Herman Yu - CFO Sharon Ng - Director of IR Robin Li - Co-Founder, Chairman, & CEO

Alex C. Yao - JP Morgan Chase & Co Alicia Yap - Citigroup Inc Eddie Leung - BofA Merrill Lynch Grace Chen - Morgan Stanley Han Joon Kim - Deutsche Bank AG Juan Lin - 86Research Limited Piyush Mubayi - Goldman Sachs Group Inc. Tian Hou - T.H. Capital, LLC Wendy Huang - Macquarie Research Thomas Chong - Crédit Suisse AG Jerry Liu - UBS Investment Bank

Hello, and thank you for standing by for Baidu's Second Quarter 2018 Earnings Conference Call. (Operator Instructions) 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 your host for today's conference, Sharon Ng, Baidu's Director of Investor Relations. Thank you, and please go ahead.

Hello, everyone, and welcome to Baidu's Second 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; and Herman Yu, Baidu's Chief Financial Officer. After our prepared remarks, we will hold a Q&A session. Please note 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 the unaudited non-GAAP measure 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 also be available on Baidu's IR website. I will now turn the call over to our CEO, Robin.

Hello, everybody, and thank you for joining our call today. We had another strong quarter in Q2, with our core business exhibiting robust revenue growth, driven by AI-powered monetization capabilities and Baidu feed continuing its strong traffic and monetization momentum. Our AI businesses are also gaining strong traction. Earlier this month, we held Baidu Create, our annual AI developer conference in Beijing, which drew over 7,300 developers and partners from around the world. Our goal for Baidu Create is to bring developers together to share ideas on how they can leverage Baidu's AI to grow their business. I'll further elaborate on our announcements at Baidu Create and how we use AI to accelerate the growth of our businesses as we review our Q2 results. Let's begin the quarter with search and feed. We continue to push the boundary of the search experience by fulfilling 37% of search queries with TOP 1 results, and 38% through Bear Paw accounts, and over 1/6 of our search page views returning high quality videos. At Baidu Create, we announced the launch of Baidu's smart mini program to take advantage of emerging trend in China, where apps with lower frequency are connecting to super apps to bypass the ever rising cost of app preinstalls. Baidu's smart mini program is unique. In that, it allows app developers to quickly convert another platform’s mini program into Baidu’s. app developers have free access to Baidu's AI capabilities, and once Baidu's open-sourced mini program is set up, the app developer can also plug into any third-party super app in Baidu's network. Baidu's smart mini program offers a seamless native-app experience to other apps, which will extend Baidu's long-tail content reach and derive better conversion. In June, average daily active users of Baidu App reached 148 million, up 17% year over year. Total user time spent on Baidu App grew approximately 30% in June year over year. Feed consumption continued to trend in a healthy manner, with video consumption up 270% year-over-year. We added over 200 MCNs in the second quarter. For example, People's Daily, a major state-owned news platform in China, joined our content network Baijiahao, making over 2,000 media sources on its platform, representing over 38,000 Baijiahao (BJH) accounts, available to Baidu feed users. We continue to leverage Baidu's strategic advantage to power our feed algorithms with Baidu's AI and massive long-tail user insight to provide users with more interesting and wider range of content. On the monetization front, Baidu's AI has also been the strategic thrust behind our ad innovation and ad effectiveness, from augmented reality to video ads to performance-based ads. Dynamic Ads has been a strong revenue driver, incorporating product and user data from our customers. This quarter, we expanded Dynamic Ads to advertisers in the education and financial services industry beyond e-commerce, travel, auto and real estate. Optimized cost per click, or oCPC, continue to be an important component of revenue growth, as we expand beyond game developers to financial services, real estate and education. For mobile action ads, we expanded beyond enabling users to call and chat with the advertisers directly from the ads on Baidu, to enabling the users to take advantage of an instant coupon or click-to-buy. With strong intent data, Baidu is able to continuously provide our customers with these useful, interest-generation and purchase features, which have generated meaningful lift in click-through-rate. We are also seeing good results from Moonrise, Baidu's deployment of deep reinforcement learning to improve ad performance. Using deep reinforcement learning, which is rewards-based, linking actions to desired results, Moonrise can suggest better keywords, photos and videos from Baidu's huge content library to