Baidu, Inc.
Q1 2018 Earnings Call Transcript

Published: 27 Apr 2018

[Operator:]

Hello, and thank you for standing by for Baidu's First Quarter 2018 Earnings Conference Call. At this time, all participants are in a listen-only mode. After management's prepared remarks, there will be a question-and-answer session. Today's conference is being recorded. If you have any objections, you may disconnect at this time. I would now like to turn the meeting over to our host for today's conference, Sharon Ng, Baidu's Director of Investor Relations.

[Sharon Ng:]

Hello, everyone, and welcome to Baidu's first quarter 2018 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, Baidu's Chief Executive Officer; Qi Lu, Baidu's Chief Operating Officer; and Herman Yu, Baidu's Chief Financial Officer. 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. Potential risks and uncertainties include, but are not limited to, those outlined in our public filings with the SEC, including our Annual Report on Form 20-F. Baidu does not undertake any obligation to update any forward-looking statement, 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 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 made available on Baidu's IR website. I will now turn the call over to our CEO, Robin.

[Robin Yanhong Li:]

Hello, everybody, and thank you for joining today's call. We had a strong start in 2018, with our core business exhibiting robust growth, and we continue to execute on our strategy to strengthen Baidu's mobile foundation and lead in AI. In the first quarter, our strong revenue growth was driven by strength in our traditional search business through better monetization capability, powered by AI technology. For search and feed, we have focused on improving the user experience of our twin engine Baidu App. As we extend the content offering from Baijiahao or BJH accounts and enhance the algorithms of feed, in March, we saw the DAU of Baidu App reach 137 million, up 18% year-over-year and total daily time spent up 30% year-over-year. Our flagship app has a strong, distinct advantage because we power our feed algorithms with Baidu's massive AI computing and big data assets to provide our users with personalized recommendations based on the wide range of activities that they conduct on our platform. Let me talk a little about Baidu's product balance. We fundamentally believe that user engagement should be viewed on content that enrich our users and have a positive societal impact. For example, we proactively clean up indecent content and limit the volume of entertainment gossip news in one sweep. Our product philosophy is to broaden our user base and expand content consumption across a wide range of high-quality content. We have improved information quality such as showing Baidu Encyclopedia content in feed, using space which was previously reserved for feed ads to help users to identify fake news. We use AI to identify and remove clickbait and vulgar content. Upholding such product values may mean putting in greater effort to grow our users and revenue. We believe such a strategy will pay-off in the long run. On search, we are pushing the boundaries of Baidu App experience. Last November, we launched the Baidu Bear Paw Account, which enables brands and businesses to link their content from various websites and social platforms, display them in a standardized format and build social following through follow, announce, payment and PIN verification features. In the first quarter, we strengthened the communication feature by enabling users to message or call friends and businesses with Bear Paw Account directly from their brand app. Such interactions are recorded in the user log enabling users to locate the brand at a later date, and enabling brands to continue the conversation with interested users through Bear Paw Account announcement. By channeling the value of search results to enable the divestment of social assets in the Baidu App, Bear Paw Account significantly improves the value proposition of Baidu Search from brands and businesses. Equally important, Bear Paw Account now covers approximately 30% of Baidu Search results, resulting in a better search experience for Baidu App users. On the advertising front, Baidu AI experienced robust growth in search and feed monetization. Dynamic ads, for example, uses machine learning and reach advertiser data sets such as product catalog to display highly relevant, personalized product ads to our users, driving higher conversion rate. In Q1, we upgraded our dynamic ads by layering in Baidu big data on top of the product catalogs from advertisers. With the new AI-based dynamic image feature, each user sees a slightly different personalized ads based on the users preference on Baidu's platform. For example, in March, S-CAR, an automotive portal, was one of our dynamic ad customers, and used our AI-based dynamic image feature to promote the most fitting car model to users on a real-time basis. Early results show dynamic app with the new dynamic image feature have increased click-through rates by double-digits and have the potential to serve advertisers with product catalog such as those from the retail, auto, real estate and travel agency spaces. We are in the early innings of leveraging AI to improve app relevance and effectiveness. Beyond search and feed, AI is enabling Baidu to address two other huge markets

[Qi Lu:]

