---
title: "Baidu conference call: AI Cloud revenue will reach 30 billion yuan by 2025, computing power subscriptions increased by 143% year-on-year in Q4, and \"Carrot Run\" orders surged by 200%"
description: "Li Yanhong emphasized in the conference call that \"application is more important than the model,\" and the value of the model ultimately needs to be reflected through application. He made it clear that"
type: "news"
locale: "en"
url: "https://longbridge.com/en/news/277058017.md"
published_at: "2026-02-26T14:29:52.000Z"
---

# Baidu conference call: AI Cloud revenue will reach 30 billion yuan by 2025, computing power subscriptions increased by 143% year-on-year in Q4, and "Carrot Run" orders surged by 200%

> Li Yanhong emphasized in the conference call that "application is more important than the model," and the value of the model ultimately needs to be reflected through application. He made it clear that Baidu will adhere to an application-driven development path, continuously feeding back and iterating model capabilities based on real demand. In terms of autonomous driving, he is confident in the global scaling expansion of Apollo Go and expects that more cities will achieve positive unit economics in the future, further validating the sustainability of its business model

Despite the lingering shadow of traditional business decline, Baidu's AI transformation results have already become evident.

On the evening of February 26, Baidu disclosed during its 2025 annual financial report conference call that **thanks to the strong drive of AI large models, the company's core AI business is accelerating to become the main growth engine.** The financial report shows that AI-driven businesses, represented by AI cloud infrastructure and autonomous driving, contributed revenue exceeding 11 billion yuan in the fourth quarter, accounting for 43% of Baidu's total revenue from general business.

Among them, **the AI cloud business performed particularly well, with annual revenue reaching 30 billion yuan for the first time.** Subscription revenue from AI high-performance computing facilities grew by 143% year-on-year in the fourth quarter, becoming a major growth driver. Market data further corroborates Baidu's leading position: in 2025, Baidu Smart Cloud ranked first in the industry in both the number of successful bids and bid amounts for large model-related tender projects throughout the year, winning the title of "bid king" in this field for two consecutive years.

However, while the AI business is booming, the decline of traditional business continues to drag down overall performance. In 2025, Baidu's total annual revenue was 129.1 billion yuan, a year-on-year decline of 3%, marking the company's third consecutive quarter of total revenue shrinkage. The financial report also disclosed a long-term asset impairment loss of up to 16.2 billion yuan, further highlighting the financial pressure during the transformation period.

In the face of challenges, **Baidu is actively adjusting its strategic layout and strengthening shareholder returns.** The company's board of directors has approved a new $5 billion share repurchase plan and introduced a dividend policy for the first time. At the same time, Baidu announced plans to spin off its AI chip subsidiary Kunlun into an independent public listing to unlock the potential value of this strategic investment that has lasted over a decade, further focusing and developing its core AI business.

Key points from the conference call:

> -   AI business drives growth: The company continues to promote AI capability upgrades, making progress in AI search, embodied intelligence, digital humans, and other fields, with significant revenue growth in smart cloud and AI application businesses, driving overall revenue to increase by 5% quarter-on-quarter to 32.7 billion yuan.
> -   Cost optimization and application innovation: By reducing production costs to enhance technology application efficiency, the company launched the "Miao Da" programming platform for non-developers, intelligent visual analysis, and M Agent autonomous problem-solving capabilities, and established PSIG to integrate consumer-side AI applications, accelerating physical AI innovation.
> -   Global expansion of autonomous driving: In the fourth quarter, total orders for fully autonomous driving reached 3.4 million, a year-on-year increase of 200%; globalization efforts accelerated, entering markets such as Abu Dhabi, Dubai, and South Korea. At the same time, the company's operating cash flow turned positive, and non-GAAP operating profit grew steadily.
> -   Financial performance remains robust: Total revenue in the fourth quarter was 32.7 billion yuan, with new AI businesses effectively offsetting the decline in traditional business; operating expenses increased due to credit losses and severance costs, but non-GAAP operating profit margin remained at a healthy level.

## Multimodal and Physical AI Become New Focus

The company stated during the conference call that it has made substantial progress in technological innovation this quarter. The company not only launched an upgraded foundational model but also deeply integrated multimodal capabilities based on this foundation, providing a more solid technological base for the development and iteration of upper-layer applications In terms of specific application implementation, **Baidu has launched a series of innovative platforms aimed at diverse scenarios**. Among them, the Mota platform serves as a real-time coding tool for non-programmers, aiming to significantly lower the barriers to application development and empower a broader group of creators. Intelligent visual analysis focuses on operational compliance scenarios, enhancing management efficiency through technological means. Additionally, **the intelligent agent M agent, capable of autonomous problem-solving, made its debut, showcasing Baidu's exploration in human-computer interaction intelligence.**

It is noteworthy that Baidu recently announced the establishment of Psig (Baidu's internal business group), aimed at integrating consumer-facing artificial intelligence applications within the group, accelerating the incubation and implementation of new products through enhanced internal collaboration. The conference call also emphasized that **Baidu is actively embracing the "physical artificial intelligence" era, committed to injecting cutting-edge AI capabilities into physical industries and real scenarios, driving technology from virtual to reality, and achieving deeper industrial empowerment.**

****

## Autonomous Driving (LuoBo KuaiPao): Globalization Accelerates, Operational Volume Soars

As a model of Baidu's AI capability commercialization, the autonomous driving business delivered an impressive performance in the fourth quarter. Data shows that **Baidu's autonomous driving mobility service platform "LuoBo KuaiPao" completed fully driverless operation services 3.4 million times in the quarter, a year-on-year increase of 200%, marking a new level in its scalable operational capability.**

In terms of globalization, Baidu has also made significant breakthroughs. Building on its stable domestic operations, the company has officially launched fully driverless mobility services in Abu Dhabi and Dubai, and successfully entered the South Korean market, demonstrating the overseas potential and cross-regional adaptability of its technological solutions.