advertisers and increase conversion and overall ads effectiveness. Baidu is an early adopter of AI. Together with massive user intent data and interest graphs, we see great opportunities to help advertisers leverage Baidu's large traffic, open-platform advantages and highly-effective performance-based ads. Turning to DuerOS, our IoT operating system, DuerOS, is expanding quite rapidly. In June, DuerOS installed base reached 90 million, nearly doubled in the last 6 months. Voice queries on DuerOS in June surpassed 400 million times, again doubling over a 3-month period. To drive the adoption of DuerOS, we offer third-party tool kits as well as first-party and second-party smart devices. In June, we launched Xiaodu Smart Speaker, which sold out 10,000 units within the first 90 seconds went on sale online. Xiaodu Video Smart Speaker, which was launched in April with AINemo, continues to sell well and has garnered overwhelmingly positive user reviews. We are seeing toddlers as young as 4-year-olds and senior citizens becoming an important customer base, reflecting the ease of use of conversational AI powered by DuerOS. In addition to retail, DuerOS is partnering with Intercontinental Hotels Group to bring smart rooms to China, where hotel customers can control their rooms and learn about hotel information through DuerOS conversational AI. Aside from the hospitality industry, DuerOS is also making inroads in the auto industry, forming partnership with the likes of BMW Group, Daimler AG, Ford Motor, Hyundai, KIA, Chery, BAIC and the FAW Group and Byton, to allow car owners to activate AI input modalities, such as voice, image and facial recognition and enable closed loop services, including ordering movie tickets and booking hotels. Incidentally, we are seeing fast adoption of voice input in China. Baidu Mobile Input saw average daily voice input grow almost 150% year-over-year in June and averaged 335 million times per day in the last week of July. Just like computers and smartphones, software will be a key determining factor for user experience of IoT devices, which means users will want better AI input modality, better search technology and better content ecosystem. On Apollo, we are moving at China Speed. Earlier this month, we worked with King Long Motors to launch the first fully autonomous Level 4 minibus, Apolong, a shuttle bus built without steering wheel. Neolix Technology also launched an L4 vehicle powered by Apollo – a mini-cart for cargo deliveries this month. We are excited about the commercialization of Apollo-powered vehicles, which is a testament to Apollo's comprehensive ecosystem, from AI technologies to OEM software-to-hardware integration, to Tier 1 parts suppliers to IoV-enabled backend infrastructure. China is the largest auto market in the world, with 28.9 million new vehicles sold last year, presenting a huge opportunity for Apollo to make driving easier and help provide environmental and traffic solutions. In March last year, I challenged our team to open source Baidu's autonomous driving platform and release a commercially viable vehicle. Within 16 months, our team worked with OEM partners to release two Level 4 vehicles off the assembly line, which is not an easy feat in the automotive industry, where manufacturing delays are common. Recently, I post another tough challenge to our team, help Baidu users identify quality, reliable natural services. In China, there are currently 31,000 licensed hospitals, but no effective mechanism to identify those with unscrupulous services. In the past, Baidu focused on banning bad ads. As a major information source in China, I believe, we should use AI and step up efforts to significantly improve the quality of medical content and weed out bad hospitals, even if they are licensed and operate within the realm of law. Having access to reliable medical service is a basic necessity to having a quality of life. Within a year, I challenge my team to come up with a solution to improve trust to online medical information on Baidu properties. Turning to iQIYI. iQIYI continues to perform well with strong growth in membership and advertising revenue in Q2. Total subscribing members reached 67 million in Q2, adding a record-high of 29 million subscribing members in the past year. Forming a large membership base lays a solid foundation for iQIYIto promote its self-produced content. We see iQIYI’s award-winning content production, coupled with Baidu's huge traffic and technology as a powerful combination in online entertainment. Our shared synergies provide iQIYI with a unique advantage, particularly as the fast-growing OTT industry in China continues to scale and conversational AI receives broader adoption into Chinese households. In summary, we are pleased with the momentum that we have built, moving forward an AI-first company. We look forward to further updating you on our progress next quarter. With that, let me turn the call over to Herman to go through the financial highlights.