Thank you, Robin. Let me provide you with a set of next level operating details that are key to powering our Q1 operating highlights that Robin just went through. For search, we are making steady improvements. Our engineering and product capabilities to index richer and higher-quality content to enable more search query modalities and to present search results via more engaging visual and interaction designs that work better on mobile devices. As a key indicator of search progress, the coverage of top one results whereby we satisfy user queries with the first search result has reached one-third of queries in Q1, up from 30% in Q4. Another big part of our effort was devoted to building out and operating the Bear Paw Account platform, as it is a fundamental part of revitalizing the mobile web and fueling the future growth of search. In Q1, we secured a series of high-quality publishers and many of them managed their video content via Bear Paw Account. For feed, we are making stride in elevating our engineering infrastructure and the product operation capabilities. For example, we deployed a new architecture that will enable much faster iterations of algorithm development, leading to a faster pace of product innovation. We are building and deploying more AI capabilities, especially in our Baijiahao content platform to offer AI-powered answering tools, which lead to more and high-quality content on our platform. As a result, our product quality and the user engagement continues to gain ground throughout Q1. Video continues to be a key focus area for us. And in Q1, we further expanded our video content coverage in feed to 48%, up from 19% one year ago. For flagship Baidu App, we continue to make good progress in elevating the underlying performance and AI capabilities of the client code base. At the same time, we systematically improved product capabilities, especially in richer user interactions through a variety of features, such as social sharing, micro channels and smoother content navigations. On the advertising front, we'll continue to build up strengths by developing more advanced AI technologies, deeper and broader data assets and stronger operation capabilities. One highlight in Q1 is that for the first time, we deployed a powerful reinforcement learning based infrastructure that can significantly improve our ability to better match ads to our users and increase click through rates and conversions. We also reached a new engineering milestone with a Dual Cloud DNN callback model, which has increased click through rates meaningfully. These are foundations on which we build up and operate an increasing suite of levers that can sustainably drive advertising performance. Our OCPC program is one such strong lever. In Q1, with OCPC, we simplified the bidding process for our customers and are able to offer more reliable online and offline conversions. As a specific example, Chang'an, a Chinese auto OEM, used OCPC and saw click through rates increase significantly and the cost per conversion decline meaningfully. As a result, Chang'an increased daily average spend on search and feed fivefold. We plan to roll out OCPC to more verticals throughout 2018. Dynamic ads and region action ads are additional such levers that will continue to enhance the monetization of our multi-channel marketing platform, as well as new advertising products, such as video advertising. Next, I will talk about progress on our AI-enabled new businesses. For Apollo, we made a strong progress in our core technologies and the platform elevation. Specifically, we launched Apollo 2.5 with much greater capabilities such as support for multiple vehicle types, including sedans, SUVs, trucks, buses and utility vehicles; support for high speed road conditions at much higher cost efficiency, enhanced development facilities and ApolloScape, a substantially more powerful open data set for training powerful models that are key to autonomous driving. The productization and commercialization of the Apollo ecosystem continues to accelerate with BYD joining us as the 100th partners last week. Even more importantly, as Robin emphasized earlier, safety is our highest priority, and we are working very hard to strengthen our development process to better safeguard vehicle safety. We are also working closely with multiple government departments and organizations to device a safety framework for autonomous driving in the future. For DuerOS, we're focused on key smart home usage scenarios and building deeper and more complete experiences that can truly gain market adoption and user engagement. For example, we made a strong progress for product experience for smart speakers with graphical display. This led to the very successful launch of the Ainemo Xiaodu @ Home product that Robin mentioned earlier. We are very focused on shipping more of such products and driving stronger usage as this type of products has genuine opportunities to be the primary control device for the future of smart homes. We also made a strong progress in platform capabilities, powering more and more devices from over 160 partners. Organizationally, we made the new Smart Living Group to better leverage our resources and drive stronger performance. Our ABC intelligent cloud business continued its rapid growth in Q1 as we build up digital expertise vertical by vertical. For example, in the financial services vertical, we just formed a strategic partnership with China UnionPay subsidiary, China UnionPay Merchant Services, which provides comprehensive payment services for millions of merchants in China. We will offer a full financial service solution to China UMS and their merchants, including ABC intelligent financial services cloud, China Merchants Bank, Ping An Life Insurance and other top-tier financial service providers that are also cooperating with Baidu Cloud. Overall, we are pleased with our progress, and we look forward to continuing our work in 2018. With that, I'll turn the call over to Herman to go through the financials.

