This conference call transcript was computer generated and almost certianly contains errors. This transcript is provided for information purposes only.EarningsCall, LLC makes no representation about the accuracy of the aforementioned transcript, and you are cautioned not to place undue reliance on the information provided by the transcript.
Operator
Hello, and thank you for standing by for Baidu's fourth quarter and 2022 earnings conference call. At this time, all participants are in 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 your host for today's conference, Juan Lin, Baidu's Director of Investor Relations.
Juan Lin
Hello, everyone, and welcome to Baidu's fourth quarter and fiscal year 2022 earnings conference call.
spk32
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, and Zhou Shen, our EVP in charge of Baidu AI Cloud Group, ACG, and Zheng Yuli, our SVP in charge of Baidu Intelligent Driving. After our prepared remarks, we will hold a Q&A session. Please note that the discussion today will contain forward-looking statements made under the safe harbor provisions of the U.S. Private Securities Litigation Reform Act of 1995. Forward-looking statements are subject to risk and uncertainties that may cause actual results to differ materially from our current expectations. For detailed discussions of these risks and uncertainties, these refer to our latest annual report and other documents filed with ICC and Hong Kong Stock Exchange. Baidu does not undertake any obligation to update any forward-looking statement, except as required under applicable law. Our earnings press release and lead call include discussions of certain unaudited non-GAAP financial measures. Our press release contains a reconciliation of the unaudited non-GAAP measures to the unaudited most directly comparable GAAP measures and is available on our IR website at ir.baidu.com. As a reminder, this conference is being recorded. In addition, a webcast of this conference call will be available on Baidu's IR website. I will now turn the call over to our CEO, Robin.
Robin Li
Hello, everyone. In Q4 2022, Baidu core revenues were impacted by a challenging involvement caused by a rapid increase in COVID-19 cases and its impact on the broader economy. However, we managed to improve our operating profit and margin year-over-year in the quarter. Beyond that, Baidu demonstrated its value as a trusted platform for users seeking accurate and authoritative information during critical moments like the peak time of COVID-19 pandemic. As restrictions in China were relaxed and many people get infected, people were most keen to monitor the progress of the epidemic in their local areas. We quickly mobilized our resources and launched COVID-19 indices based on Baidu search queries, providing real-time updates and forecasts for better transparency. Also, we partnered with reputable hospitals and medical experts to offer guidance and advice for public health. In Q4, health-related search queries grow about 40% year-over-year, reaching a record-breaking high. Today, we are on the threshold of exciting new developments in our industry. For this call, I will focus on three major themes, and then I'll provide quarterly reviews across our BIM slides. And finally, I'll talk about our outlook for 2023. Regarding the major themes, first, although our fourth quarter results were impacted by COVID-19, we are optimistic about our path to re-accelerate revenue growth in the coming quarters. So far in Q1, we have already seen a recovery of our online marketing revenue from the trial, in particular from verticals in healthcare, travel, business services, and lifestyle. Assuming no significant new disruptions We expect this recovery trend to continue and boost our online marketing revenue through increased ad demand and take advantage of pent-up consumption in China. We also anticipate resuming cloud project implementation as business operations return to normal, signing new contracts, and accelerating AI cloud revenue growth. Second, We are riding high on the wave of AI as our earlier investments make us well positioned to benefit. Now that the industry is about to experience significant growth and monetization. Over the past many years, we have doubled down on AI, recognizing its potential to drive market expansion and increase enterprise value. Today, AI has become a cornerstone of many of our products and services. For example, in search, the early family of large models have been instrumental in delivering improved search results since March of 2019. These models enable improved ranking and multi-model search capabilities. In cloud, we leverage our AI expertise to offer industry-specific AI solutions and applications for traditional industries such as transportation, manufacturing, energy, and utilities. Our intelligent driving business built on the backbone of Baidu's computer vision and other machine learning algorithms is another testament to our commitment to this field. Recently, with users reading about ChatGPT, large language models using generated AI have created a megatrend that will revolutionize many businesses. Baidu is well positioned to capitalize on the imminent inflection point in AI, thanks to our extensive talent pool, well-developed infrastructure, a diverse array of businesses, and leading technology. We are also excited about the upcoming launch of Ernie Bot, our new conversational AI bots powered by our latest in-house large language models. We plan to fully integrate Ernie Bot across all of our operations. For our consumer-facing products, we plan to embed Ernie Bot into Baidu Search first. We believe this will reshape information generation and presentation and create a new entry point for the next generation internet, helping us attract more users and gain market share in a profitable manner. In the future, we will also integrate ErnieBot into XiaoDu, which will allow XiaoDu to significantly upgrade all the smart devices and services. We believe AIGC will also provide notable potential for ITE on content generation and user acquisition. For our to-be businesses, we plan to make this technology widely available to our customers, developers, and ecosystem partners to help boost productivity across industries. By opening the platform to generate a large language model to the public, We expect more and more business owners and entrepreneurs to build their own models and applications on our AI cloud. For intelligent driving, we plan to integrate Ernie Bot into our auto solutions, further enhancing user experience. We also believe generative pre-trained transformers will play an important role in driving the expansion of our operations. Clearly, we can benefit from the new AI inflection point as we have timely prepared our overall business for this new megatrend. Turning to my third plea, we will continue to focus on the long-term sustainable development of our company. As you all know, 2022 was an incredibly challenging year. However, we have used the time wisely to lay a strong foundation by building a more efficient organization through cost optimization, organizational adjustment, and management reshuffle. As a result, in the second half of 2022, Baidu Core's non-GAAP operating margin and profit both improved year over year. We believe a more streamlined operation will help us to achieve sustainable growth. Now, let's review the fourth quarter operating highlights. AI Cloud recorded 4% year-over-year growth in Q4. Revenue growth decelerated from the prior quarter due to the impact of COVID-19 outbreak at the end of the quarter. Nevertheless, the margins of AI Cloud in Q4 improved year-over-year. two major factors were at play. First, we continue to scale back low margin business for both app and cloud solution projects. And second, we continue to standardize our AI solutions and applications for specific use cases in traditional industries. For example, our quality inspection and the patrol inspection solutions have been adopted by many leading companies in traditional industries like chemical fiber, electric power, and renewable energy. We have seen margin improvement as we replicate our AI solutions and applications which tackle industry-specific challenges. By working with industry leaders, we are enhancing our knowledge of industry-specific challenges and standardizing AI cloud solutions and applications for scale. As an example, thermal power, which is mainly coal-fired, makes up the majority of the power supply in China, generating around 5.9 trillion kilowatt-hours in 2022, according to the National Bureau of Statistics of China. Recently, we developed an AI solution for a leading coal-fired power plant in Shanxi Province. to improve the efficiency of power generation through methods such as automated cooling system. According to our internal estimates, this solution resulted in almost a 5 gram of coal consumption per kilowatt hour, well above our customers' expectation of 1 gram reduction. This also means a 12.5 gram reduction in carbon dioxide emissions per kilowatt hour. We aim to create more similar projects to gain scale. Turning to intelligent driving, we continue to make significant progress in the fourth quarter, further reinforcing our beliefs in our strategy and BIMS model. ApolloGo remains the largest autonomous driving service provider in the world. measured by the number of rides completed. In Q4, ApolloGo provided 561,000 rides to the public on open roads. That's up 162% year-over-year. By the end of January, the cumulative number of rides exceeded 2 million. This demonstrates users' increasing acceptance of ApolloGo Autonomous Ride Hailing Service. As we scale up our operations, we are able to continuously improve our technologies and refine our services. These accomplishments also build our safety record and to bolster both users' and regulators' trust and fueling the development of the fully driverless ride-hailing market. In 2022, We made some notable breakthroughs in fully driverless operations on public roads. Since last August, we have begun to offer fully driverless ride-hailing services in Wuhan and Chongqing. In late December, we were granted permission to test drive vehicles with no safety operator in the car on public roads in the Yizhuang region of Beijing. A critical step before the capital city allows fully driverless ride-hailing operation. As I stated in the past, labor cost is a major cost component in ride-hailing operations. Removing safety operators from vehicles can significantly reduce this cost. In Q4, fully driverless rides provided to the public are increasing very fast, helping us to expand our operation and reduce cost per mile. In 2022, we also unveiled RT6, aiming to bring the cost of robot taxi to the price range of mass market EVs for the first time in China. This initiative will allow us to further reduce our cost in the future. Baidu Apollo Auto Solutions made a breakthrough engagement with a multinational automaker. It intends to cooperate with us on AVP functionality for one of their popular car models in China. Moving into mobile ecosystem. Our traffic remains strong. Both total mobile search queries and feeds distributed through Baidu app continue to experience double digit year over year growth in the quarter. In December, MA use for the Baidu app increased 4.2 percent year-over-year. Despite revenue declining in Q4, our mobile ecosystem continued to generate high operating profit, operating margin, and substantial cash flow thanks to increased efficiency. Encouragingly, we have seen a recovery in our online marketing revenues after the Chinese New Year holiday. On another positive note, Retail and e-commerce remained an outperforming vertical with more than 20% year-over-year growth, accounting for over 10% of the total advertising revenue. Users not only come to Baidu to search for information and knowledge, but also increasingly come to us to search for services and merchandise, driving up growth in merchandise search queries and transactions on Baidu. Quarterly GMB facilitated by Baidu Search grow by 84% year-over-year in Q4. Short videos contributed to year-over-year increase in feed revenue in Q4 despite macro challenges. Apart from feed, we believe short videos also benefit search in both user experience and monetization. and the growing prevalence of short videos within Baidu mobile ecosystems should drive long-term online marketing revenue growth. We continue to help content creators to generate content through AI, enriching our mobile ecosystem, in particular, producing short videos. As we move forward into 2023, our priorities are clear and focused. First, we are committed to revenue growth as we have seen a clear pathway for a successful business recovery. Second, we are poised to take advantage of the huge potential of AI to drive business growth. We aim to integrate generative AI into most of our products and services and make our cutting-edge large language models available to both consumers and businesses. We will maintain healthy growth while also strengthening our foundation. Our mobile ecosystem should continue to generate strong margins and cash flow. We aim to continue expanding margins for AI cloud and outgrowing our internet peers. We also will continue to make strategic and prudent investments in intelligent driving to capture the long-term market opportunity. With that, let me turn the call over to Rong to go through financial results.