More critically, the business model of the autonomous driving sector is maturing. The conference call emphasized that **the business has achieved positive operating cash flow and operating profit under non-GAAP standards, showcasing strong growth resilience and self-sustaining capabilities**, further validating its sustainable commercialization path.

Baidu reiterated that **the company will continue to use AI technology as the core driving force to promote changes in mobility, committed to making future travel safer, more economical, and more comfortable.**

****

## Strategic Focus: AI Cloud and Capital Allocation

Looking ahead, **Baidu has clarified the growth path for its AI cloud business, which will continue to expand its market footprint by building a competitive infrastructure system and a diversified application ecosystem, solidifying its core position in the enterprise-level AI service sector.**

In terms of capital allocation, Baidu has demonstrated a firm commitment to returning value to shareholders and a strong confidence in long-term development. The company announced the launch of a stock repurchase plan of up to $5 billion and simultaneously introduced a new dividend policy to continuously enhance shareholder returns At the same time, Baidu is advancing a strategic long-term restructuring within the group, aimed at further optimizing operational efficiency, concentrating resources on core AI businesses, and unlocking potential value to attract a more diverse group of investors, laying a solid foundation for high-quality growth in the next phase.

The transcript of Baidu's Q4 2025 earnings call is as follows (assisted by AI tools):

> Operator:
> 
> Hello, thank you for joining Baidu's Q4 2025 earnings call. Currently, all participants are in listen-only mode. After the management's prepared remarks, we will have a Q&A session. Today's meeting is being recorded.
> 
> If you have any objections, you may disconnect at this time. Now, I would like to introduce today's host, Baidu's Director of Investor Relations, Juan Lin.
> 
> Juan Lin, Director of Investor Relations:
> 
> Hello everyone, welcome to Baidu's Q4 2025 earnings call. Baidu's earnings report was released earlier today, and you can find a copy on our website and through our news wire services. Joining us today are: our co-founder and CEO Robin Li; EVP Julius Rong Luo, who is responsible for Baidu's Mobile Ecosystem Group (MEG); EVP Dou Shen, who is responsible for Baidu's AI Cloud Group (ACG); and our interim CFO Jackson Junjie He.
> 
> After our prepared remarks, we will have a Q&A session. Please note that today's discussion will include forward-looking statements made under the safe harbor provisions of the Private Securities Litigation Reform Act of 1995. Forward-looking statements involve risks and uncertainties that may cause actual results to differ materially from our current expectations. For a detailed discussion of these risks and uncertainties, please refer to our latest annual report and other documents submitted to the U.S. Securities and Exchange Commission and the Hong Kong Stock Exchange. Baidu does not undertake any obligation to update any forward-looking statements unless required by applicable law. Our earnings press release and this call include discussions of certain unaudited non-GAAP financial metrics. Our press release contains a reconciliation table of the unaudited non-GAAP metrics to the most directly comparable unaudited GAAP metrics, which can be found on our investor relations website.
> 
> A reminder, this meeting is being recorded. Additionally, the live webcast of this call will be available on Baidu's investor relations website. Now, I will turn the call over to our CEO Robin.
> 
> Robin Li, Co-founder, Chairman, and CEO:
> 
> Hello everyone. In the fourth quarter, Baidu's total revenue was RMB 26.1 billion. Our core AI-driven business revenue exceeded RMB 11 billion, accounting for 43% of Baidu's total business revenue. In terms of AI cloud infrastructure, subscription revenue from AI accelerator infrastructure grew 143% year-over-year, further accelerating from 128% in the third quarter At the same time, Apollo Go has maintained strong momentum, providing 3.4 million fully autonomous driving trips this quarter. The total trips increased by over 200% year-on-year. 2025 marks the third year of our generative AI journey and a key year for AI to become the new core of our business. In 2025, we made substantial progress in expanding AI across various businesses, accelerating AI cloud growth, expanding autonomous taxi operations and improving unit economics, as well as deepening AI integration into the mobility ecosystem. By examining our business portfolio from an AI-native perspective, the momentum of our core AI-driven business continues to strengthen in 2025.

AI cloud infrastructure has gained strong market appeal due to its efficient and cost-effective training and inference capabilities. In 2025, AI cloud infrastructure revenue reached approximately RMB 20 billion, a year-on-year increase of 34%, surpassing industry growth rates. Our AI application portfolio is one of the most comprehensive in the industry, combining AI-enabled flagship products and AI-native applications that unlock new use cases. For the entire year of 2025, AI application revenue exceeded RMB 10 billion.

Apollo Go has reached an important milestone. In just 2025 alone, we provided over 10 million fully autonomous driving trips. To date, we have cumulatively provided over 20 million travel services to the public. As we accelerate our global expansion, Apollo Go's business footprint has now extended to 26 cities worldwide, solidifying our leadership position in the autonomous mobility service sector. Finally, our AI-native marketing services, including digital humans and intelligent agents, have maintained strong growth, with revenue increasing by 110% year-on-year. These achievements collectively demonstrate the growing contribution of AI to Baidu's value creation and our ability to translate AI capabilities into scalable business impact. Now, let me share the key highlights of this quarter, starting with our self-developed AI chips. This quarter, we announced the proposed spin-off and independent listing of Kunlun Chip.

After more than a decade of steadfast investment in self-developed AI chips, we are proud to see the market increasingly recognize their value and proven performance. This milestone event validates our long-term strategic vision and opens new opportunities for value creation. Our AI chips are built on a proprietary architecture developed in-house from day one. They provide stable, high-performance large-scale AI computing capabilities and are compatible with different models and frameworks.