[Cheng-Chun Yu:]

Thank you, Qi. Hello, everyone. Welcome to Baidu's first quarter 2018 call. As Robin mentioned, iQIYI successfully IPO-ed at the end of March raising approximately $2.37 billion. Also in March, Baidu raised $1.5 billion on debt offering [Technical Difficulty] (00

[Operator:]

Thank you very much. The question-and-answer session of this conference will start in a moment. Your first question today comes from the line of Eddie Leung from Merrill Lynch. Please go ahead.

[Eddie Leung:]

Hi. Good morning. Thank you for taking my question. I'm wondering if you could share a little bit more color with us on the performance of your news feed advertising product in the first quarter? And a follow-up on that, could you also comment on your thought on the pricing? For example, the eCPM of news feed advertisement in the industry, because it seems like, just in general for the industry, the eCPM went up quite a bit, especially, for example, for some of your competitors in news feed. So just wondering how much potential in general for the industry in this monetization front? Thanks.

[Qi Lu:]

Okay. Let me take this question. There's two parts. First is about our overall advertising performance in our feed part. First, as you all know, feed as a product experience is a natural extension of a search platform based advertising so that we can fully leverage our existing platform capability and our customer base. At the same times, our feed also open up growth opportunities to get into brand advertising, because the user engagement is very much moving up into the consideration funnels towards upper part of the funnel. So therefore, our product is able to enable our other driving product (00

[Cheng-Chun Yu:]

Yeah. Let me just add a few more points. Thanks, Qi. Yeah. So with regards to – I just want to clarify that Qi had talked about product ads and also [Technical Difficulty] (00

[Eddie Leung:]

Thank you.

[Operator:]

Your next question comes from the line of Alicia Yap from Citigroup. Please go ahead.

[Alicia Yap:]

Hi. Good morning, Robin, Dr. Lu, Herman and Sharon. Thanks for taking my questions and congrats on the solid results. Actually, my question is one or two follow up on a lot of the technology improvements. So can you elaborate a bit of the reinforcement learning for this Phoenix Nest online marketing system that you highlight on your prepared remarks? And what are the industry verticals that will be benefiting from that? And then, similarly for your OCPC, other than automobile, since – what other industry vertical could also be potentially benefit from these features the rest of this year? And if you also good quantify the potential upside of the revenues and monetization ability driven by all these technology enhancements that you have implemented over the past few months, if you can share some of that, it will be good.

[Qi Lu:]

Okay. Let me take that question. First is about the new infrastructures deployed in Q1 that's based on reinforcement learning. A little bit background. Machine learning, as we all know, is very, very powerful capabilities that drives a lot of our industry. Reinforcement learning is a particular kind. The key strength of reinforcement learning is that you do not – one is, you do not need the label data, meaning that you learn by the actual activity data, whether it's view or clicks or conversion data by itself, so the learning is a lot more efficient. And second, it is able to tap into more data signals, so that the ability to elevating economic yield and ad quality will be further improved. And third part is reinforcement learning is very much can be done in what we call online learning, meaning that you do not have to train the model offline then, move it to online. So the speed of development is also efficient. As an example, in our industry, as we all know, reinforcement learning is the underlying technology delivered (00

[Cheng-Chun Yu:]

Yeah, great. And let me just add specific examples of what just Qi just said, OCPC, as mentioned, this is optimized cost per click. And as you know, Baidu system is based on number of clicks. But for many industries, what advertisers want is something that is more certain. For example, if you are in the business of developing apps, you are a app developer, what you want to know is not how much the cost per click is, you want to know, for example, how much cost would it be for a user to look at your ads and be able to download your apps, right. And there are industries, for example, in autos and so forth that you want to know what's the cost per lead. What Baidu has done is, because we have a rich history of all these customers in the past, we can then run it through our huge AI computing power and be able to predict basically based on the type of ads you have, the kind of words you have, the kind of photos that you use and so forth and we help you optimize the content and so forth that we have – we're able to forecast if you spend this much on CPC, therefore, your result for CPA for these actions that you would like would be approximately this. That way it reduces the uncertainty of an advertiser and they will be willing to pay more because of that less uncertainty. So you can think of it that way. So for an industry that doesn't have a long history and we have a lot of data, we can help them optimize, and those are the potential that we have.