Thank you, Robin. Hello, everyone. Welcome to Baidu's Second Quarter 2018 Call. Before I begin Q2's review, let me make a few notes. All monetary amounts in my discussion are in RMB, unless stated otherwise. Turning on January 1, we adopted ASC 606, a new revenue accounting standard that nets value-added tax from the revenue and cost of revenue lines. To increase comparability with 2018 numbers, 2017 revenue numbers have been adjusted net of value-added tax. We completed the spinoff of our Global DU business, in addition to Baidu Deliveries and Baidu Games, and are in the process of spinning off Du Xiaoman, or DXM, our financial services business. For today's discussion on Baidu Core, we have excluded the spun-off and to-be-spin-off businesses, including Du Xiaoman to give more visibility on the performance of Baidu Core, not including the spun off in planned-to-spin-off businesses. With that, let's turn to Q2. We had another solid quarter in the second quarter. Total revenues reached RMB 26.0 billion, up 32% year-over-year and non-GAAP operating income reached RMB 6.5 billion, up 31% year-over-year. Revenue from Baidu Core, excluding spin offs reached RMB 19.0 billion, up 30% year over year and non-GAAP operating profit of Baidu Core, excluding spin offs, reached RMB 7.7 billion with non-GAAP operating margin reaching 41%.In the second quarter, our sales and marketing expenses came in under budget. We're targeting non-GAAP operating margin in the third quarter to be several points lower than in the second quarter, assuming we're able to acquire the traffic acquisition and channel pre-installs as planned. Revenue from announced spin offs was slightly above RMB 1.0 billion in the second quarter. We expect this portion of the revenue to decline in the future, upon the completion of each divestiture. For example, in July, we completed divestiture of Global DU, and as a result, we no longer consolidate its revenue. We're providing this level of revenue detail to give investors more visibility to gauge Baidu's revenues post the completion of the expected divestitures. Our strategy to transform Baidu into an AI-first company to drive synergy and accelerate long-term growth, while redirecting management attention and resources from non-core businesses to new businesses, is seeing good traction. In the second quarter, feed plus AI businesses together made up almost 20% of Baidu Core revenue excluding spin offs. Together feed plus AI businesses grew over 150% year-over-year, AI is powering search with better user experience and continuously increasing Baidu's ad effectiveness. AI is also powering the growth of feed and new AI businesses, which, we believe, will shift Baidu's revenue structure proportionally to higher-growth areas over time. By divesting non-core businesses, we expect the announced spin offs together will generate approximately USD 1.8 billion in cash, which we plan to repurpose for better return on capital. In June, we announced a share repurchase program for up to USD 1 billion over the next 12 months. Program-to-date, we have given back to our shareholders approximately USD 187 million. Let me now go through the rest of the financial highlights. Online marketing revenues were RMB 21.1 billion, up 25% year-over-year. We had approximately 511,000 online marketing customers, up 9% year-over-year, and revenue per customer was approximately RMB 41,200, up 16% year over year. Other revenues were RMB 4.9 billion, up 75% year-over-year, mainly resulting from the robust growth in iQIYI membership and other businesses. Revenue from iQIYI in the second quarter reached RMB 6.2 billion, up 51% year-over-year. Cost of sales, excluding stock compensation, in the second quarter was RMB 12.0 billion, up 28% year-over-year. Content cost in the second quarter was up 68% year-over-year to RMB 5.2 billion, mainly due to the increased content purchases by iQIYI, and for Baijiahao, Baidu's feed content network. SG&A expenses, excluding stock compensation in the second quarter were RMB 4.2 billion, up 56% year-over-year, mainly due to the increase in channel and promotional marketing, partly offset by the cutback in Baidu Deliveries and O2O promotions. R&D expenses, excluding stock compensation, in the second quarter were RMB 3.3 billion, up 24% year-over-year, mainly due to an increase in personnel-related costs. Operating income in the second quarter was RMB 5.4 billion, up 29% year over year. Non-GAAP operating income was RMB 6.5 billion, up 31% year-over-year. Non-GAAP operating income of Baidu Core, excluding spin offs, was RMB 7.7 billion and non-GAAP operating margin for Baidu Core, excluding spin offs, was 41%. Income tax expense in the second quarter was RMB 1.1 billion compared to RMB 0.6 billion last year. Effective tax rate was 18% compared to 11% last year, which benefited from the disposal of certain subsidiaries. Net income attributable to Baidu in the second quarter was RMB 6.4 billion, and diluted EPS was RMB 18. Non-GAAP net income attributable to Baidu was RMB 7.4 billion, up 57% year-over-year, and non-GAAP diluted EPS was RMB 21. Adjusted EBITDA in the second quarter reached RMB 7.4 billion and adjusted EBITDA margin reached 29%. Adjusted EBITDA for Baidu Core reached RMB 8.6 billion, and adjusted EBITDA margin was 43%. As of June 30, 2018, cash and short-term investments was RMB 128.3 billion. For the second quarter, cash flow from operating activities was RMB 7.1 billion, and capital expenditures were RMB 1.5 billion. Total headcount as of June 30 was approximately 39,700, down 6% year-over-year. As of June 30, 2018, excluding iQIYI, cash and short-term investments was RMB 115.2 billion. For the second quarter, cash flow from operating activities was RMB 7.0 billion, and capital expenditures were RMB 1.3 billion. Excluding iQIYI, total headcount as of June 30 was approximately 32,900, down 9% year-over-year. Turning to third quarter guidance, we expect total revenues to be between RMB 27.37 billion and RMB 28.77 billion, representing a 23% to 30% increase year-over-year. Excluding the impact from announced divestitures, including Global DU and Du Xiaoman, Baidu expects revenues to be between RMB 26.56 billion and RMB 27.92 billion, representing a 26% to 33% increase year-over-year. This forecast is our current and preliminary view and is subject to change. I will now open the call to questions. Operator, please?