This enables customers to deploy faster at lower integration costs. What sets our AI chips apart is that they have been validated through large-scale, real-world deployments in leading enterprises across various industries, including financial services, telecommunications, IT, and the internet. Customers choose our chips for their reliable performance, stable scalable supply, excellent software compatibility, and strong efficiency, especially in inference workloads. Looking ahead, as the demand for AI infrastructure continues to accelerate, we see significant opportunities for both Baidu and Kunlun Chip. Next, I will discuss our AI cloud infrastructure. Our infrastructure is one of the most advanced in China, supported by a diverse range of domestic and international high-performance computing resources In the fourth quarter, subscription revenue from AI accelerator infrastructure grew by 143% year-on-year, achieving triple-digit growth for the full year of 2025. Importantly, we see a continued shift in the revenue model towards a more recurring and structurally healthier direction.

The rapid expansion of enterprise AI applications has driven strong growth. As customers integrate AI into their core operations, the unique value of our full-stack, end-to-end AI architecture has become increasingly evident. By owning and optimizing all four layers, we have achieved sustainability. We have gained a continuous advantage in stability and cost-effectiveness, better meeting enterprises' demands for AI deployment.

These advantages are translating into tangible market momentum, accelerating the adoption of our AI cloud infrastructure. In the fourth quarter, we further expanded our customer coverage. Leading enterprise customers deepened their collaboration with us, driving growth in usage and spending. We also saw healthy growth contributions from mid-sized customers.

We continue to strengthen our presence across various industries, including internet services, gaming, autonomous driving, and embodied intelligence, highlighting the versatility of our infrastructure. In particular, embodied intelligence has shown significant momentum. In the fourth quarter, revenue in this vertical doubled quarter-on-quarter. We welcomed a new wave of leading humanoid robotics companies, solidifying our position as the preferred cloud service provider for China's rapidly growing embodied intelligence industry. Next, I will introduce our foundational model progress, which is a key part of our AI capabilities.

We remain fully committed to advancing our proprietary foundational model, Wenxin Yiyan (ERNIE). Following the launch of Wenxin Yiyan 5.0 last quarter, we released an updated version in January. In advancing Wenxin Yiyan, we have consistently followed a clear application-driven approach, ensuring that Wenxin Yiyan performs strongest in the most critical areas of our business portfolio. To execute this approach more effectively, we recently restructured our model development organization into two dedicated teams.

One team is responsible for advancing the state-of-the-art foundational model capabilities of Wenxin Yiyan, maintaining our technological advantage in this rapidly evolving field. The other team customizes models for specific business needs, reducing costs, improving response latency, and optimizing model size and efficiency to ensure our technology is not only cutting-edge but also easily scalable across our various businesses. The close collaboration between the two teams ensures that our technology is grounded in real-world demands while our applications benefit from ongoing technological advancements. Now, turning to AI applications.

We believe the greatest value of AI will ultimately be reflected here. We are pioneering AI applications to solve complex real-world problems for individuals and enterprises. Let me share our progress in several key areas, starting with AI-driven search. In the fourth quarter, we continued our AI search transformation, pursuing one of the most comprehensive and ambitious transformations globally.

Our focus remains on continuously improving the quality of AI search results while expanding the tasks users can complete directly in search. For example, this quarter, we introduced AI-generated infographics in search results, visualizing text-based information where appropriate, making key insights immediately visible and easy to understand. We also integrated more MCP capabilities in key scenarios, including e-commerce, healthcare, and local services