[Sharon Ng:]

Operator, next question.

[Operator:]

Certainly. The next question comes from the line of Grace Chen from Morgan Stanley. Please go ahead.

[Grace Chen:]

Thank you. Thank you for taking my questions and congratulations on your strong results. My question is about the progress of your autonomous cars. We understand that Baidu just celebrated one year anniversary for the Apollo project. We understand that development is still in the early stage, but in retrospect, what do you view as the key achievements in the past one year? And what would you do to do better? Looking into the future, what would be your key focus to stay ahead of the curve, given the intensified competition domestically? Thank you.

[Qi Lu:]

Yes. So first of all, with regard to the key progress over the last one year for the Apollo platform, there's several important area to quote. One is the platform capabilities. We deployed Apollo 1.0 within short span. Now, we just deployed Apollo 2.5. The platform capability and technology is moving at a rapid pace. Second is, the ecosystem helps. Our developer community, member partners, the different agencies, government departments that are embracing, supporting Apollo has been growing leaps and bounds. Third, which is more important, is the productization and the commercialization of our partners for L3 and L4-based vehicles is moving at a much, much faster pace than we originally anticipated, much, much faster. Now going forward, our key focus areas will be continue to build up our technology foundations, but more importantly, to speed up the productization and the commercialization of vehicles. For example, we mentioned in this year, we have a partner to commercially deploy L4 minibus vehicles. In next calendar years, we have several OEM partners deploy L3 commercially available vehicles. And as long as we keep excellent pace onto all those, the Apollo platform will continue to gain strength against the competition. As we mentioned in the earlier calls, the Apollo ecosystem also have a very clear and healthy business model. We are very similar to Android, the platform technology is open for participating, open for use, but we also sell our services, just like Google, if you use Android to build a handset, you use Google Search services. In this case, it would be our HD map services and the future services that we'll deploy. So overall, our focus is continue to speed up the development ecosystem, particularly the productization and the commercialization of our partners' vehicles. That will further increase Baidu's competitive strengths.

[Sharon Ng:]

Operator, next question please.

[Operator:]

Certainly. Your next question comes from the line of Gregory Zhao from Barclays. Please go ahead.

[Gregory Zhao:]

Hi, management. Thanks for taking my question and congratulations on the strong quarter. So my first question (00

[Cheng-Chun Yu:]

Okay. Hi, Greg. So let me try to answer your questions on revenues and on margins. Yeah, when Baidu – in our guidance, we build in and we factor in iQIYI, but I think we also consider just overall as a consolidated number. So we don't really separate what's iQIYI and then the rest is Baidu, just as an upright. But we expect actually, based on our guidance that our growth rate should be similar to what we saw in Q1 approximately on a year-over-year basis, okay. Your question was on media censorship. I think as Robin mentioned in his prepared remarks that the way we position our products, we believe is really trying to optimize the user experience over the long term life of the user on our platform. So having clean content rather than having very quick wins and using content that's not very appropriate is really not our style. We would focus on more wholesome content, we focus on more diversified content and leverage on our AI to actually promote content that's – what's most relevant and best for the user. So I think as a result, I think that – we think that we're better positioned for high-quality content. And with regards to what happened with Toutiao, I don't think that when you look at the size of our advertising dollar, our advertising market is focused head to head with Toutiao. I think what to look at is the current size of our sales and marketing market, which is very significant in China, and we then also look at our opportunity to actually go into key accounts and be able to get some of that brand budget as we see with our feed doing very well, with the content consumption on feed doing very well, gives us more inventory to actually do display and so forth. So that's where we see the opportunity with the SME market with, as Qi mentioned, a lot of the new products that we've done with advertising leveraging big data so that it becomes more effective for our customers and with the KA (00

[Qi Lu:]

So just to sum up, I think the strong revenue and guidance can be attributed to technology innovation, both on the user front and on the customer front and less from impact of the regulatory environment changes.