MA
Thank you, Robin. Now let me go through the details of our first quarter and full year 2022 financial results. We closed the year 2022 with solid financial results. Baidu Core's pure portfolio revenue was RMB 25.7 billion, decreasing 1% year-over-year. In 2022, Baidu Core generated RMB 95.4 billion, or $13.8 billion in revenue, basically flat from last year. Now, online marketing business continue to grow very fast and reach 30% of by-the-court revenue in the first quarter, up from 26% a year ago, demonstrating how cloud and other AI-powered businesses have fueled our business now and will continue to do so in mid or long term. For the full year 2022, Now online marketing business accounting for 27% of Baidu core revenues up from 22% in 2021. Within the non-online marketing business, revenue from Baidu AI Cloud increased by 4% year-over-year to RMB 5.1 billion Q4 on an Apple to Apple comparison and were up by 23% year-over-year to RMB 17.7 billion in 2022 on an Apple to Apple comparison. Xiaoduo is another revenue contributor to the non-online marketing business. Xiaoduo continues to achieve solid revenue growth in both Q4 and the year 2022. By the call, online marketing revenue decreased by 6% year-over-year and 4% sequentially in Q4 due to the wage demand for advertisers in the challenging macro environment caused by COVID. By the call, online marketing revenue was down by 6% year-over-year in the year 2022. IT revenue was RMB 7.6 billion in Q4, increasing 3% year-over-year. IT revenue was RMB 29 billion in the year 2022, decreasing 5% year-over-year. Cost of revenue was RMB 16.9 billion in Q4, decreasing 2% year-over-year, primarily due to the reduction of content costs, cost of goods sold, and other costs related to new AI business, partially offset by the increase in traffic acquisition costs. Cost of revenue was IMB 63.9 billion in 2022, decreasing 1% year-over-year, primarily due to an increase in traffic acquisition costs, bandwidth costs, cost of goods sold, and other costs related to new AI business, offset by the decrease in common causes. Operating expenses were IMB 11.5 billion Q4, a decrease of 17% one-seventh year-over-year, which was primarily due to a reduction in staff-related expenses. Operating expenses were RMB 43.8 billion in 2022, decreasing by 12% year-over-year, which had primarily due to a reduction in channel spending, promotional marketing, and personnel-related expenses. Non-GAAP operating income was RMB 6.5 billion in Q4, and non-GAAP operating margin was 20%. Longan operating income was RMB 23.2 billion in 2022, and Longan operating margin was 19.19% in 2022. For Baidu Core, Longan operating income was RMB 5.5 billion, and operating margin was 21% in Q4. Longan operating income was RMB 20.9 billion, and Longan Baidu Core operating margin was 22% in the year 2022. Regarding the non-operating items, in Q4, total other income net was RMB 1.8 billion, mainly resulting from fair value gain of RMB 1.6 billion from long-term investments. Income tax expenses was RMB 1.3 billion. In 2022, total other loss net was 5.8 billion, mainly resulting from fair value loss of RMB 3.9 billion and impairment loss of RMB 3 billion from long-term investments. Income tax expenditures was RMB 2.6 billion. In Q4, net income attributable to Baidu was RMB 5 billion, and diluted earnings per ADS was RMB 13.59. Net income attributable to Baidu Core was RMB 4.8 billion, and net margin for Baidu Core was 19.19%. Non-GAAP net income attributable to Baidu was RMB 5.4 billion, Longan diluted earnings per ADS was RMB 15.25. Longan net income attributable to Baidu Core was RMB 4.9 billion and Longan net margin for Baidu Core was 19% from that. In the year 2022, net income attributable to Baidu was RMB 7.6 billion and diluted earnings per ADS was RMB 19.85. Net income attributable to Baidu Corp was RMB 7.6 billion, and net margin for Baidu Corp was 8%. Non-GAER net income attributable to Baidu was RMB 20.7 billion, and Non-GAER diluting earnings payday amounting for RMB 58.93. Non-GAER net income attributable to Baidu Corp was RMB 19.9 billion, and long-term net margin for Baidu Corp was 21%. Cash and short-term investments for Baidu Corp as of December 1, 2022 were RMB $177.4 billion or USD $25.7 billion. Free cash flow for Baidu Corp excluding IT was RMB $18.1 billion or USD $2.6 billion in 2022. Baidu Call had approximately 36,300 employees as of December 31, 2022. On a separate note, we're excited about IT's continuous efforts to improve operating efficiency. In the quarter, IT once again generated positive non-GAAP operating profit and great cash flow. On the user side, IT's average daily number of totals Subscribing members increased to 111.6 million in Q4, a net addition of 10.6 million from the previous quarter. IT maintained its largest market share for the drama category in terms of effective video views, according to ELITEN data. In addition, in the past 12 months, IT raised around US$1.3 billion post-VIP fundraisers. by to continue to hold controlling voting power in IT and consolidate IT's financial results. Operator, with that, let's now open the call to questions.
Robin
Operator, please. Sure.
Operator
Ladies and gentlemen, we will now begin the question and answer session. If you wish to ask a question, please press star 1 on your telephone and wait for your name to be announced. If you wish to cancel your request, please press star 2. Participants are requested to restrict one question at each time. Your first question comes from Alex Yao from JP Morgan. Please ask your question.
Alex Yao
Thank you management for taking my question. Congrats on a strong quarter. I have a couple of questions for the online advertising business. Can you share with us the trends you are seeing in February post the Chinese New Year holidays from an industry perspective? What are the stronger recovery industry and what are the relatively weaker recovery industry and particularly I'm interested in the offline radio verticals such as travel, local services, franchising and healthcare. Can you comment on your regular exposure to these offline radio verticals and how are they recovering since the Chinese New Year holiday? And then secondly, can you Share with us your preliminary thoughts on the advertising recovery trend in the coming months. Particularly, I think the second quarter will be a little bit special because consumption has already been normalized and you are coming against a low base on your real basis. How should we think about the advertising growth trend in the next couple of months, particularly second quarter? Thank you.
Robin Li
Hi, Alex. This is Robin. Our absence is highly correlated with COVID. As you know, when we entered the first quarter, the number of COVID infections has passed its peak and is now on a downward trend. People's daily lives are gradually back to normal and the operations of many companies are recovering as well. As the economy continues to recover, our ad revenue is gradually improving from the 12. When we look into our ad verticals, we have seen substantial improvements in our largest verticals, including healthcare, travel, lifestyle, and franchises. These are all pretty much offline businesses. And online businesses, especially online games, are relatively weaker. And regarding to the recovery by region, I think the COVID infection number has passed its peak in all of China. The regions passing the peak earlier started to recover earlier. And the post Chinese New Year holiday, Tier 2 and Tier 3 cities recovered faster than Tier 1 cities. That's because a lot of people took longer breaks from their families before going back to Tier 1 cities for work. And again, the magnitude of the recovery correlates heavily with the recovery of our top verticals. As we look beyond the first quarter and into the latter part of 2023, we're confident that our admins will benefit from China's reopening and macro recovery. And with that, we look forward to demand recovery from advertisers, which should provide significant boost to our ad revenue. I also want to highlight that advertisers prefer to spend on performance-based ads rather than pure brand promotion during the process of economic and business recovery. And search has proven to be the most effective form of performance-based advertising because users have a clear intent when using search. Search ads connect the user's intentions with the most relevant product and service offer. That's why we expect to gain more marketing dollars from our advertisers. Over the long term, we will continue to benefit from the sustained growth of China's macroeconomy improvement. Also, e-commerce and short videos will continue to drive long-term growth of our ad business. Overall, we believe our ad revenue should outgrow China's GDP in the long run. Beyond that, we are very excited about the opportunities around generative AI. We believe by incorporating Ernie Bott Baidu apps and particularly Baidu search will have an enhanced user experience. The new features powered by Ernie Bot should help us to attract new users and drive user engagement and also drive advertiser interest in Baidu too. So this will power our long-term revenue growth.