(Operator Instructions) Your first question comes from the line of Eddie Leung of Merrill Lynch.

Quickly, just wonder if you could share more color with us on the programs you have made on news feed, in terms of both usage and monetization? And how does that compare with our competitors? And then just a quick follow-up on your guidance. We have known some scandals related to P2P financing and that we’ve seen recently. Have we seen any impact on our revenue in the third quarter from these sorts of events?

Yes. Eddie, on the news feed progress, as I mentioned during the prepared remarks, things have been growing very, very quickly. And we have seen growth both in terms of DAUs as well as time spent per user. And the relevancy of those recommended news feed content keep improving. We believe we have the best user experience now. And comparing to most of the other super apps, I think, we are gaining share. But our news feed is designed to enrich peoples’ lives, not really just help them to kill time. So we want our users to learn something valuable anytime they come to the Baidu App. And that's why we think news feed is actually a part of our core business, and it's an integral part of the search plus feed super app. And we do expect this kind of trend will continue. And we think our news feed already reaches the largest audience among all the comparable products in China. With more than 600 million monthly users for our search product, we still have a lot of room to grow in terms of the news feed product. And on the P2P advertisers, we always had a very tight control and high bar for P2P advertisers. We haven't seen any meaningful impact on our revenue as of today, and probably going forward, we wouldn't see any meaningful impact either.