> This allows operations such as shopping, booking, and medical consultations to be seamlessly completed within the search experience. During the Spring Festival, we acted quickly by directly integrating the recently popular open-source agent framework OpenClaw into the Baidu App with one click, embracing the latest AI agent innovations, enabling our users to immediately benefit from cutting-edge agent AI capabilities. With approximately 700 million monthly active users (MAU), we provide convenient access to OpenClaw for nearly half of China's population. For Ernie Assistant, our platform-integrated AI chatbot, we enhanced the user experience by introducing broader multimodal capabilities.
> 
> These improvements have been well received by users, driving the monthly active users of Ernie Assistant to exceed 200 million in December. Meanwhile, as we expand the AI search API, its adoption accelerated in the fourth quarter, with a month-on-month increase in call volume of over 110%. With industry-leading authority, comprehensiveness, and newly added multilingual capabilities, our AI search API now opens up broader possibilities for international markets. Next is the digital human, which represents a compelling form of AI application.
> 
> They combine visual imagery, sound, and real-time interaction to create a more engaging and effective experience. By December 2025, the number of digital humans live-streaming on our platform increased by nearly 200% year-on-year. In addition to Baidu's own platform, our digital human technology is expanding into a wider range of industries. Leading companies have partnered with us, including JD.com, Zuoyebang, and TikTok, validating the performance and efficiency of our digital humans.
> 
> At the forefront of technology, we believe our ultra-realistic digital humans represent the next generation of capabilities. This quarter, production costs have dropped to about one-third of the previous quarter's level, bringing industry-leading cost-effectiveness and laying the foundation for broader adoption of this technology. Another area of progress is Miaoda.
> 
> Another area of progress is our no-code development platform Miaoda, which enables users without coding experience to build applications through natural language, including WeChat mini-programs, websites, mini-games, and more. Following the launch of the international version MeDo of Miaoda in the fourth quarter, by early February, global users had created over one million AI applications, all without writing a single line of code. Looking ahead, we see significant opportunities to unlock greater possibilities in the field of AI application development.
> 
> Finally, we are leveraging AI to solve operational issues and drive efficiency improvements across various industries. One example is our advanced visual intelligence platform ETN. ETN enables businesses to automate operational compliance and safety inspections through intelligent visual analysis, and well-known brands in coffee chains, fast food restaurants, and fine dining are now using ETN to ensure high standards of operation across their thousands of stores. Another example is our self-evolving agent FM Agent, designed to address complex operational challenges.
> 
> By autonomously reasoning through data, rules, and real-world constraints, it simulates countless scenarios to find the best solutions. We have already seen strong validation internally through our own cloud resources, as well as externally across industries such as manufacturing, energy, finance, and logistics (where efficiency improvement is a universal priority) In terms of organization, we recently established the Personal Super Intelligence Group (PSIG).
> 
> PSIG integrates our two flagship AI applications aimed at consumers—Baidu Wenku and Baidu Wangpan. Even before this organizational integration, these two teams had already collaborated at the product level, providing innovations such as FreeCanvas and Jamflow. As we accelerate the launch of new applications to drive a strong growth curve fueled by application-level innovations, this new business group will achieve deeper collaboration in the future. Turning to physical AI, Apollo Go represents our largest AI application in the physical world.
> 
> The year 2025 is a year of accelerated expansion for Apollo Go, during which we have consolidated our leadership position in operational scale and made significant progress in global expansion. We continue to rapidly expand our fully autonomous driving operations, providing 3.4 million fully autonomous driving trips in the fourth quarter, with weekly trip peaks exceeding 300,000. Total trips have increased by over 200% year-on-year. As of February 2026, we have provided over 20 million rides to the public, firmly establishing our leading position in the global autonomous driving mobility service sector.
> 
> We enter 2026 with momentum in major international markets. In the UK, we are advancing our partnerships with Uber and Lyft, pushing forward plans to pilot autonomous vehicles in London, with testing expected to begin in the first half of 2026. This represents an important step in Apollo Go's international expansion, extending our right-hand drive autonomous taxi capabilities from Hong Kong to another strategically significant market. In Switzerland, following our market entry last quarter, we have begun testing in St. Gallen. In the Middle East, we have made progress in both Abu Dhabi and Dubai.
> 
> In Abu Dhabi, we launched a fully autonomous ride-hailing service with Auto Go on Yos Island in January. In Dubai, we obtained the city's first fully autonomous testing license from the Roads and Transport Authority. We also announced the next phase of our global partnership with Uber, bringing our fully autonomous ride-hailing service to Dubai through the Uber platform. These are key milestones accelerating our progress in the UAE. In Asia, we have entered a new market—South Korea, starting from the Seoul metropolitan area, further expanding our business footprint in the region. Meanwhile, in Hong Kong, we are expanding road testing to Tsuen Wan and launching cross-district testing between Airport Island and Tung Chung, bringing us closer to commercial readiness there. As of February 2026, Apollo Go's global business footprint has reached 26 cities, demonstrating the scalability of our autonomous driving technology in diverse regulatory and operational environments. Looking ahead, we focus on accelerating expansion into more cities globally while continuously enhancing operational excellence and unit economics.
> 
> Our growing experience in different markets has strengthened our confidence in further expansion, and we expect that over time, more cities will achieve positive unit economics. Supporting this expansion is safety, which is our top priority and foundation for all our work. Our autonomous ride-hailing service is the safest in the world, with our fully autonomous vehicles averaging one airbag deployment incident every 12 million kilometers As we scale up, we will continue to strengthen safety standards to ensure lasting reliability.
> 
> Ultimately, our mission is to leverage AI to transform mobility, making it fundamentally safer, more economical, and more comfortable, while improving the travel, work, and lifestyles of billions of people. In summary, with AI now firmly integrated into our business portfolio, we believe we are well-prepared to deliver sustainable value and shape the next phase of the AI era.
> 
> Now, I will hand the call over to Henry to introduce the financial performance.
> 
> Haijian He, Interim Chief Financial Officer:
> 
> Thank you, Robin, hello everyone. We have made progress in key focus areas. In recent quarters, we have enhanced disclosures to improve transparency and driven operational efficiency.
> 
> This quarter, we took an important step by proposing the spinoff and independent listing of Kunlun Chip, unlocking value from our strategic AI chip investments, and we are particularly pleased with this milestone event. We also announced a new $5 billion stock repurchase plan and adopted a dividend policy for the first time. Additionally, by establishing the PSIG business group and integrating Baidu Wenku and Baidu Wangpan, we further focused on strategic priorities for high-potential AI applications. These actions reflect our consistent execution and ongoing commitment to creating value for shareholders.
> 
> Looking at the fourth quarter performance, we saw positive momentum. Baidu's total business revenue grew by 6% quarter-over-quarter, and non-GAAP operating profit increased by 28% quarter-over-quarter, reaching RMB 2.8 billion. Baidu's operating cash flow turned positive in the third quarter and remained positive in the fourth quarter, generating a total of RMB 3.9 billion over the two quarters. For our core AI-driven business, fourth-quarter revenue exceeded RMB 11 billion, accounting for 43% of Baidu's total business revenue.