Robin
Thank you. Operator, next question please. Thank you. Your next question comes from Alicia Yap from Citigroup.
Operator
Please ask your question.
spk28
Hi, yeah, thank you. Good evening, Robin, Julius and management team. Thanks for taking my questions. My question is related to chat GPT. So there's been a lot of expectation on Baidu progress in chat GPT and also the AI generated content. Can management share with us what could be the new opportunities that you are seeing in this field and how is Baidu positioned in this trend? Any color that management can share would be appreciated. Thank you.
Robin Li
Thank you for your questions, Alicia. We are obviously excited about Chinese EPD and AIGC. It represents a megatrend that could change a lot of things. We are working on Ernie Bot, a new version of Conversational AI Bot that uses our latest technology in large language models. We will embed Ernie Bot into Baidu Search first and we'll open it to the public in March.
Robin
Baidu is the leader in China's technology innovation.
Robin Li
Remember, we launched Ernie in March 2019 and have scaled it up with well over 100 billion parameters. It is trained by serving billions of user search requests and other applications every day. So in terms of NLP capabilities, Ernie is considered as the state of the art Chinese language model. And it is not only about language, but also about understanding Chinese culture. ERMI 3.0 is already a very localized AI foundation model for China market, which means the generated large language model we are working on right now will be more suitable in Chinese language and to the China market than models developed overseas. In addition, The deep learning framework Petal-Petal, on which Ernie is based, has gained a strong momentum. Millions of developers use Petal for their AI works. There is a strong synergy between the framework layer and the model layer. AI pre-training, as you know, is very expensive. We believe our full-stack AI capabilities will allow us to build the most efficient large language model and support all kinds of applications from search to content generation or any vertical areas that could improve productivity significantly. For Baidu Search, I just mentioned that we're working hard on a revolutionary version by the search built upon ErnieBot that incorporates generative AI into our search algorithm as well as content creation. And we are adding interactive features too. Users will soon be able to interact directly with the new generative large language model. It could be, it would be complimentary or even upgrading the traditional search experience and attract more users. ChatGPT type of features could potentially become a new traffic entry point for Internet users and therefore expand the market size of search. Meanwhile, it will also help our advertisers, our content creators, and merchants, et cetera. With our strong AI capabilities, we should be able to keep iterating and upgrading the model. Search is just one example. There are many other applications we believe will also benefit from it. And for AI Cloud, our AI Cloud provides full stack offerings of four layers, from the cloud infrastructure layer to deep learning open source framework layer, then to the large foundation model layer, and ultimately to application. So these are the four layers. And by opening the generated large language model to the public, aka model as a service, we should help many business owners and entrepreneurs build their own models and applications on our cloud, bring about significant positive change and improvement in a number of areas, including increased efficiency, better decision-making, and improved customer experiences. So to recap, Baidu's strong AI capabilities build our large language model and AI foundation models. In addition to expand the market size for search, we are able to help many, many industries to build their own models, develop their own applications, and significantly improve their productivity. We believe it will be a game changer for cloud computing. AI is transforming many industries in a big way, and we are super excited about what's to come. we're building an AI ecosystem around Ernie Bot. As of today, a number of organizations have already decided to integrate Ernie Bot into their products and services. And remember, this is just the beginning of the journey. Thank you.
Ernie
Thank you, Robin.
Operator
Thank you. Your next question comes from Gary Yu, with Morgan Stanley. Please ask your question.
Gary Yu
Hi, thank you, management, and congratulations on a decent set of results. So I have a couple of questions regarding the AI cloud business. First one is for the revenue growth. Will the disruption in revenue collection that we saw in fourth quarter push out the benefit first quarter? And have you seen the reopening kicked off in projects of cloud so far? And what are the overall cloud revenue growth for 2023 that management expects to see? And what kind of product offering we can see from the cloud service as the major growth driver for 2023? And how do we expect the company growth rate and travel of the cloud market in the coming years to be? And I think a second question related to that is, can you also comment on overall IT spending budget by enterprises in the public sector versus the, so the private sector versus the public sector? And also, do you expect to win a decent number of new cloud projects this year? And do you think that the uptake will be gradual given the overall macro remains weak? And just lastly, any comment on breakeven timeline and what is the key bottleneck so far reaching break-even.
spk07
Thank you.
Dong
Hi, Gary.
Gary
Thanks for your questions. This is Dao. I will answer the first two questions and then let Rong take care of the last one. So we do have seen a gradual recovery of cloud revenues as our R&R team has begun to work on the ongoing projects post-re-opening. However, the cycle from sales lead to revenue recognition varies from a few months to a few quarters, considering that our AI cloud revenue growth should be back loaded this year. But in the meanwhile, we will keep phasing out low margin businesses for both us and cloud solution projects as we focus on margin improvement. for the cloud. This initiative will have an impact on our revenue growth going forward. But still, we aim to continue outgrowing our internet peers in 2023. If we look into the long term, the big trend for traditional industries to move business onto cloud and use AI to improve their efficiency remained unchanged. I believe the digital and intelligent transformation will be the key driver for the GDP growth in the near future. And for sure, we will benefit the most from this transformation because we think Baidu is uniquely positioned. Actually, compared to others, our AI cloud provides full stack offerings on four layers, as Robin just mentioned, you know, cloud infrastructure to deep learning platform, you know, that's PaddlePaddle, as you know, to large foundation models and applications. And all the four layers can work seamlessly together, you know, to offer end-to-end solutions, you know, to achieve optimal outcomes for our clients, you know, even at lower costs, you know, versus other solutions, you know, which is built from different providers for the cloud platform and models. In addition, end-to-end solutions can actually upgrade in much shorter time once we have improvements on any of these layers, which will further benefit the clients. So if you are in the sectors of transportation, manufacturing, energy and utilities or other industries, I believe Baidu's cloud service should be the most efficient one because in addition to the technical advantage I just explained, we already know the industry and we know the pinpoints and we have accumulated solid experience by successfully delivering high quality products already in the industries. Most importantly, as Robin just mentioned, we plan to integrate ErnieBot into our cloud and AI products. So this should help many business owners and entrepreneurs build their innovative applications, explore new business, or improve their operations. So we believe this will be a significant step for our AI cloud. and potentially reshape the whole China's cloud industry. So we will help more business owners and entrepreneurs to live from digital era to AI era, where AI technologies will play a vital role in productivity in all industries. With that, I will turn to Rong for your last part.
MA
Thanks, Dong. Yeah, I'm going to address your question regarding to our AI cloud margins. We continue to improve margins for cloud by cutting the low margin businesses and continue to build in the AI applications, which can scale up. This is a consistent strategy for years. I think as we have mentioned before, our AI cloud is actually including two parts. The first one will be the personal cloud, which continues to generate a decent operating profit margin. While the enterprise cloud is still loss-making, the margins is improving both in the growth margin and operating margin studies. In fact, if you look at the new projects we signed, the projected margins continue to improve in the past quarter. Robin has talked a lot about how we can achieve the margin improvement in his previous remarks, and this will help us continue improving our margins in the future. Note that it takes time to build a sizable AI application portfolio, which can be used by many companies in the same industries. But I'm confident that Baidu's full stack of AI capabilities can help us to . And lastly, we're pretty much on track to make the AI cloud a profitable and healthy business in the coming few years.
Dong
Thank you.
Operator
Thank you. Your next question comes from Eddie Leung with Bank of America. Please ask your question.
Eddie Leung
Good evening, guys. So I have a couple of quick questions around your investment and margins. As we are having an economic recovery, can you talk a bit about your investment plan for 2023, such as user acquisition and headcount? A bit related to that, I think Robin also mentioned that the AI chatbot could be pretty expensive to operate, right? So also you're related to that. Could you talk a little bit about the margin profile of that? And then could you give us some color on the margin profile of your core marketing pieces as well as the long-term R&D and G&A expense target? Thank you.