[Cheng-Chun Yu:]

And then, Gregory, on your question on margin. The way we see margin is, I think if you look at Baidu as a platform, last year, we were doing things such as O2O, on Nuomi, on Baidu Deliveries and so forth and we have other businesses that have since spun off. So last year was a transformation year. And through that, our strategy has been more clear, strengthening our mobile and then also strategy in AI and as we restrategize our businesses around AI-based platform, there's a lot of synergy for that, even though we're in search, we're in feed, we're in autonomous driving, we're – to activate Internet. So what you've seen in the first quarter since we have one lap around us (00

[Operator:]

Your next question today comes from the line of Piyush Mubayi from Goldman Sachs. Please go ahead.

[Piyush Mubayi:]

And thank you for taking my question. Dr. Qi, a question on autonomous vehicle for you. We understand that the commercial application for AV may still be in an early stage, but you mentioned commercial vehicles twice. And I wondered if you could just compare and contrast the passenger segment and the commercial segment and take us through which ones could come to market faster and if you could potentially talk through the monetization opportunities relative to the two businesses.

[Qi Lu:]

Okay. So let me comment on the three aspects of what you just mentioned. So first of all, just as a matter of clarification and to put the facts in our statement. When we talk about commercially available vehicles, that's on the passenger side, meaning these are the L4 vehicles. These are six-seaters minibuses that will be commercially available later part of this calendar year and as several OEM partners will be available in 2019 and some in 2020. These are sedans. These are L3 vehicles meaning that the passengers still need to sit in the car in front of the wheel, but the vehicle is able to perform increasing more and more autonomous driving capabilities. Those are commercially available vehicles for passengers, not for commercial, let's say, trucks or logistic vehicle. That's the – make sure that we are clear and state the facts. That's number one. Number two, on the commercial opportunities, the key thing I want to emphasize is that we need to think about the commercial deployment at the lowest speed, the more controlled environment, and that's the key. There are a lot more commercially viable opportunities, individual consumers or business customers are willing to pay for, but autonomous vehicle driving behaviors is at the lowest speed and more controlled environment. The six-seater bus is a prime example of that because commercially, there's already stream of incoming bookings for that vehicle, you can imagine deploying those minibuses in a industrial park, in a tourist regions, they are operating in the lowest speed in a very much controlled environment. Therefore, safety can be guaranteed, can be fully protected, and similar for our L3 vehicles for next year. And having said all of that, let me also talk about the relative areas of commercially viable opportunity for passenger vehicles and commercial vehicles. Indeed, in our inbound partnership requirements that we're working with more and more partners on commercial vehicle, for example, logistic vehicles, trucks and autos because there's even more opportunity to deploy those at the lowest speed and a more controlled environment. So that's the landscape that's evolving at the very fast pace, and Apollo is tend to gain substantial more opportunities. Lastly is about the business model. For Apollo, we are very clear upfront in our business models. We make the platform available open – free for our partner to use, but they would need to use a key set of services, and one of those is HD maps, which stands for high-definition maps. These are maps not meant for human eye balls to see. These are maps for the sensor of the devices that's in the car to see. And this is the key part of our economic model and we're seeing very healthy adoptions of our HD maps by our customers, current customers and the future customers.

[Operator:]

Your next question comes from the line of Juan Lin from 86Research. Please go ahead.

[Juan Lin:]

Hi. Good morning. This is Juan. Robin, Dr. Lu, Herman and Sharon, thank you for taking my questions and congratulations on the strong set of results. So my first question is on short video. I'm wondering if you could share with us the current progress of monetization for the short video content and what is the strategy in short video content production? And the second question is regarding Xiaodu @ Home. I wonder what is your target for Xiaodu @ Home and any new hardware products that are going to be launched later this year. And when do we expect DuerOS partnership with hardware manufacturers to start generating revenue? Thank you.

[Robin Yanhong Li:]

I'll answer the short video question and Qi will address the Xiaodu @ Home one. Short video is an important content form for the Baidu App. The percentage or the penetration of short video content on the Baidu feed has been growing very, very quickly from the teens to almost half of our total distribution, so we're seeing strong demand on that. And in terms of monetization, we also see quickly improving monetization capabilities for short form video. And we believe this is going to be a very profitable business going forward.