> 
> We have seen strong momentum in multiple areas. AI cloud infrastructure continues to attract market attention, growing faster than the industry average. Our AI application portfolio is rapidly expanding in enterprise adoption. Combining AI cloud infrastructure and AI applications, our AI cloud revenue is expected to reach RMB 30 billion for the full year of 2025.
> 
> Meanwhile, Apollo Go has solidified its position as a global leader in autonomous driving mobility, boasting one of the largest business footprints in the industry and the strongest growth momentum. AI-native marketing services are also experiencing rapid growth. These achievements showcase our progress, and we believe this is just the beginning. We have a strong pipeline of initiatives for the future, and we are confident in creating lasting shareholder value.
> 
> Now, let me detail our financial performance for the fourth quarter and the full year of 2025. Total revenue for the fourth quarter was RMB 32.7 billion, a 5% quarter-over-quarter increase, primarily due to growth in Baidu's core AI-driven business. Total revenue for the full year of 2025 is expected to be RMB 129.1 billion, a 3% year-over-year decline, mainly due to a decline in traditional businesses, partially offset by growth in Baidu's core AI-driven business. Fourth-quarter cost of revenue was RMB 18.3 billion, flat quarter-over-quarter. For 2025, cost of revenue is expected to be RMB 72.4 billion, a 10% year-over-year increase, primarily due to increased costs associated with Baidu's core AI-driven business In the fourth quarter, operating expenses were RMB 13 billion, a quarter-on-quarter increase of 10%, mainly due to increased expected credit losses and one-time employee severance costs incurred to improve efficiency. Operating expenses for 2025 are projected to be RMB 46.3 billion, a year-on-year increase of 1%. Long-term asset impairment for 2025 is RMB 16.2 billion, attributed to impairment losses on core asset groups.
> 
> In the fourth quarter, operating revenue was RMB 1.5 billion, with an operating profit margin of 5%. The operating loss for 2025 is RMB 5.8 billion, with an operating loss margin of 5%. Excluding the impact of long-term asset impairment, operating revenue for 2025 is RMB 10.4 billion. Non-GAAP operating revenue for the fourth quarter was RMB 3 billion, with a non-GAAP operating profit margin of 9%.
> 
> Non-GAAP operating revenue for 2025 is RMB 15 billion, with a non-GAAP operating profit margin of 12%. In the fourth quarter, net other income was RMB 1.2 billion, compared to RMB 1.9 billion in the previous quarter. Income tax expense was RMB 1 billion, while the previous quarter had an income tax benefit of RMB 1.8 billion. For 2025, net other income is RMB 12.5 billion, compared to RMB 7.4 billion in the same period last year.
> 
> Income tax expense was RMB 1.3 billion, compared to RMB 4.4 billion in the same period last year. In the fourth quarter, net profit attributable to Baidu was RMB 1.8 billion. Baidu's net profit margin was 5%, and diluted earnings per American Depositary Share (ADS) were RMB 3.71. Non-GAAP net profit attributable to Baidu was RMB 3.9 billion. Baidu's non-GAAP net profit margin was 12%, and non-GAAP diluted earnings per ADS were RMB 10.62. For 2025, net profit attributable to Baidu is RMB 5.6 billion. Baidu's net profit margin is 4%, and diluted earnings per ADS are RMB 11.78.
> 
> Excluding the impact of long-term asset impairment, net profit attributable to Baidu is RMB 19.4 billion. Non-GAAP net profit attributable to Baidu is RMB 18.9 billion. Baidu's non-GAAP net profit margin is 15%, and non-GAAP diluted earnings per ADS are RMB 53.41. We define total cash and investments as cash, cash equivalents, restricted cash, short-term investments, net long-term time deposits, held-to-maturity investments, and adjusted long-term investments.
> 
> As of December 31, 2025, total cash and investments were RMB 294.1 billion. In the fourth quarter, operating cash flow was RMB 2.6 billion. For 2025, operating cash flow was negative RMB 3 billion, but it has remained positive for the past two quarters. As of December 31, 2025, Baidu's overall business had approximately 29,000 employees.
> 
> So, operator, let's now begin the Q&A session of the conference call.
> 
> Q&A Session
> 
> Operator:
> 
> Q&A Session
> 
> Operator:
> 
> Thank you. Ladies and gentlemen, we will now begin the Q&A session. (Operator instructions) Your first question comes from Alicia Yap of Citigroup. Please go ahead
> 
> Alicia Yap:
> 
> Thank you. Good evening, Robin, Haifeng, Haijian. Thank you for answering my questions. I have a question about the model. We have noticed that model iterations have been very active recently. How does management view the current competitive landscape? Baidu has also recently released an updated version of Wenxin Yiyan 5.0 and made some organizational adjustments. Can management discuss the strategic reasons behind these initiatives and how the company views the relationship between model evolution and application in your overall AI strategy? Thank you.
> 
> Robin Li, Co-founder, Chairman, and CEO:
> 
> Hello Alicia, I am Robin. We have indeed seen a lot of activity in model releases recently. The market competition is fierce and developing rapidly, but in all this competition, we always believe that application is more important than the model, as the model ultimately creates value through application.
> 
> That’s why we have always taken an application-driven approach with Wenxin Yiyan. Model improvements are guided by the most valuable and promising use cases. This has been consistently reflected in every iteration of Wenxin Yiyan. As I mentioned earlier, we recently released an updated version of Wenxin Yiyan 5.0.
> 
> At the same time, we have been actively undergoing organizational changes to remain agile in this fast-developing market. We have restructured the model team into different focus areas. One team continues to push frontier capabilities at the foundational model level to maintain technological leadership. Wenxin Yiyan has clear advantages in several key areas, such as creative writing, multimodal understanding, and instruction following.
> 
> We believe that we will continue to enhance the performance of Wenxin Yiyan in key application scenarios. Meanwhile, high-value application scenarios continuously provide real data and feedback for Wenxin Yiyan, driving model iterations and making Wenxin Yiyan better and better. Another team is closer to specific needs and application scenarios, focusing on reducing costs, increasing speed, and enhancing efficiency, or leveraging the best available models for specific use cases, all aimed at helping enterprises better utilize AI according to their actual needs. We recognize that model capabilities are broad, and application scenarios can be highly diverse, with no single model being the best everywhere.
> 
> Therefore, we fully leverage the areas where Wenxin Yiyan has clear advantages, and we are willing to use other models where they are more suitable. The goal is always to achieve the best application outcomes. In summary, we will continue to adhere to an application-driven approach, continuously iterating and optimizing our models based on real application needs, while also continuously improving the applications themselves to deliver better and better results, ultimately creating tangible value for users and enterprises.
> 
> Operator:
> 
> Your next question comes from Alex Yao of JP Morgan.
> 
> Please ask your question.
> 
> Alex Yao:
> 
> Good evening, management team, thank you for answering my question. I have a question about Baidu AI Cloud. We have noticed that Baidu AI Cloud revenue achieved strong growth for the full year of 2025. Could you elaborate and help us understand the key growth drivers behind the strong revenue figures? And how should we view the revenue growth prospects for AI Cloud in 2026? Thank you.
> 
> Haijian He, Interim Chief Financial Officer:
> 
> Thank you, Alex. I am Dou. In 2025, our AI cloud revenue (including revenue from AI cloud infrastructure and AI applications) reached RMB 30 billion. AI cloud infrastructure revenue grew by 34% year-on-year, outpacing the broader market.
> 
> Within AI cloud infrastructure, subscription revenue from AI accelerator infrastructure grew by 143% year-on-year in the fourth quarter, becoming a major growth driver and showing strong momentum. We are confident in maintaining strong growth momentum in 2026. Supporting our growth is the acceleration of enterprise AI adoption. We see increasing demand for both training and inference workloads.
> 