MA
Thanks so much for your questions, and let me try to address your questions. If we still remember, actually we start to cut cost and expenses since the end of the year 2021, which is ahead of our counterparts in the industry in China. And we carefully review and restructure our business. For mobile ecosystem, with size down and even phase out, some businesses where Baidu has obviously we have no competitive advantages or we cannot be profitable for long term. By doing so, the MEG mobile ecosystem growth today is more healthy than one year ago and having much thinner operations than before. And during the past years, the total headcount of Baidu call actually reduced by 8%. And we are now focusing on the areas where we are best in class or where we are first in class. And at the same time, for AI Cloud, we start to gradually eliminate the low margin businesses. We also tighten our controls on variable costs. You probably can see that as a result in 2022 by the cost operating expenses decreased by 9% year-over-year. If we separate them out, SG&A was down by 15% one-five year-over-year, and R&D was down by 3% year-over-year. In the second half of the year 2022, by due cost and non-GAAP operating profit margin is increasing year-over-year. And now we have a healthier cost structure, stronger organization than one year ago. So looking forward to the year 2023, we will remain very disciplined in spending. and we will continue to balance the revenue growth and margin as we are building the long-term sustainable development for the whole organization. If we look into each like items, we will continue to monitor ROI for all promotional activities, particularly the use acquisition costs, and we will carefully manage all the other variable causes. For IND, you have asked just now, we're working very hard to improve the efficiency of our team. The IND team today are required to focus on building the technologies, products, or services which can create more values for users and customers. We believe over the long term, we should be able to see the operating leverage from IND's side. At the same time, we will continue to invest for growth. and new opportunities, and closely monitor the returns. For CapEx's study, we will keep investing in cutting-edge technologies like large language models or generative AI because we believe it will transform many businesses and provide huge opportunities for IT. While we're able to further improve our efficiencies of reallocating some resources to supporting the new initiatives, In short, we will not hesitate to invest in our new promising business. So as a summary, we will look into each business. Mobile ecosystem should continue to generate high profits and cash flow in the future, which is our foundation. And for AI cloud, and to continue to expand our margins within AI cloud, enterprise cloud will continue to cut down on margin businesses and focus on key industry. As Robin has just said, we will continue to standardize our AI solutions and applications. For Personal Cloud, we will continue to earn decent profits and margins in the future. For ApolloGo, we will continue to invest for the future. But meanwhile, we will carefully monitor its operational and financial metrics to evaluate our performance and investment returns. Over the long run, we believe we still have room to optimize our cost at any expenses. But during this process, we will dynamically allocate resources and balance the growth with margins and cash flow. Thank you so much for your questions, Adrian.
Operator
Thank you. Thank you. Your next question comes from Wei Jiang with UBS. Please ask your question.
Wei Jiang
Hi, good evening, management. Congratulations on a strong quarter, and thank you for taking my question. I want to ask about your robo-taxi business. I'm wondering, what is the current unit economics for this business, and how should we think about the break-even timeline for the robo-taxi? And also, if management can share any color on your technology roadmap, that would be appreciated as well. Thank you.
Dong
Hi, thanks for your question.
spk16
This is Jamie. For the question regarding unit economics, as you know, the major cost of a taxi are labor cost and vehicle cost. Firstly, labor cost. As you know, fully driverless operation means no safety officer in the car, which helps to lower labor cost. In 2022, we made a big progress in providing fully driverless service on the open road. Now, we are providing free driverless ride-hailing service in Wuhan and Chongqing. For example, in Wuhan, we keep expanding operations by covering the wide area and longer daily operation, operating hours, and also by growing our fleet of vehicles. In December, we were allowed to test the free driverless cars on the open roads in Beijing, Yizhuang, This is an important step towards fully rehabilitative operations in the capital city. In 2023, we plan to further expand our fleet and drive for fully rehabilitative operations. We will continue to reduce labor costs by allowing more vehicles to remove in-car safety operators. More and more cities in China are supportive to autonomous ride-hailing operations. As a master leader, Baidu is benefiting from this change. Also, as we said before, scaling up operations help us to refine our whole technology and further improve our safety performance. As we scale up, we will further improving our whole technology and gaining trust from regulators. Ablego remains the largest autonomous ride-hailing service provider globally. We provide more than 1.5 million rides to the public. Now, on average, each vehicle in top cities, like Beijing, provides more than 15 rides a day. This number is already close to what a taxi can do. We believe by continuously improving our technology, more and more cities will give green light to our fully-driven ride-hailing service in the future. Secondly, reducing vehicle cost. With Apollo IT6, which is designed to offer large scale for the driverless ride-hailing services, we are bringing the cost of RoboTaxi to the price range of mass market electric vehicles for the first time. We believe vehicle cost will continue to decline in the future as China's EV market is developing very fast. and the entire supply chain is highly localized. The cost of auto parts still has a lot of room to decrease. Compiling those two factors, we believe that the unit economy will continue to improve. As Robin said, once we scale up, we can leverage large language model to improve technology. And this will help us to further improve safety performance. Thank you.
Jamie
Thank you.
Operator
Thank you. Your next question comes from Kenneth Fong with Credit Suisse. Please ask your question.
Kenneth Fong
Hi. Thank you, management, for taking my questions and congrats on a very strong set of results. I have a question on the user traffic and user time spans. Have you seen any impact or changes in these two parameters as China reopens, especially in the big cities? Thank you.
Robin Li
Hi, Ken. This is Robin. People started to spend more time outdoors post the reopening. But actually, user time spent for our major apps as a whole increased 6.6% year-over-year, I think, in January. and that outperforms the overall mobile market. We have a wide variety of apps that cater to various needs of users. Baidu app is one of the very few super apps in China, and Baidu Search has a unique value proposition. We help people quickly find the most relevant and authoritative information online through technology. So whenever users want to search for information and knowledge, they come to Baidu. When they come back to work, this becomes more apparent. Even if they want to check for academic development in their regions or check entrance exam scores, this kind of need can be better and better met by the Baidu search and other Baidu apps. Over the past years, we continue to use technology to enrich content and services so we can provide our users with better experiences in search, in feed, and in many other products. Today, people not only come to us to search for information, but also to find products and services. This effort has strengthened Baidu's leading position in China's mobile Internet space. And another highlight is, of course, Ernie Bot. As I mentioned before, it has the potential to become a new traffic entry point for people online. Also, using AI to generate short video content is a typical application for generative AI. We believe AIGC will help us accumulate more short videos on our platform over time. And a direct benefit would be increasing video viewership and ad revenue. Thank you.
Operator
Thank you, Benjamin. Your next question comes from Lincoln Kong with Goldman Sachs. Please ask your question.
Benjamin
Thank you, management, for the opportunity.
Benjamin
I have a follow-up in terms of ChatGPT and our Ernie bot. So, it does seem that, you know, many domestic companies are recently keep announcing their partnering with us or used by those, you know, they are learning about technology. So, can you imagine elaborate a bit more on the format of this partnership and what could be the monetization opportunity here and how big should we think about this revenue potential? and also in terms of competition. So how should we think about the Internet companies coming offering a similar service to challenges that limits our modernization opportunity here?
spk19
Thank you.
Robin Li
Sure. We're glad to see that after we announced the plan to launch ErnieBot, many organizations reached out to us and expressed their strong interest to work with EarningBot. Some already announced their plan to partner with us to integrate EarningBot capabilities into their products or services or operations. More and more enterprises realize that the generated large language models are going to change their industries, and they would like to capture this opportunity. EarlyBot has attracted so much interest, also shows that Baidu is well-recognized for our leading AI capabilities and the age of AI has finally come. And regarding to monetization opportunities, EarlyBot will be integrated into Baidu Search, enhance the user experience, and users will be much more dependent on us for all kinds of tasks and needs. therefore significantly expand the market size of search. We are using AIGC to expand our content like text, images, or videos, and this will create a great opportunity for us to attract new users, increase user time spent and user stickiness, help us gain share in the online ad market over the long run. And for cloud customers, they will be able to leverage our full stack AI capabilities, not just the basic stuff, storage, or bandwidth, or database, this kind of thing. They can develop their apps based on our AI framework, Pedal to Pedal, and the foundation model, Earning. It will be much easier, much more efficient, much more powerful. So this will propel our cloud revenue, too. And our full stack of AI capabilities is quite unique. It contains cloud infrastructure, deep learning platform like PedoPedo, and large language models, and widely used applications. We have a strong presence in all four layers, and these four layers work together efficiently. Our large language model, Ernie 3.0, has been trained with billions of daily user search requests and other popular Baidu family of apps. Baidu's Ernie bot will be using this kind of hugely well-tagged data pool to help improve adapt quickly over time. The barrier to entry for this is very high. It requires multiple years of significant investment, and we have the first mover advantage. The technology behind Bernie Box is called reinforcement learning with human feedback. So here, the human feedback part is critical. As the first mover, we get more and more human feedback. and further improve our capabilities way ahead of anyone else. And right now we are focused on preparing the launch of Ernie Bot and make continuous improvement after the launch. And we are also focusing on areas that we believe Ernie Bot will create immediate value. So stay tuned. Thank you.
Benjamin
Thank you.
Robin Li
Thank you.