The next question comes from the line of Alicia Yap of Citigroup.

I do have some follow-ups on the reasons operating environments for the overall news feed and advertising content censorship. Have we actually seen more step up by the regulators on paying more closely attentions to some of these ad contents format and the content that carry the ads? Is the change of this regulatory environment temporary? Or should we expect more normalized ad growth for the industry after that? Specifically, any impact for Baidu family on this step up? And then just on the housekeeping question, for the SG&A, there's quite a bit of increase this quarter. Are there any onetime cost associated with the business divestiture? Or is it mainly from the promotional channel cost increase?

We always hold a very high standard on our content for news feed. We do not want to give users something that's illegal or something that's on the borderline. Again, we want to enrich our users life, and we want our users learn something valuable. So we haven't seen any negative impact over the past couple of quarters because our standards have always been higher.

So on the SG&A, the majority of the growth year-over-year has been related to promotional activities and also channel costs. There were over RMB 100 million related to a certain bad debt that we had to reserve for, but I think that will be more of a one-off than a continuing item.

Your next question comes from the line of Grace Chen of Morgan Stanley.

My question is about -- in the recent Baidu Create conference, Baidu announced introduction of smart mini program. Can you provide a bit more color about the progress of implementation, feedback from customers, any performance metrics that you can share with us? Such that which verticals have adopted the smart mini program first and what other verticals may present bigger potential? And also, can you elaborate a bit more about how the smart mini program would fit in your overall strategy for search or feed business? What would be the sustainable long-term growth rate of social feed business? And lastly, it will be great if you could give us some guidance in terms of the margins in the second half year, in particular, some guidance for the variance cost items, such as traffic acquisition costs, content costs and R&D?

Grace, is that right? Your first question was on smart mini program and the progress there?

Yes, the first question is smart mini program and in terms of the implementation right now, your feedbacks, and in terms of which verticals they have adopted the smart mini program? And second question is about the guidance for the margins in the second half, especially, in terms of variance cost items such as traffic acquisition costs, content costs and R&D?

Okay. On the smart mini program, we received overwhelmingly positive responses from the partners we have contacted with. And our latest version of Baidu App already supports the mini program. It will give users much better experience and well drive higher conversion in the future. Right now, some of the verticals are already ready, including games and travel-related verticals. And we are implementing other verticals to well continue to evolve very quickly. We have promised to open source this by the end of this year, which means that other large apps can just plug into the API and be ready to support all kinds of mini programs in the same ecosystem. So we're seeing a very high level of interest, and we're just implementing all kinds of features. We're already seeing a number of exemplary partners online, for example, train tickets, the conversion rate is much higher than before.

Grace, on your question on second half margin, as I mentioned in the prepared remarks, we think that we should see our margin dip several points from second quarter. We've been talking about this since the beginning of the year that we intend to increase our investments in traffic acquisition. I think that's probably the main line item. In addition to that, we'll also continue to increase our investment in content acquisition, so that our feed can continue to be able to acquire traffic and push long tail content. So I think those are the main items that could potentially increase our budget, sales and marketing and also cost of revenues in the content line.

Your next question comes from the line of Thomas Chong of Crédit Suisse.

I've got questions about AI initiatives in DuerOS and Apollo. Can management give us any additional color in terms of the monetization time line? And when should we expect revenue contribution in the coming quarters? And my other question is about on the regulatory front, should we expect there will be more regulations to come up in the search or on the news feed side?