> We expect the demand for AI computing to continue to expand, creating significant opportunities in the future. Baidu's full-stack end-to-end AI architecture is a key differentiator in capturing such opportunities. The foundation of this architecture is our industry-leading AI infrastructure, which achieves an excellent balance between performance, efficiency, and cost.
> 
> Our AI infrastructure is supported by a diverse combination of chips. We have built deep expertise in heterogeneous computing and unified scheduling, enabling us to efficiently manage computing resources from different chip vendors and achieve industry-leading performance and efficiency. At the same time, our self-developed chip capabilities provide a significant competitive advantage. As Robin just mentioned, our self-developed Kunlun AI chip offers powerful performance, compatibility, and cost-effectiveness.
> 
> They have been deployed at scale among leading enterprise clients in industries such as financial services, telecommunications, energy, and the internet, with very positive market feedback. Kunlun chips are a key component of our own cloud platform's computing capabilities and play an important role in our overall AI infrastructure. As AI demand grows, the advantages of our AI infrastructure will become increasingly apparent. In addition to the AI infrastructure we just discussed, we are also continuously developing our top-tier agent infrastructure to help enterprises quickly build and scale AI agents.
> 
> We are constantly introducing the latest and most cutting-edge capabilities. For example, we recently launched a simplified Open Cloud deployment on Baidu AI Cloud, streamlining the process so that even users without coding experience can quickly deploy their own Open Cloud agents. Looking ahead to 2026, as enterprise AI deployment deepens further, we believe our cloud business will continue to grow at a rate faster than the industry.
> 
> We expect AI cloud infrastructure to maintain strong momentum, with AI accelerator infrastructure continuing to serve as a core driver, pushing our entire cloud business toward a more sustainable and high-quality growth model. Thank you.
> 
> Operator:
> 
> Your next question comes from Lincoln Kong of Goldman Sachs. Please go ahead.
> 
> Lincoln Kong:
> 
> Thank you, Mr. Chen, for answering my question. In fact, this quarter we have seen AI-enabled businesses continue to achieve quite robust growth. How does management view the current development stage of these AI-enabled businesses? When can we expect to see, for example, 50% of Baidu's overall business, and what are the key drivers for future AI-enabled businesses? Thank you.
> 
> Wang Haifeng, Chief Technology Officer:
> 
> Okay, let me first share how we view our core AI-enabled business. This includes AI cloud infrastructure, AI applications like Baidu Wenku and Baidu Wangpan, our autonomous taxi service Apollo Go, and our AI-native marketing services, including intelligent agents and digital humans. The AI-enabled business is organized based on the nature of our products and services, where AI is empowering every business to create meaningful customer value and commercial impact. In the fourth quarter, the revenue from AI-enabled business exceeded RMB 11 billion. This accounts for approximately 43% of Baidu's overall business revenue.
> 
> This percentage has rapidly increased in recent quarters, and the AI-enabled business is becoming the core driver of our overall revenue growth. Each of our AI-enabled businesses has a clear strategic positioning and competitive advantage. First is AI infrastructure, namely AI cloud infrastructure. We see enterprises expanding AI from pilot projects to production, and our full-stack end-to-end AI capabilities enable them to achieve strong performance at competitive costs. Revenue from AI cloud infrastructure is growing faster than the industry average by 2025, with subscription-based AI accelerator infrastructure revenue accelerating sharply in the fourth quarter.
> 
> Second is AI applications. We have always believed that the ultimate value of AI will be realized at the application layer, and we have built one of the most comprehensive AI application portfolios in China. As AI capabilities continue to evolve and new use cases emerge, we see significant expansion potential in this business. Third is the autonomous taxi service Apollo Go. Apollo Go is rapidly expanding while also internationalizing. We are globally leading in operational scale, safety records, efficiency, and cost structure.
> 
> Fourth is AI-native marketing services, such as intelligent agents and digital humans. They enhance interaction and conversion, and we have seen strong market acceptance and tremendous potential for the future. Therefore, looking ahead to the medium and long term, as enterprise AI deployment deepens, the monetization capabilities of AI applications improve, and physical AI applications (such as autonomous driving) continue to expand, we are confident in the growth trajectory of our AI-enabled business.
> 
> These AI-enabled businesses do not exist in isolation; they continuously strengthen each other through our full-stack capabilities. Based on current visibility, we believe that in the foreseeable future, our core AI-enabled business will become a major part of Baidu's overall business.
> 
> Operator:
> 
> Your next question comes from Wei Xiong.
> 
> Please go ahead.
> 
> Wei Xiong:
> 
> Okay. Good evening, management. Thank you for answering my question. Could management elaborate on the framework you use for capital allocation, including shareholder returns, internal investments, and potential strategic opportunities? Additionally, could management comment on the long-term strategic positioning of Kunlun Core within Baidu Group? Thank you.
> 
> Haijian He, Interim Chief Financial Officer:
> 
> Thank you, Xiong Wei. I am Henry. I believe many of you may have noticed our recent series of initiatives
> 
> These measures include strengthening disclosures, improving operational efficiency, optimizing organizational structure, advancing the proposed spinoff and independent listing of Kunlun Core, as well as announcing our new stock repurchase plan and initial dividend policy. Recently, we also reorganized PSIG, the Personal Super Intelligence Group, integrating Baidu Wenku and Baidu Wangpan. In summary, these measures reflect a coherent execution framework, demonstrating our improved management execution capability and our ongoing commitment to creating shareholder value. Take our new stock repurchase plan as an example.
> 
> We are very focused on providing clear and sustainable returns to our shareholders. Therefore, recently, in February, the board approved a new $5 billion stock repurchase plan, which we plan to execute regularly in a very disciplined and transparent manner. We also introduced Baidu's initial dividend policy. We believe that the introduction of this policy, along with the large-scale stock repurchase plan, will further enhance our shareholder return profile and attract a broader range of investors, thereby further diversifying our investor base.
> 
> As you mentioned, the proposed spinoff and independent listing of Kunlun Core is another great example. We have made very good progress in the listing process. Kunlun Core is the result of over a decade of investment, and it represents a key infrastructure component in our four-layer AI stack. We believe this spinoff and independent listing will receive high recognition from the market and unlock significant value for Baidu's shareholders. So looking ahead, we firmly believe the company has tremendous value, and we will continue to unlock it through various initiatives.
> 
> We remain committed to providing our shareholders with sustainable and consistent returns. Therefore, more initiatives will follow at the appropriate time, so please stay tuned.
> 
> Wei Xiong:
> 
> Thank you.
> 
> Operator:
> 
> Your next question comes from Gary Yu of Morgan Stanley.
> 
> Please go ahead.
> 
> Gary Yu:
> 
> Hi, thank you, management. My question is about autonomous taxi services. First, congratulations on the expansion of autonomous taxis to more countries, especially my hometown, Hong Kong. Can you share your overseas strategy for 2026? What are your key competitive advantages there? Additionally, recently Waymo's valuation reached 126 billion RMB; how does management consider unlocking the value of Apollo Go? Would you consider a spinoff? Thank you.
> 
> Robin Li, Co-founder, Chairman, and CEO:
> 
> Hi, Gary. As I mentioned last quarter, I believe that autonomous taxis have reached a critical point globally. By continuously providing safe autonomous driving mileage and positive driving mileage, we see more and more countries and regions creating supportive environments for autonomous taxi operations. We believe this industry will accelerate its development in 2026.
> 