Operator
Ladies and gentlemen, that does conclude our conference for today. Thank you for participating. You may all disconnect. Thank you. Thank you. Bye. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. you Thank you. Thank you.
spk00
Thank you. music music
Operator
Hello, and thank you for standing by for Baidu's fourth quarter and 2022 earnings conference call. At this time, all participants are in 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 your host for today's conference, Juan Lin, Baidu's Director of Investor Relations.
Juan Lin
Hello, everyone, and welcome to Baidu's fourth quarter and fiscal year 2022 earnings conference call.
spk32
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, Zhou Shen, our EVP in charge of Baidu AI Cloud Group, ACG, and Zheng Yuli, our SVP in charge of Baidu Intelligent Driving. After our prepared remarks, we will hold a Q&A session. Please note that the discussion today will contain forward-looking statements made under the safe harbor provisions of the U.S. Private Security Litigation Reform Act of 1995. Forward-looking statements are subject to risk 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 ICC and Hong Kong Stock Exchange. Baidu does not undertake any obligation to update any forward-looking statement, except as required under applicable law. Our earnings press release and lead call include discussions of certain unaudited non-GAAP financial measures. Our press release contains a reconciliation of the unaudited non-GAAP measures to the unaudited most directly comparable GAAP measures and is available on our IR website at ir.baidu.com. As a reminder, this conference is being recorded. In addition, a webcast of this conference call will be available on Baidu's IR website. I will now turn the call over to our CEO, Robin.
Robin Li
Hello, everyone. In Q4 2022, Baidu core revenues were impacted by a challenging involvement caused by a rapid increase in COVID-19 cases and its impact on the broader economy. However, we managed to improve our operating profit and margin year over year in the quarter. Beyond that, Baidu demonstrated its value as a trusted platform for users seeking accurate and authoritative information during critical moments like the peak time of COVID-19 pandemic. As restrictions in China were relaxed and many people get infected, people were most keen to monitor the progress of the epidemic in their local areas. We quickly mobilized our resources and launched COVID-19 indices based on Baidu search queries, providing real-time updates and forecasts for better transparency. Also, we partnered with reputable hospitals and medical experts to offer guidance and advice for public health. In Q4, health-related search queries grow about 40% year-over-year, reaching a record-breaking high. Today, we are on the threshold of exciting new developments in our industry. For this call, I will focus on three major themes, and then I'll provide quarterly reviews across our BIM slides. And finally, I'll talk about our outlook for 2023. Regarding the major themes, first, although our fourth quarter results were impacted by COVID-19, we are optimistic about our path to re-accelerate revenue growth in the coming quarters. So far in Q1, we have already seen a recovery of our online marketing revenue from the trial, in particular from verticals in healthcare, travel, business services, and lifestyle. Assuming no significant new disruption We expect this recovery trend to continue and boost our online marketing revenue through increased ad demand and take advantage of pent-up consumption in China. We also anticipate resuming cloud project implementation as business operations return to normal, signing new contracts, and accelerating AI cloud revenue growth.
spk22
Second,
Robin Li
We are riding high on the wave of AI, and our earlier investments make us well positioned to benefit. Now that the industry is about to experience significant growth and monetization. Over the past many years, we have doubled down on AI, recognizing its potential to drive market expansion and increase enterprise value. Today, AI has become a cornerstone of many of our products and services. For example, in search, the early family of large models have been instrumental in delivering improved search results since March of 2019. These models enable improved ranking and multi-model search capabilities. In cloud, we leverage our AI expertise to offer industry-specific AI solutions and applications for traditional industries such as transportation, manufacturing, energy, and utilities. Our intelligent driving business built on the backbone of Baidu's computer vision and other machine learning algorithms is another testament to our commitment to this field. Recently, with users reading about ChatGPT, large language models using generated AI have created a mega-trend that will revolutionize many businesses. Baidu is well-positioned to capitalize on the imminent inflection point in AI, thanks to our extensive talent pool, well-developed infrastructure, a diverse array of businesses, and leading technology. We're also excited about the upcoming launch of Ernie Bot, our new conversational AI bot powered by our latest in-house large language models. We plan to fully integrate Ernie Bot across all of our operations. For our consumer-facing products, we plan to embed Ernie Bot into Baidu Search first. We believe this will reshape information generation and presentation and create a new entry point for the next generation internet, helping us attract more users and gain market share in a profitable manner. In the future, we will also integrate ErnieBot into XiaoDu, which will allow XiaoDu to significantly upgrade all the smart devices and services. We believe AIGC will also provide notable potential for ITE on content generation and user acquisition. For our to-be businesses, we plan to make this technology widely available to our customers, developers, and ecosystem partners to help boost productivity across industries. By opening the platform to generate a large language model to the public, We expect more and more business owners and entrepreneurs to build their own models and applications on our AI cloud. For intelligent driving, we plan to integrate Ernie Bot into our auto solutions, further enhancing user experience. We also believe generative pre-trained transformers will play an important role in driving the expansion of our operations. Clearly, we can benefit from the new AI inflection point as we have timely prepared our overall business for this new megatrend. Turning to my third theme, we will continue to focus on the long-term sustainable development of our company. As you all know, 2022 was an incredibly challenging year. However, we have used the time wisely to lay a strong foundation by building a more efficient organization through cost optimization, organizational adjustment, and management reshuffle. As a result, in the second half of 2022, Baidu Core's non-GAAP operating margin and profit both improved year over year. We believe a more streamlined operation will help us to achieve sustainable growth. Now, let's review the fourth quarter operating highlights. AI Cloud recorded 4% year-over-year growth in Q4. Revenue growth decelerated from the prior quarter due to the impact of COVID-19 outbreak at the end of the quarter. Nevertheless, the margins of AI Cloud in Q4 improved year-over-year. two major factors were at play. First, we continue to scale back low-margin business for both app and cloud solution projects. And second, we continue to standardize our AI solutions and applications for specific use cases in traditional industries. For example, our quality inspection and control inspection solutions have been adopted by many leading companies in traditional industries like chemical fiber, electric power, and renewable energy. We have seen margin improvement as we replicate our AI solutions and applications which tackle industry-specific challenges. By working with industry leaders, we are enhancing our knowledge of industry-specific challenges and standardizing AI cloud solutions and applications for scale. As an example, thermal power, which is mainly coal-fired, makes up the majority of the power supply in China, generating around 5.9 trillion kilowatt-hours in 2022, according to the National Bureau of Statistics of China. Recently, we developed an AI solution for a leading coal-fired power plant in Shanxi Province. to improve the efficiency of power generation through methods such as automated cooling systems. According to our internal estimates, this solution resulted in almost a 5 gram of coal consumption per kilowatt hour, well above our customers' expectation of 1 gram reduction. This also means a 12.5 gram reduction in carbon dioxide emissions per kilowatt hour. We aim to create more similar projects to gain scale. Turning to intelligent driving, we continue to make significant progress in the fourth quarter, further reinforcing our beliefs in our strategy and BIMS model. ApolloGo remains the largest autonomous driving service provider in the world. measured by the number of rides completed. In Q4, ApolloGo provided 561,000 rides to the public on open roads. That's up 162% year over year. By the end of January, the cumulative number of rides exceeded two million. This demonstrates users' increasing acceptance of ApolloGo Autonomous Ride Hailing Service. As we scale up our operations, we are able to continuously improve our technologies and refine our services. These accomplishments also build our safety record and to bolster both users' and regulators' trust and fueling the development of the fully driverless ride-hailing market. In 2022, We made some notable breakthroughs in fully driverless operations on public roads. Since last August, we have begun to offer fully driverless ride-hailing services in Wuhan and Chongqing. In late December, we were granted permission to test drive vehicles with no safety operator in the car on public roads in the Yizhuang region of Beijing, a critical step before the capital city allows fully driverless ride-hailing operation. As I stated in the past, labor cost is a major cost component in ride-hailing operations. Removing safety operators from vehicles can significantly reduce this cost. In Q4, fully driverless rides provided to the public are increasing very fast, helping us to expand our operation and reduce cost per mile. In 2022, we also unveiled RT6, aiming to bring the cost of robot taxi to the price range of mass market EVs for the first time in China. This initiative will allow us to further reduce our cost in the future. Baidu Apollo Auto Solutions made a breakthrough engagement with a multinational automaker. It intends to cooperate with us on AVP functionality for one of their popular car models in China. Moving into mobile ecosystem. Our traffic remains strong. Both total mobile search queries and feeds distributed through Baidu app continue to experience double digit year over year growth in the quarter. In December, MAUs for the Baidu app increased 4.2 percent year-over-year. Despite revenue declining in Q4, our mobile ecosystem continued to generate high operating profit, operating margin, and substantial cash flow thanks to increased efficiency. Encouragingly, we have seen a recovery in our online marketing revenues after the Chinese New Year holiday. On another positive note, Retail and e-commerce remained an outperforming vertical with more than 20% year-over-year growth, accounting for over 10% of the total advertising revenue. Users not only come to Baidu to search for information and knowledge, but also increasingly come to us to search for services and merchandise, driving up growth in merchandise search queries and transactions on Baidu. Quarterly GMB facilitated by Baidu Search grow by 84% year-over-year in Q4. Short videos contributed to year-over-year increase in feed revenue in Q4 despite macro challenges. Apart from feed, we believe short videos also benefit search in both user experience and monetization. and the growing prevalence of short videos within Baidu mobile ecosystems should drive long-term online marketing revenue growth. We continue to help content creators to generate content through AI, enriching our mobile ecosystem, in particular, producing short videos. As we move forward into 2023, our priorities are clear and focused. First, we are committed to revenue growth as we have seen a clear pathway for a successful business recovery. Second, we are poised to take advantage of the huge potential of AI to drive business growth. We aim to integrate generative AI into most of our products and services and make our cutting-edge large language models available to both consumers and businesses. Third, We will maintain healthy growth while also strengthening our foundation. Our mobile ecosystem should continue to generate strong margins and cash flow. We aim to continue expanding margins for AI cloud and outgrowing our Internet peers. We also will continue to make strategic and prudent investments in intelligent driving to capture the long-term market opportunities. With that, let me turn the call over to Rong to go through financial results.