[Qi Lu:]

And let me answer the second part of the question. First of all, just as kind of the strategic backdrop, as Herman and Robin made, the conversational Internet or voice-active Internet [voice-activated Internet] or conversational AI platform has tremendous and larger economic opportunity. For Baidu, we have a portfolio of investment on devices that can become hit devices that elevates the amount of usage and the user engagement enable Baidu to build up our DuerOS platform. So, among all those devices, we have our first-party devices, which – we are the principals of building those devices, and third party devices are partner-built. And this particular device, Xiaodu @ Home, typically called a second-party devices. They are a strategic partner, meaning that they collaborate with us, our teams, very, very closely. The brand experience, the operation of this product, the design engineering is very much done together by the two teams. And we're super pleased with the momentum this device is able to generate. At the same times, we fully expect, as part of – the remaining 2018, the DuerOS platform, we will roll out more devices across several categories. First is smart speakers with a graphical display. We believe this has the highly likelihood of becoming the future control devices for the smart home environment because there is the video screens. This product is a lot more programmable and operatable. In addition to music, simple instructions, people are increasingly using for content consumptions, for communication among family members and including to verticals for learnings, for telemedicines, there's tremendous growth potentials for that device. So we are planning to support more third parties and the future first-party devices in that category. Having said that, there is also multiple devices we roll out. For example, speakers without a screen is also a important part of the mix because imagine a home where you have multiple room, cover larger space, you need those type of speakers who can respond to voice commands providing useful services back. In addition, we are fully expecting developing additional first-party hardware product and third-party hardware product. But the intent is to be super choosy and selective in those products. Each of those products will have the potential to become a hit device in terms of number of device volume that we can ship, but more importantly, the usage amount, the engagement that we can generate on those devices. Lastly, let me talk about the general economics. This is a broad platform. The economics can be divied up into several segments. First, give device, give Baidu the opportunity to have future entry points for information, for content, for services, and advertising is the natural affinity. As a matter of fact, we already have ongoing early-stage advertising operations with partners for our DuerOS platforms. That's one. The second is, the DuerOS is also a platform. Platform has economics such as licensing economics in the long term and are taking a bid, just like iOS will take a bid on any services sold to the platform, but these are very long term. Short term, we're very focused on building hit devices generating great experiences for users and usages. And focusing on little bit of short-term advertising because they are very natural and we do operate some advertising products with our partners as of today for our DuerOS platforms. The emphasize is really long term we can grow into substantial economics, but it will take time.

[Operator:]

Your next question comes from the line of Thomas Chong from Credit Suisse. Please go ahead.

[Thomas Chong:]

Hi. Thanks management for taking my questions. I have a question on the competition. Can management give us some color how we see the intensifying competition in conversational platform as well as autonomous driving? Should we expect us to step up the investment on this front because of competition? Thank you.

[Qi Lu:]

Yeah. So let me first talk about the competition on the autonomous vehicle front. Within China market, which is our key focus areas based on the – everything we have been seeing from the industry landscape, we believe Baidu is in a very strong position to drive the autonomous vehicle productization and the commercialization and the entire ecosystem forward. There's a few important reason behind that – we are enjoying a competitive edge. First is, Baidu as a company, we invest in the autonomous driving technology much earlier than most of the others. We started working on this more than four years ago. As a result, our technological capabilities is stronger. The second is, we took a very important position that stand for open access to technology, building a collaborative ecosystem. As a result, we have been very broadly embraced pretty much by all OEM partners, tier 1 partners. And also, as a result of that, we are also being very much supported by various government agencies and departments. For example, the Chinese government designated Apollo as one of the four national open platforms. Particularly, Apollo is the platform of choice for autonomous driving. So, overall, in combination, we very much see Baidu is in a strong competitive position in the space of autonomous vehicle. Relative to conversational AI, the competition is more intense as you observed. Having said that, Baidu is in a strong position for a number of important reasons

[Thomas Chong:]

Herman, do you want to address the margin question?

[Cheng-Chun Yu:]

Yeah. I think we talked about the margins a little bit earlier. So with our margin – as you saw in Q1, our margin improved significantly from the year before, but part of it is the reduction in marketing costs with regards to our O2O. And we also increased the spending for our channels and so forth. And I think going into our Q2, our margins – I think that we have the capability of having higher margins, but we want to spend opportunistically with regards to TAC and also with regards to the marketing dollars. So I think depending on what opportunity there would be, our margins may go down, but it all depends on the market situation.

[Operator:]

Your next question comes from the line of Han-Joon Kim from Deutsche Bank. Please go ahead.