> Apollo Go is a clear global leader in this field. We have completed over 20 million cumulative trips. At peak times, our weekly fully autonomous trips exceed 300,000. To date, the Apollo Go fleet has accumulated over 300 million autonomous driving kilometers, of which over 190 million kilometers are fully autonomous, and it has an excellent safety record We continue to advance our industry-leading technology to make journeys safer and more comfortable.
> 
> We are also accelerating our international expansion to seize global opportunities. Today, our global business footprint spans 26 cities across major continents, covering both left-hand and right-hand drive autonomous taxi markets. Our autonomous driving system operates reliably anywhere, across different traffic modes and urban environments. Notably, few players have entered the right-hand drive autonomous taxi market, while we have established a business and are making rapid progress.
> 
> Additionally, we have a fundamental cost advantage. The RT6 is the world's first mass-produced vehicle designed and manufactured from the ground up for Level 4 autonomous driving. With a cost of under $30,000 per vehicle, the RT6 offers the industry's best cost structure, combined with our leading operational efficiency, allowing us to achieve the lowest cost per mile globally while maintaining exceptional safety. We are the first company to achieve unit economic (UE) breakeven in Wuhan by the end of 2024, as you know, ride-hailing prices in most major cities are higher than in Wuhan. To accelerate global expansion, we are leveraging diverse strategic partnerships. For example, this year we will collaborate with Uber and Lyft to launch services in London, and we are also partnering with Uber in Dubai.
> 
> These partnerships drive faster and more efficient market expansion. We view Apollo Go as a strategic growth engine with tremendous long-term potential. Many major cities lack human drivers. The capacity provided by our autonomous taxi service not only offers a safer riding experience but also stimulates ride-hailing demand, thereby increasing tax revenue for governments.
> 
> It also frees up valuable land from parking lots and provides additional monetization opportunities for these real estate assets. We focus on three areas. First, actively expanding the scale of safe and comfortable operations by deploying more vehicles. Second, continuously improving unit economics. This is aimed at achieving UE breakeven in more cities.
> 
> Third, expanding through flexible business models, both domestically and internationally. As for strategic choices, we will remain flexible and assess the best path to maximize long-term shareholder returns. Of course, our focus is always on execution and sustainable growth. We believe the entire autonomous driving mobility service sector is still undervalued.
> 
> Over time, we expect the valuation of the entire industry to better reflect the transformative potential of this technology, creating meaningful upside opportunities for Apollo Go. Thank you.
> 
> Operator:
> 
> Your next question comes from Miranda Zhuang of Bank of America Securities. Please go ahead.
> 
> Miranda Zhuang:
> 
> Thank you to the management for answering my question. Happy New Year to everyone. So my question is about competition.
> 
> We have seen the consumer-facing AI competition intensify recently, especially during the Spring Festival. How do you assess the current competitive landscape? Where do you see the differentiation and market positioning of Baidu's AI to C products, such as the Wenxin Yiyan assistant? Finally, how do you consider its monetization path? Thank you.
> 
> Julius Rong Luo, Executive Vice President of Baidu Mobile Ecosystem Group (MEG):
> 
> I am Julius. The market competition for AI to C products is very fierce. We have seen some competitors adopt very aggressive marketing strategies during the past Spring Festival to rapidly expand their user base. However, with the rapid development of technology and products, we still believe that our core strategy should be based on real user needs. We are committed to continuously enhancing the capabilities of our existing products and services through AI innovation to better serve our users. In our consumer-facing flagship product, the Baidu App, we have built the Wenxin Yiyan assistant to strengthen our service capabilities throughout the user journey (from information acquisition to providing solutions and completing tasks).
> 
> In terms of information acquisition, we have significantly enhanced the way users obtain information through the Wenxin Yiyan assistant. For example, we have improved the accuracy and relevance of answers through RAG, and the error rate of the Wenxin Yiyan assistant remains low, with very few hallucinations, providing users with highly credible content. We have also integrated multilingual AI search API capabilities, allowing users to access a wider range of information sources in conversations, enhancing the richness and availability of information, which is particularly useful in scenarios like travel planning. In December, the monthly active users (MAU) of the Wenxin Yiyan assistant exceeded 200 million, and the number of conversation rounds and user engagement has grown rapidly.
> 
> In terms of task completion, we are integrating MCP agents to connect users with tools and real-world services. Just in this quarter, we have added nearly 100 service capabilities, especially in the fields of healthcare, travel, education, and e-commerce. For example, through the Baidu Health MCP integrated into the Wenxin Yiyan assistant, users can access a range of healthcare capabilities, from online consultations to offline services. In the e-commerce sector, our MCP module has achieved very strong quarter-on-quarter GMV growth.
> 
> At the same time, we have taken a different approach to the standalone Wenxin Yiyan App, positioning it as a platform for innovation and experimentation. Our early multimodal AI features have gained good attention among young users. Recently, we have added AI capabilities focused on workplace productivity, exploring its potential to handle complex tasks in professional scenarios. We have seen early positive signals in these productivity scenarios.
> 
> We take a prudent approach to monetization of AI to C products, prioritizing product excellence and user experience. As the products mature, monetization will naturally follow. Thank you for your question.
> 
> Operator:
> 
> Your next question comes from Ellie Jiang of Macquarie.
> 
> Please ask your question.
> 
> Ellie Jiang:
> 
> Hi, thank you for answering my question, good evening. My question mainly focuses on AI investment. How do you view capital expenditures related to AI over the next 12 to 24 months? How should we view the returns on these AI investments, and the expected impact on long-term profitability? Overall, where do you think we can find further opportunities for efficiency improvements to support profitability and cash flow enhancement? Thank you.
> 
> Haijian He, Interim Chief Financial Officer:
> 
> Thank you. I am Henry. First, regarding capital expenditures and AI investments, since we launched Wenxin Yiyan in March 2023, we have invested over 100 billion RMB. In the future, we will continue to maintain this level of investment.
> 
> Second, we are very focused on returns and understand investors' concerns about the return on invested capital. That is why we strive to improve financial performance and have achieved good results in key metrics over the past few quarters. For example, in the fourth quarter, Baidu's gross profit achieved double-digit quarter-on-quarter growth, and Baidu's non-GAAP operating revenue grew approximately 30% quarter-on-quarter. We also performed better in terms of margins.
> 
> Importantly, Baidu's operating cash flow turned positive in the third quarter and remained positive in the fourth quarter. In the second half of the year, operating cash flow approached 4 billion RMB. Baidu's free cash flow also turned positive in the fourth quarter. Third, we are also looking for and exploring other ways to meet our financial needs, including, for example, operating and financing leases, as well as our channels to obtain low-interest bank loans.
> 
> For instance, some of these bank loans and lease financing have interest rates as low as below 2%. Therefore, these methods help us maintain a healthy long-term financing structure while sustaining AI investments and supporting business growth. In summary, we will continue to maintain our level of AI investment while balancing investors' concerns about profitability and return cycles. We believe that even with substantial AI investments, future operating cash flow will continue to grow positively. Thank you.
> 
> Operator:
> 
> Thank you, ladies and gentlemen, our meeting today has come to an end. Thank you for your participation. You may now disconnect.
> 
> The event has ended