MA
Thank you, Robin. Now let me go through the details of our first quarter and full year 2022 financial results. We closed the year 2022 with solid financial results. Baidu Core's pure portfolio revenue was RMB 25.7 billion, decreasing 1% year-over-year. In 2022, Baidu Core generated RMB 95.4 billion, or $13.8 billion in revenue, basically flat from last year. Now, online marketing business continue to grow very fast and reach 30% of by-the-court revenue in the first quarter, up from 26% a year ago, demonstrating how cloud and other AI-powered businesses have fueled our business now and will continue to do so in mid or long term. For the full year 2022, Now, online marketing business accounting for 27% of Baidu Cloud revenues, up from 22% in 2021. Within the non-online marketing business, revenue from Baidu AI Cloud increased by 4% year-over-year to RMB 5.1 billion Q4 on Apple-to-Apple comparison, and will up by 23% year-over-year to RMB 17.7 billion in 2022 on Apple-to-Apple comparison. Xiaoduo is another revenue contributor to the non-online marketing business. Xiaoduo continues to achieve solid revenue growth in both Q4 and the year 2022. By the call, online marketing revenue decreased by 6% year-over-year and 4% sequentially in Q4 due to the weak demand for advertisers in the challenging macro environment caused by COVID. By the call, online marketing revenue was down by 6% year-over-year in the year 2022. IT revenue was RMB 7.6 billion in Q4, increasing 3% year-over-year. IT revenue was RMB 29 billion in the year 2022, decreasing 5% year-over-year. Cost of revenue was RMB 16.9 billion in Q4, decreasing 2% year-over-year, primarily due to the reduction in quantum costs, cost of goods sold, and other costs related to new AI business, partially offset by the increase in traffic acquisition costs. Cost of revenue was IMB 63.9 billion in 2022, decreasing 1% year-over-year, primarily due to an increase in traffic acquisition costs, bandwidth costs, cost of goods sold, and other costs related to new AI business, offset by the decrease in common causes. Operating expenses were IMB 11.5 billion Q4, a decrease of 17% one-seventh year-over-year, which was primarily due to a reduction in stock-related expenses. Operating expenses were RMB 43.8 billion in 2022, decreasing by 12% year-over-year, which has primarily due to a reduction in channel spending, promotional marketing, and personnel-related expenses. Non-GAAP operating income was RMB 6.5 billion in Q4, and non-GAAP operating margin was 20%. Nongen operating income was RMB 23.2 billion in 2022, and Nongen operating margin was 19.19% in 2022. For Baidu Core, Nongen operating income was RMB 5.5 billion, and operating margin was 21% in Q4. Nongen operating income was RMB 20.9 billion, and Nongen Baidu Core operating margin was 22% in the year 2022. Regarding the non-offering items, in Q4, total other income net was RMB 1.8 billion, mainly resulting from a fair value gain of RMB 1.6 billion from long-term investments. Income tax expenses was RMB 1.3 billion. In 2022, total other loss net was 5.8 billion, mainly resulting from a fair value loss of RMB 3.9 billion and an impairment loss of RMB 3 billion from long-term investments. Income tax expenditures was RMB 2.6 billion. In Q4, net income attributable to Baidu was RMB 5 billion, and diluted earnings per ADS was RMB 13.59. Net income attributable to Baidu Core was RMB 4.8 billion, and net margin for Baidu Core was 19.19%. Non-GAAP net income attributable to Baidu was RMB 5.4 billion, Nongad diluted earnings per ADS was RMB 15.25. Nongad net income attributable to Baidu Core was RMB 4.9 billion and Nongad net margin for Baidu Core was 19% from that. In the year 2022, net income attributable to Baidu was RMB 7.6 billion and diluted earnings per ADS was RMB 19.85. Net income attributable to Baidu Corp was RMB 7.6 billion, and net margin for Baidu Corp was 8%. Non-GAER net income attributable to Baidu was RMB 20.7 billion, and Non-GAER's diluting earnings payday amounting for RMB 58.93. Non-GAER net income attributable to Baidu Corp was RMB 19.9 billion, and long-term net margin for Baidu Corp was 21%. Cash and short-term investments for Baidu Corp as of December 1, 2022 were RMB $177.4 billion or USD $25.7 billion. Free cash flow for Baidu Corp excluding ITE was RMB $18.1 billion or USD $2.6 billion in 2022. Baidu Call had approximately 36,300 employees as of December 31, 2022. On a separate note, we're excited about IT's continuous efforts to improve operating efficiency. In the quarter, IT once again generated positive non-GAAP operating profit and great cash flow. On the user side, IT's average daily number of totals Subscribing members increased to 111.6 million in Q4, a net addition of 10.6 million from the previous quarter. IT maintained its largest market share for the drama category in terms of effective video views, according to enlightened data. In addition, in the past 12 months, IT raised around US$1.3 billion post-Viz fundraisers. Baidu continue to hold controlling voting power in IT and consolidate IT's financial results. Operator, with that, let's now open the call to questions.
Robin
Operator, please.
Operator
Sure. Ladies and gentlemen, we will now begin the question and answer session. If you wish to ask a question, please press star one on your telephone and wait for your name to be announced. If you wish to cancel your request, please press star two. Participants are requested to restrict one question at each time. Your first question comes from Alex Yao from JP Morgan. Please ask your question.
Robin
Thank you management for taking my question.
Alex Yao
Congrats on a strong quarter. I have a couple of questions for the online advertising business. Can you share with us the trends you are seeing in February post the Chinese New Year holidays from an industry perspective? What are the stronger recovery industry and what are the relatively weaker recovery industry and particularly I'm interested in the offline related verticals such as travel, local services, franchising, and healthcare. Can you comment on your revenue exposure to these offline related verticals and how are they recovering since the Chinese New Year holiday? And then secondly, can you Share with us your preliminary thoughts on the advertising recovery trend in the coming months. Particularly, I think the second quarter will be a little bit special because consumption has already been normalized and you are coming against a low base on your real basis. How should we think about the advertising growth trend in the next couple of months, particularly second quarter?
Robin
Thank you.
Robin Li
Hi, Alex. This is Robin. Our absence is highly correlated with COVID. As you know, when we entered the first quarter, the number of COVID infections has passed its peak, and it's now on a downward trend. People's daily lives are gradually back to normal, and the operations of many companies are recovering as well. As the economy continues to recover, our ad revenue is gradually improving from the 12th. When we look into our ad verticals, we have seen substantial improvements in our largest verticals, including healthcare, travel, lifestyle, and franchises. These are all pretty much offline businesses. And the online business, especially online games, is relatively weaker. And regarding to the recovery by region, I think the COVID infection number has passed its peak in all of China. The regions passing the peak earlier started to recover earlier, and the post-Chinese New Year holiday, Tier 2 and Tier 3 cities recovered faster than Tier 1 cities. That's because a lot of people took longer breaks from their families before going back to Tier 1 cities for work. And again, the magnitude of the recovery correlates heavily with the recovery of our top verticals. As we look beyond the first quarter and into the latter part of 2023, we're confident that our admins will benefit from China's reopening and macro recovery. And with that, we look forward to demand recovery from advertisers, which should provide significant boost to our ad revenue. I also want to highlight that advertisers prefer to spend on performance-based ads rather than pure brand promotion during the process of economic and business recovery. And search has proven to be the most effective form of performance-based advertising because users have a clear intent when using search. Search ads connect the user's intentions with the most relevant product and service offer. That's why we expect to gain more marketing dollars from our advertisers. Over the long term, we will continue to benefit from the sustained growth of China's macro economy improvement. Also, e-commerce and short videos will continue to drive long-term growth of our admins. Overall, we believe our ad revenue should outgrow China's GDP in the long run. Beyond that, we are very excited about the opportunities around generative AI. We believe by incorporating Ernie Bott Baidu apps, and particularly Baidu Search, will have an enhanced user experience. The new features powered by Ernie Bot should help us to attract new users and drive user engagement and also drive advertiser interest in Baidu too. So this will power our long-term revenue growth.