On the AI initiatives, both DuerOS and Apollo are growing very, very quickly. We're seeing tremendous growth or adoption rate from the users who use our IoT devices, and we see a lot of OEMs’ strong interest to work with us. But this 2 business lines are in its very early stage, so we do not exact any meaningful revenue contribution in the coming quarters. But having said that, I would say that these 2 business lines are the most advanced ones among its peers. If anyone starts to be able to generate meaningful revenue, we will be the first one to achieve that goal. Especially for Apollo, I think we enjoy an advantage that the Chinese government is very supportive of this kind of technological innovation and are willing to cooperate on the infrastructure level, which means that they want to install sensors on the roads, so that cars do not need to buy very expensive LiDAR or other type of sensors in order to drive autonomously. So we are very excited about the future of these new initiatives. And on the regulatory environment, for both search and news feeds, we have been in existence in the search business for a very long time, and we're seeing changes or fluctuations from time to time. And we think we are able to cope with that going forward as usual. And news feed, we've also been operating this for a little more than 2 years. And we are able to maintain a high standard and close dialogue with the regulator. And we do not see anything changing significantly going forward.

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

My first question is on short video. I wonder whether there is a crackdown against short video apps that are not in compliance with content censorship has put some weight on the short video supply of our platform and whether that will impact our short video strategy going forward? Also, I wonder what is the current revenue contribution and the monetization progress for our video content? How does video ads compare to news feed ads in terms of the level of monetization? Second question is on TAC. You mentioned that company will step up spending on traffic acquisition for the second half of the year, given that traffic acquisition prices remain pretty high, I wonder whether we're going to adjust our OEM traffic acquisition strategy and the measurement of ROI for traffic acquisition going forward?

All right, Juan, I'll answer your short video question and Herman will answer the TAC question. Short video is growing very, very quickly. We're new to this business, but we have seen tremendous growth over the past few quarters. And video content represents well over half of our total news feed content distribution volume. And short video is also very positive to our revenue generation, which means it actually makes money from day 1 if you look at the monetization capability angle. And in terms of regulations regarding to short video, I think, it's the same standard with text and images. And we have a sophisticated system to identify unwanted content, and we're able to use AI, use machine learning to really filter out those unwanted content and send the most relevant content to our users.

The question, Juan, on TAC, I think in terms of overall traffic acquisition cost, the way we look at it is, on TAC, in the past, we look at trying to get absolute margins -- gross margin from TAC purchases, and we've changed that strategy since the beginning of the year. We believe that maximizing profit rather than maximizing margin will be a better strategy, and we've been doing that since the beginning of the year. So to the extent that we think that there is incremental dollars to be had, we'll continue to acquire TAC. With that said, we're also looking at the ROIs coming from TAC versus app downloads and trying to grow Baidu App. And so far, we think that we're able to actually increase Baidu App spending, app installs, faster than TAC because we see that ROI has been more positive. So I think in the last 2 quarters, you're seeing us trying to increase our investments in traffic acquisition. TAC we have not increased too significantly. However, under sales and marketing channel cost, you've seen that increase significantly because we've found ways to get pretty good ROI by doing such marketing investments.

Your next question comes from the line of Jerry Liu of UBS.

My question is on advertising. Wanted to ask about revenue growth in the next few quarters. If we're to maintain roughly our current revenue growth rate, do we see a lot of legs left in dynamic ads and oCPC and some of the initiatives today? And when can we see mini programs start to kick in?

Yes. As you know, the Internet population in China is not growing as quickly as before, so I think the total pie is not getting much larger. But that really means the technology will play a much more important role, both in terms of user experience and in terms of monetization. And regarding to ads, we do see a lot of room in terms of better monetization, technology, including oCPC and including mini programs. Once the advertisers adopt the mini programs, they should see a much better conversion because the user experience is more like a native app instead of just a web page. And user data are connected and the relevancy should be improved. So going forward, over the next few quarters, I think a big revenue driver will be technology innovation on the monetization front.

The next question is from Han Joon Kim of Deutsche Bank.

And just had quick questions. I've noticed that the revenue mix from mobile actually decreased for the first time. So I just wanted to get your perspective on how you see the PC and the mobile mix shifting? And if this is still a one-off or some signs of change in the market? And as a follow-up to your question or topic about #1 kind of results having a better impact, does that mean that the #1 keyword basically gets better pricing? And so it dilutes the impact of #2 and #5? How does the blended eCPM rate kind of shifts when you're #1 results becomes much more dominant result out of all the results?

