned smart transportation. The customers went up threefold year-over-year. So the new customers, they're going to adapt the margin. On the other hand, half of that business is with repeat customers. That's going to improve. So I think over time, we see as we go out for the next few years that more and more revenue is going to be recurring, which is going to [indiscernible].

Our next question comes from Gary Yu from Morgan Stanley.

I have one follow-up question regarding the opportunities with the open up -- further open up of the Internet. It seems like there is a great opportunity from both the kind of e-commerce space and also potentially content space. But at the same time, given users have already got into the habit of doing those transactions in some of the other competing platforms, how are we going to kind of change some of the user behavior to start doing more kind of e-commerce transaction on Baidu? And as a result of that, are we going to kind of go into another kind of investment cycle in terms of getting traffic and users on these transaction-based traffic?

Yes, Gary, I'll take this question. Actually, as we see the users' behaviors in our web search, before the people do their -- make their final decision to buy something or not, so they come to Baidu for more information. So this is how we step in to help them to do the transactions actually. So also, as Robin just mentioned, right, so the infrastructure for transactions in terms of the payment delivery or -- is pretty mature in China. So that said, once we give the user a comprehensive information for them to make the right decisions, people will come back again for their another transaction. In terms of getting more traffic, I think for now, we still have tons of actually user search volumes in Baidu for not just the typical e-commerce queries and all other queries related to transactions. So that's why at this stage, we are trying to make the transaction experience as smooth as possible on Baidu's platform. And actually, we have been working on this for a while since we started building Smart Mini Programs, Managed Pages and Baidu Health. So that's how we can not only get the right information but also have the right support for these transaction-related queries.

Our next question comes from Eddie Leung from Bank of America.

May I have 2 questions on your cloud business? The first one is about hybrid cloud and private cloud. As the government is pushing for the awareness of data security, will there be a potential increase in demand for private cloud and hybrid cloud versus public cloud? And how might that offset your basis? And then secondly, I remember, Robin, you mentioned earlier a pretty interesting point. You guys have been focusing your cloud business on SaaS and end-to-end solutions while some of your industry peers seem to be emphasizing more on the -- shall we say, the modules or the component capabilities and performance. So probably a bit more about PaaS, at least from a marketing perspective, right? So could you share your thoughts on any difference between the clients that these 2 approaches may get?

I think you're right. Given the regulatory environment changes, I think more and more customers would want private deployment or a private cloud or hybrid cloud rather than public cloud. And like Herman mentioned before that our higher growth areas and also higher-margin area comes from those kind of private placements worked or solutions offered to our customers, typically end-to-end solutions, leveraging our strong capability in all kinds of AI value chain, PaaS, all included, AI chips frameworks, deep learning frameworks. And when we say SaaS, it means more like end-to-end solution, which typically would include the PaaS part. It's just an AI PaaS in China or PaaS in general in China. It's not a big market yet. It's growing very fast, but it's very hard to charge a lot of money for a middle layer of the solution. But yes, I think a lot of customers do value our end-to-end capability and solution in terms of AI. And I think, overall, it's a very large market. And our strength is to really leverage Baidu's investments in AI and provide end-to-end solution to our customers in important verticals like industrial Internet, smart transportation, energy, financial services, these kind of verticals that we think we have a very strong proposition.

So we have reached the end of the question-and-answer session. So with that, we conclude our conference for today. Thank you for participating. You may all disconnect.