Robin
Thank you. Next question, please. Thank you. Your next question comes from Alicia Yap from Citigroup.
Operator
Please ask your question.
spk28
Hi, yeah, thank you. Good evening, Robin, Julius and management team. Thanks for taking my questions. My question is related to chat GPT. So there's been a lot of expectation on Baidu progress in chat GPT and also the AI generated content. Can management share with us what could be the new opportunities that you are seeing in this field and how is Baidu positioned in this trend? Any color that management can share would be appreciated. Thank you.
Robin Li
Thank you for your questions, Alicia. We are obviously excited about Chinese EPD and AIGC. It represents a megatrend that could change a lot of things. We are working on ErnieBot, a new version of conversational AI bot that uses our latest technology in large language models. We will embed ErnieBot into Baidu Search first and we'll open it to the public in March. Baidu is the leader in China's technology innovation. We have been working on large language models for a few years. Remember, we launched Ernie in March 2019 and have scaled it up with well over 100 billion parameters. It is trained by serving billions of user search requests and other applications every day. So in terms of NLP capabilities, Ernie is considered as the state-of-the-art Chinese language model. And it is not only about language, but also about understanding Chinese culture. Fermi 3.0 is already a very localized AI foundation model for China market, which means the generated large language model we are working on right now will be more suitable in Chinese language and to the China market than models developed overseas. In addition, our deep learning framework, Petal-Petal, on which Ernie is based, has gained a strong momentum. Millions of developers use Petal for their AI works. There is a strong synergy between the framework layer and the model layer. AI pre-training, as you know, is very expensive. We believe our full stack AI capabilities will allow us to build the most efficient large language model and support all kinds of applications from search to content generation or any vertical areas that could improve productivity significantly. For Baidu Search, I just mentioned that We're working hard on a revolutionary version by research built upon Ernie Bot that incorporates generative AI into our search algorithm as well as content creation. And we are adding interactive features too. Users will soon be able to interact directly with the new generated large language model. It could be, it would be complimentary or even upgrading the traditional search experience and attract more users. Catchy PT type of features could potentially become a new traffic entry point for internet users and therefore expand the market size of search. Meanwhile, it will also help our advertisers, our content creators, and merchants, et cetera. With our strong AI capabilities, we should be able to keep iterating and upgrading the model. Search is just one example. There are many other applications we believe will also benefit from it. And for AI Cloud, our AI Cloud provides full stack offerings of four layers, from the cloud infrastructure layer to deep learning open source framework layer, then to the large foundation model layer, and ultimately to application. So these are the four layers. And by opening the generated large language model to the public, aka model as a service, we should help many business owners and entrepreneurs build their own models and applications on our cloud, bring about significant positive change and improvement in a number of areas, including increased efficiency, better decision making, and improved customer experiences. So to recap, Baidu's strong AI capabilities build our large language model and AI foundation models. In addition to expand the market size for search, we are able to help many, many industries to build their own models, develop their own applications, and significantly improve their productivity. We believe it will be a game changer for cloud computing. AI is transforming many industries in a big way, and we are super excited about what's to come. We're building an AI ecosystem around Ernie Bot. As of today, a number of organizations have already decided to integrate Ernie Bot into their products and services. And remember, this is just the beginning of the journey. Thank you.
Ernie
Thank you, Robin.
Operator
Thank you. Your next question comes from Gary Yu with Morgan Stanley. Please ask your question.
Gary Yu
Hi, thank you, management, and congratulations on a decent set of results. So I have a couple questions regarding the AI cloud business. First one is for the revenue growth. Will the disruption in revenue collection that we saw in fourth quarter push out the benefit first quarter? And have you seen the reopening kicked off in projects of cloud so far? And what are the overall cloud revenue growth for 2023 that management expects to see? And what kind of product offering we can see from the cloud service as the major growth driver for 2023? And how do we expect the company growth rate and travel of the cloud market in the coming years to be? And I think a second question related to that is, can you also comment on overall IT spending budget by enterprises in the public sector versus the, so the private sector versus the public sector? And also do you expect to win a decent number of new cloud project this year? And do you think that the uptake will be gradual given the overall macro remains weak? And just lastly, any comment on breakeven timeline
spk07
and what is the key bottleneck so far reaching break-even.
Dong
Thank you. Hi, Gary.
Gary
Thanks for your questions. This is Dao. I will answer the first two questions and then let Rong take care of the last one. So we do have seen a gradual recovery of cloud revenues as our R&D team has begun to work on the ongoing projects post-re-opening. However, the cycle from sales lead to revenue recognition varies from a few months to a few quarters. Considering that, our AI cloud revenue growth should be back loaded this year. In the meanwhile, we will keep phasing out low margin businesses for both us and cloud solution projects as we focus on margin improvement. for the cloud. This initiative will have an impact on our revenue growth going forward. But still, we aim to continue outgrowing our internet peers in 2023. If we look into the long term, the big twine for traditional industries to move business onto cloud and use AI to improve their efficiency remained unchanged. I believe the digital and intelligent transformation will be the key driver for the GDP growth in the near future, and for sure we will benefit the most from this transformation because we think Baidu is uniquely positioned. Actually, compared to others, our AI cloud provides full-stack offerings on four layers, as Robin just mentioned, you know, cloud infrastructure to deep learning platform, you know, that's PaddlePaddle, as you know, to large foundation models and applications. And all the four layers can work seamlessly together, you know, to offer end-to-end solutions, you know, to achieve optimal outcomes for our clients, you know, even at lower costs versus other solutions, you know, which is built from different providers for the cloud platform and models. In addition, end-to-end solutions can actually upgrade in much shorter time once we have improvements on any of these layers, which will further benefit the clients. So if you are in the sectors of transportation, manufacturing, energy, and utilities or other industries, I believe Baidu's cloud service should be the most efficient one because in addition to the technical advantage I just explained, we already know the industry and we know the pinpoints. And we have accumulated solid experience by successfully delivering high quality products already in the industries. Most importantly, as Robin just mentioned, we plan to integrate Ernie Bot into our cloud and AI products. So this should help many business owners and entrepreneurs build their innovative applications, explore new business, or improve their operations. So we believe this will be a significant step for our AI cloud. and potentially reshape the whole China's cloud industry. So we will help more business owners and entrepreneurs to live from digital era to AI era, where AI technologies will play a vital role in productivity in all industries. With that, I will turn to Rong for your last part.
MA
Yeah, I'm going to address your question regarding to our AI cloud margins. We continue to improve margins for cloud by cutting the low-margin businesses and continue to build in the AI applications, which can scale up. This is a consistent strategy for years. I think as we have mentioned before, our AI cloud is actually including two parts. The first one will be the personal cloud, which continues to generate a decent operating profit margin, while the enterprise cloud is still loss-making. while the margins is improving, both in the growth margin and operating margin status. In fact, if you look at the new projects we signed, the projected margins continued improving in the past quarter. Robin has talked a lot about how we can achieve the margin improvement in his previous remarks, and this will help us continue improving our margins in the future. Note that it takes time to build a sizeable AI application portfolio, which can be used by many companies in the industries. But I'm confident that Baidu's full-style AI capability can help us to do so. And lastly, we're pretty much on track to make the AI cloud a profitable and healthy business in the coming few years.
Robin
Thank you. Thank you.
Operator
Your next question comes from Eddie Leung with Bank of America. Please ask your question.
Eddie Leung
Good evening, guys. So I have a couple of quick questions around your investment and margins. As we are having an economic recovery, can you talk a bit about your investment plan for 2023, such as user acquisition and headcount? A bit related to that, I think Robin also mentioned that the AI chatbot could be pretty expensive to operate, right? So also you'll be there to that. Could you talk a little bit about the margin profile of that? And then could you give us some color on the margin profile of your core marketing pieces as well as the long-term R&D and GMA expense target? Thank you.