: Yes. Han Joon, I think you're referring to our mobile revenue. When you're looking year-over-year, our mobile revenue, I think, actually grew as a percentage of total. I think it's just a seasonal fluctuation between Q1 and Q2. And that change was 1%, which I don't think is material. I don't think I kind of read through a 1% variation quarter-over-quarter is too meaningful. I think, overall, we're still seeing a faster growth from mobile side. And I'll have Robin answer the Top 1 results from our search.

When we think about how to develop our products, we always consider the users’ needs first. So we want to try to meet users' needs as quickly as possible, providing the right answer as the #1 result achieves that. And in the meantime, I don't think it will hurt monetization significantly, because in a lot of cases such kind of #1 result has commercial values too, and users are well served in the meantime and we can make some money out of it. For example, when people are looking for airline tickets, the first results get very well satisfying users' needs without them landing on the website, and in the meantime, it has got clear commercial value. So when it comes to the ranking of search results, we always think of the users' interest first, and we think in the long run, users will come back more often and contribute more time to our app, and eventually, we will also be able to monetize the overall user base better.

The next question is from Wendy Huang of Macquarie.

I just have 3 housekeeping questions. The first is to follow up on the PC mobile revenue mix. So on the flip side, given that the PC dynamics exposure actually increased, does that imply the PC revenue growth, actually in the quarter, is more than 30%? And also, what's CPC trend for both PC and mobile behind that? And secondly, you disclosed about the Baidu App DAU and user time spent year-over-year trend. I just wonder if you can also share a sequential trend or maybe the Q1 data for the DAU and time spent? Lastly, on the spinoff business, if my calculation is correct, I think the revenue of the spinoff businesses Internet financing and Global DU is about RMB 800 million to RMB 850 million. So what is actually the margin of this spinoff business? And how shall we take in to consideration of this spinoff impact on the margin side in the Q3 as well?

Let me answer your first question on the mobile revenue. So mobile revenue as a percentage of total was 76.6%. So we were 1% less than last quarter. But when you compared to last year, we were 72%. So when you look at the last 4 quarters and so forth, mobile revenue was still growing. So as I mentioned at the question before, I don't think a 1% variation is something to worry about. I think, overall, we're seeing, if you look at the last 4 quarters, mobile revenue is actually trending up because fortunately traffic from mobile is also higher.

On the DAU and time spent metric, I think, the trend for Q1 and Q2 was very consistent. As you know, Q1 is seasonally a weak quarter. So in Q2, if we compare to Q1, the sequential growth is very significant. So it's better to compare Q1 with the year-over-year numbers and Q2 also year-over-year numbers. This numbers, we have said in the last earnings call that both are growing. And comparing to most of the super apps, we are growing faster than most of them.

And then the last question with regards to spinoff revenues and margins, I think that question is going to be hard to answer. Really depends on whether we're able to close the deal during the quarter or at the end of the quarter. We presented in our prepared remarks the amount excluding the spinoffs. So the difference you can see what we're estimating. For the announced spinoffs in Q3, we're anticipating approximately may be RMB 800 million from the spinoff, that's assuming full quarter of our Du Xiaoman financial services. But should we close this deal beforehand, then obviously, we would only have a portion. With regards to the profitability of the spinoff as a whole, we're seeing profitability, but that is pretty minor compared to all of Baidu. So I think after the spinoff, we would probably lose the revenue between RMB 800 million to RMB 1 billion a quarter. And then with regards to profitability, we'll lose some profitability, but it will not be too significant.

Our next question comes from the line of Alex Yao of JPMorgan. Alex C. Yao: I have a question on the core ads growth outlook. If my math is right, I think, the core ads growth rate for third quarter revenue guidance is around high-teens to mid-20s. Can you give us a sense of what's driving this core as growth rate, i.e. pay click growth versus monetization growth? And then is it fair to say majority of the pay clicks growth is driven by the newly added inventory from feeds? And secondly, is Du Xiaoman the last and non-core asset that we're going to spinoff?