[Han-Joon Kim:]

Great. Thanks for the opportunity to ask a question. The question revolves around kind of the pace of commercialization of autonomous driving versus kind of the voice-enabled speakers and so forth. So monetization of the HD maps on autonomous driving is pretty clear. What will be the comparable kind of element of how we monetize voice interactions or voice modality? And between the pace of the two, autonomous driving and voice search, which kind of do you think is tracking faster?

[Cheng-Chun Yu:]

Yeah. Let me address two parts of the question. First of all, with regard to the commercialization of autonomous driving and relative clarity of our business model, based on everything we are seeing, pretty much every week when we talk to our OEM customers, our tier 1 customers, we see more commercially viable scenarios that our technology can be deployed. So, overall, it is really accelerating. For example, this week there is Beijing auto shows. As a matter of fact, we just discovered new commercially viable opportunities just by talking to the partners during auto show exhibitions. So – and that is – it is really accelerating. And with regard to the Baidu economic opportunity, the HD maps itself is very long term, I emphasize long term, is very meaningful by itself. Having said that, we fully expect to tap into future economic opportunities for Baidu. There's more and more commercially viable opportunities as part of the Apollo platform Baidu will tap into. For example, we publicly talked about building something called ACU, which stands for Apollo Computing Unit. You can imagine that. Essentially, we are doing a lot of R&D. It's a hardware box that you can plug a sensor into it, you can plug silicon into it and silicon-wise, you can use NVIDIA, Intel or Baidu has its own silicon solutions through FPGA. Another product will be ramped up very quickly. That can grow into another future economic opportunities. And there's a number of different services. As part of the announcement of Apollo 2.5, we launched a new set of cyber-security services initiatives. That itself – because of the future vehicles, there's a lot of security requirements. These security services can also be virtually monetized over a period of time. So that's the Apollo side of our economic opportunities long term. On the conversational AI side, let me just recalibrate what I mentioned earlier. It's very, very important to talk through the economic opportunities and the relative speed of deployment. The speed of deployment – let me first talk about the speed of deployment. Based on the industry structure, the speed of the deployment of that conversational AI platform really depend on the hit devices. In the United States, it's because Echo, Amazon pushed it into hit devices and then that platform started to take off. In China, it is going to go through a similar structural evolution. And we very much see Baidu is in the frontier of pushing that hit device, which we believe will be a smart speaker with graphical display because the China home environment is substantially different than the United States. So that's number one. And we will be able to on the leading front. In that case, there will be – long term, there will be hardware economics for Baidu to tap into that's naturally. And then once the hit devices start to take off, the platform will be anchored. The platform will be used on more and more devices. On the platform side, first, these platform give Baidu new entry points, and directly these entry points will be able to create advertising revenue opportunities as I mentioned earlier. Right now, we are indeed operating with a few advertising partners as part of a deployment for our DuerOS devices, but I want to emphasize, we are very early – just a few partners we're working with. And then second, outside of the entry points, advertising revenues. The platform itself has economics attached to it. The typical form – platform will be licensing revenues because the device will be richer and richer. We envision right now with speakers, can be many different form of devices. The platform will have licensing economics. And the platform will have VICs (01

[Operator:]

Your next question today comes from the line of Karen Chan from Jefferies. Please go ahead.

[Karen Chan:]

Thank you management for taking my question, and congratulations on the strong set of results. So just a housekeeping question, sorry if I have missed it previously. Just wondering what's the revenue contribution from the news feed during the quarter. And how are we thinking about the growth trajectory going for the rest of this year? Are we still maintaining a 20% sequential growth target? Thanks a lot.

[Cheng-Chun Yu:]

Yeah. Hi, Karen. As I mentioned earlier, feed this quarter was slightly up on a sequential basis because from Q4 to Q1, seasonally in China, advertising usually has dropped. So the fact that we can actually counter the drop and have a sequential quarter I think just shows the strength of our feed. And as I mentioned also on the last call that from Q4 to Q1 you should expect that. But Q1 to Q2, you probably have a seasonality. So we expect that will be up higher on a sequential basis in Q2 versus Q1. Thank you.

[Cheng-Chun Yu:]

And that concludes our call today. Thank you, everyone, for dialing in. We'll join you guys next quarter.

[Operator:]

We are now approaching the end of the conference call. Thank you for your participation in today's conference. You may now disconnect. Good day.