MA
Thanks so much for your questions, and let me try to address your questions. If we still remember, actually we start to cut cost and expenses since the end of the year 2021, which is ahead of our counterparts in the industry in China. And we carefully review and restructure our business. For mobile ecosystem, with size down and even face down, some businesses where Baidu has obviously will have no competitive advantages or we cannot be profitable for long term. By doing so, the MEG mobile ecosystem growth today is more healthy than one year ago and having much leaner operations than before. And during the past years, the total headcount of Baidu call actually reduced by 8%. And we are now focusing on the areas where we are best in class or where we are first in class. And at the same time, for AI Cloud, we start to gradually eliminate the low-margin businesses. We also tighten our controls on variable costs. You probably can see that as a result, in 2022, by the cost of operating expenses decreased by 9% year-over-year. If we separate them out, SG&A was down by 15% one-five year-over-year, and IMD was down by 3% year-over-year. In the second half of the year 2022, Baidu calls a non-GAAP operating profit margin is increasing year-over-year. And now we have a healthier cost structure, stronger organization than one year ago. So looking forward to the year 2023, we will remain very disciplined in spending. and we will continue to balance the revenue growth and margin as we are building the long-term sustainable development for the whole organization. If we look into each like items, we will continue to monitor ROI for all promotional activities, particularly the use acquisition costs, and we will carefully manage all the other variable causes. For IND, you have asked just now, we're working very hard to improve the efficiency of our team. The IND team today are required to focus on building the technologies, products, or services which can create more values for users and customers. We believe over the long term, we should be able to see the operating leverage from IND's side. At the same time, we will continue to invest for growth. and new opportunities, and closely monitor the returns. For CapEx's study, we will keep investing in cutting-edge technologies like large language models or generative AI because we believe it will transform many businesses and provide huge opportunities for IT. While we're able to further improve our efficiencies of reallocating some resources to supporting the new initiatives, such as the earning ball. In short, we will not hesitate to invest in our new promising business. So as a summary, we will look into each business. Mobile ecosystem should continue to generate high profits and cash flow in the future, which is our foundation. And for AI cloud, we aim to continue to expand our margins. Within AI cloud, enterprise cloud will continue to cut down on margin businesses and focus on key industry. As Robin and Doug have just said, we will continue to standardize our AI solutions and applications. For personal cloud, we will continue to earn decent profits and margins in the future. For ApolloGo, we will continue to invest for the future. But meanwhile, we will carefully monitor its operational and financial metrics to evaluate our performance and investment returns. Over the long run, we believe we still have room to optimize our cost and expenses. But during this process, we will dynamically allocate resources and balance the growth with margins and cash flow. Thank you so much for your questions, Adrian.
Operator
Thank you. Thank you. Your next question comes from Wei Jiang with UBS. Please ask your question.
Wei Jiang
Hi. Good evening, management. Congratulations on a strong quarter, and thank you for taking my question. I want to ask about your RoboTaxi business. I'm wondering what is the current unit economics for this business, and how should we think about the break-even timeline for the RoboTaxi? And also, if management can share any color on your technology roadmap, that would be appreciated as well. Thank you.
Dong
Hi, thanks for your question.
spk16
This is James. For the question regarding unit economics, as you know, the major cost of a taxi are labor cost and vehicle cost. Firstly, labor cost. As you know, fully driverless operation means no safety officer in the car, which helps to lower labor cost. In 2022, we made a big progress in providing fully driverless service on the open road. Now we are providing fleet driverless ride-hailing service in Wuhan and Chongqing. For example, in Wuhan, we keep expanding operations by covering the wide area and longer daily operations operating hours and also by growing all fleet vehicles. In December, we were allowed to test the fleet driverless cars on the open roads in Beijing, Yizhuang, This is an important step towards fully rehabilitative operations in the capital city. In 2023, we plan to further expand our fleet and drive for fully rehabilitative operations. We will continue to reduce labor costs by allowing more vehicles to remove in-car safety operators. More and more cities in China are supportive to autonomous ride-hailing operations. As a master leader, Baidu is benefiting from this change. Also, as we said before, scaling up operations help us to refine our whole technology and further improve our safety performance. As we scale up, we will further improving our whole technology and gaining trust from regulators. Ablego remains the largest autonomous ride-hailing service provider globally. We provided more than 1.5 million rides to the public. Now, on average, each vehicle in top cities, like Beijing, provides more than 15 rides a day. This number is already close to what a taxi can do. We believe by continuously improving our technology, more and more cities will give green lights to our fully-drived ride-hailing service in the future. Secondly, reducing vehicle cost. With Apollo IT6, which is designed to offer large-scale fully driverless ride-hailing services, they are bringing the cost of Robtex to the price range of mass-market electric vehicles for the first time. We believe vehicle costs will continue to decline in the future as China's EV market is developing very fast. and the entire supply chain is highly localized. The cost of auto parts still has a lot of room to decrease. Compiling those two factors, we believe that the unique economics of Oploco will continue to improve. As Robin said, once we scale up, we can leverage large language model to improve technology. This will help us to further improve safety performance.
Dong
Thank you.
Jamie
Thank you.
Operator
Thank you. Your next question comes from Kenneth Fong with Credit Suisse. Please ask your question.
Kenneth Fong
Hi. Thank you, management, for taking my questions and congrats on the very strong set of results. I have a question on the user traffic and user time spans. Have you seen any impact or changes in these two parameters as China reopens, especially in the big cities? Thank you.
Robin Li
Hi, Ken. This is Robin. People started to spend more time outdoors post the reopening, but actually user time spent for our major apps as a whole increased 6.6% year-over-year, I think, in January. and that outperformed the overall mobile market. We have a wide variety of apps that cater to various needs of users. Baidu App is one of the very few super apps in China, and Baidu Search has a unique value proposition. We help people quickly find the most relevant and authoritative information online through technology. So whenever users want to search for information and knowledge, they come to Baidu. When they come back to work, this becomes more apparent. Even if they want to check for academic development in their regions or check entrance exam scores, this kind of need can be better and better met by the Baidu Search and other Baidu apps. Over the past years, we continue to use technology to enrich content and services so we can provide our users with better experiences in search, in feed, and in many other products. Today, people not only come to us to search for information, but also to find products and services. These efforts have strengthened Baidu's leading position in China's mobile Internet space. And another highlight is, of course, Ernie Bot. As I mentioned before, it has the potential to become a new traffic entry point for people online. Also, using AI to generate short video content is a typical application for generated AI. We believe AIGC will help us accumulate more short videos on our platform over time. And a direct benefit would be increasing video viewership and ad revenue. Thank you.
Operator
Thank you, Benjamin. Your next question comes from Lincoln Kong with Goldman Sachs. Please ask your question.
Benjamin
Thank you, management, for the opportunity. I have a follow-up in terms of ChatGPT and our Ernie bot. So it does seem that, you know, many domestic companies recently keep announcing they're partnering with us. We're used by those, you know, they are learning about technology. So can you imagine elaborate a bit more on the format of this partnership and what could be the monetization opportunity here and how big should we think about this revenue potential? and also in terms of competition. So how should we think about the Internet companies coming offering a similar service to challenges and limit our monetization opportunity here?
spk19
Thank you.
Robin Li
Sure. We're glad to see that after we announced the plan to launch ErnieBot, many organizations reached out to us and expressed their strong interest to work with the earning bot. Some already announced their plan to partner with us to integrate earning bot capabilities into their products or services or operations. More and more enterprises realize that generated large language models are going to change their industries and they would like to capture this opportunity. EarlyBot has attracted so much interest, also shows that Baidu is well recognized for our leading AI capabilities, and the age of AI has finally come. And regarding to monetization opportunities, EarlyBot will be integrated into Baidu Search, enhance the user experience, and users will be much more dependent on us for all kinds of tasks and needs. Therefore, significantly expand the market size of search. We are using AIGC to expand our content like text, images, or videos, and this will create a great opportunity for us to attract new users, increase user time spent, and user stickiness. Help us gain share in the online ad market over the long run. And for cloud customers, they will be able to leverage our full-stack AI capabilities, not just the basic stuff, storage or bandwidth or database, this kind of thing. They can develop their apps based on our AI framework, PedoPedo, and the foundation model, Earning. It will be much easier, much more efficient, much more powerful. So this will propel our cloud revenue too. And our full stack of AI capabilities is quite unique. It contains cloud infrastructure, deep learning platform like PedoPedo, and large language models, and widely used applications. We have a strong presence in all four layers, and these four layers work together efficiently. Our large language model, Ernie 3.0, has been trained with billions of daily user search requests and other popular Baidu family of apps. Baidu's Ernie bot will be using this kind of hugely well-tagged data pool to help improve adapt quickly over time. The barrier to entry for this is very high. It requires multiple years of significant investment, and we have the first mover advantage. The technology behind Bernie Box is called reinforcement learning with human feedback. So here, the human feedback part is critical. As the first mover, we get more and more human feedback. and further improve our capabilities way ahead of anyone else. And right now we are focused on preparing the launch of EarlyBot and make continuous improvement after the launch. And we are also focusing on areas that we believe EarlyBot will create immediate value. So stay tuned. Thank you.
Benjamin
Thank you. Thank you.
Operator
Ladies and gentlemen, that does conclude our conference for today. Thank you for participating. You may all disconnect.
Disclaimer