I'll take that, Alex. When you say core business, I guess, you're meaning Search. Is that correct? Alex C. Yao: Yes, I mean taking out the video -- the iQIYI business everything else left over.

Okay. So our Search plus feed, I think, overall, is growing pretty significantly. We don't really break out in terms of how much of that is traffic, how much of that is clicks. I think it is a mixture. When you're looking at Search in itself, obviously, you're seeing the effectiveness of using different type of AI technologies, coupled with our large pool of data to continue to increase the efficiency of Search, that's helping drive our revenue. And then with feed, I think, you're seeing both. You're seeing not only additional traffic, you're seeing increasing user time spent. You're seeing us introducing some of the new search products, monetization product into feed. You're also seeing different formats, for example, like AR that's helping drive some of the display ads that we didn't have before. So there are several dimensions that are driving feeds. With regards to Du Xiaoman -- what was the question, again? Alex C. Yao: You guys have been spinning off or disposing non-core assets for quite a few quarters. Is Du Xiaoman the last non-core asset you are going to spinoff?

Yes. Normally for these business decisions, we will announce it publicly when we have made a decision internally. So thus far, Du Xiaoman is the last one that we have announced, and that's the one that in the process we're spinning off.

Your next question comes from the line of Piyush Mubayi of Goldman Sachs.

My question is around DuerOS, which has now reached 90 million installed base. Is there any monetization that we could build into our forecast? That's the first question. Second, following up on Alex's question, it appears that your guidance suggest a slight slowing down into the third quarter on year-on-year basis for the Core, Herman? I think, if you could just comment on that? Or is it just because your range is wide enough and we're reading too much into that number?

On the DuerOS, we are still focusing on expanding the installed base and improving user experience. There is a clear path to revenue and profit, because users will increasingly rely on voice-activated devices for all kinds of information, content and services. So eventually, we wouldn’t worry about monetization, but for the next few quarters, we would not think that DuerOS will contribute a significant portion of our revenue.

And Piyush, on your question with regards to following up on Alex's question on Search plus feed. I think the reason why you're seeing on a year-over-year basis that we're slowing down from second quarter is because of the 1 year lap from last year. If we recall the growth rate of our advertising last year, 2017, I think, first quarter we were at a negative. Second quarter, we were going, like, 7% year-over-year. And by third quarter, we significantly increased and we were growing 29%. As you probably know, we had a medical situation back in 2016. And by the third quarter of last year, basically, our revenue started going over 29%. So we have a higher base, obviously, coming into this year. So I think it's that 1 year lap of going to be a more normalized growth rate -- sorry, in third quarter last year, that's causing this quarter to be like this.

And we'll take the final question from Tian Hou of T.H. Capital.

My question is related to your Apollo. So Apollo has started to make it into a car or minibus and in the Level 4. I think it's a great accomplishment. So I wonder from here to Level 5, what are some of the milestones you guys are going to do to accomplish that? How long it may take so from here to Level 5? That's my question.

Yes. As you know, auto is a huge industry, it involves a lot of ecosystem partners, and we're working with many of them on Level 3, Level 4. And we're also working with the government to design better infrastructure or smarter roads for autonomous driving. So that's why you see that the Level 4 minibus actually launched before any passenger cars with Level 3 technologies. They're going to be used in designated areas at a slower speed, which will make sure that the safety is not an issue. So the Apollo ecosystem is a very comprehensive one. We will be selling simulation software. We will be selling HD maps. We'll be selling ACU, which is Apollo Computing Unit consisting of both hardware and software. We will have solutions for valet parking. And we have a broad spectrum of services offering to our partners. As we move forward, we'll be able to monetize from both Level 3 and Level 4, and probably some other related areas.

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.