### Related Stocks

- [09888.HK - BIDU-SW](https://longbridge.com/en/quote/09888.HK.md)
- [CLOU.US - Global X Cloud Computing ETF](https://longbridge.com/en/quote/CLOU.US.md)
- [BIDU.US - Baidu](https://longbridge.com/en/quote/BIDU.US.md)

## Related News & Research

| Title | Description | URL |
|-------|-------------|-----|
| 百度 AI 云 2025 年营收 300 亿元 | 百度预计 2025 年 AI 云营收将达到 300 亿元，四季度基于订阅的 AI 高性能计算设施收入同比增长 143%。此外，百度智能云在 2025 年大模型相关中标项目数和金额上均位居第一，连续两年成为云厂商中标项目数和金额最多者。 | [Link](https://longbridge.com/en/news/277047146.md) |
| 阿里云推出低价 AI 编程套餐，集成四大顶级开源模型 | 阿里云推出低价编程工具订阅套餐，集成 Qwen3.5、GLM-5、MiniMax M2.5、Kimi K2.5 四大开源模型 API 服务，旨在争夺开发者市场。轻量版首月售价 7.9 元，专业版首月 39.9 元。此举引发市场关注，尤其是在 | [Link](https://longbridge.com/en/news/276950888.md) |
| AI 资本支出激增，电网更吃紧！高盛大幅上调全球 AI 用电预期：2030 年需求暴增 220% | 高盛最新报告上调 2030 年全球数据中心用电需求增幅至 220%，其中美国占据六成。AI 投资增量正从算力外溢至电力供应链，引发基础设施 “可靠性超级周期”。尽管云厂商资本开支剧增挤压自由现金流，但其仍有财力为清洁电力溢价买单。目前行业瓶 | [Link](https://longbridge.com/en/news/276852501.md) |
| AI 冲击引发恐慌？红杉林君叡：软件公司能够挺过来 | 红杉资本的林君叡表示，软件公司能够应对人工智能带来的冲击，并相信 AI 将改善大多数公司。他提到，AI 的影响是真实的，能够提升工作效率。尽管市场对 AI 的担忧影响了软件股，但红杉仍专注于利用 AI 增强人类工作成果，并投资于 Rowsp | [Link](https://longbridge.com/en/news/276940068.md) |
| 无视海外 “AI 恐慌”，中国市场狂炒 AI 赢家 | 同一个 AI，截然相反的命运。美国市场陷入” 恐慌性抛售”，中国投资者疯狂追捧：智谱 AI 一度暴涨 524%，MiniMax 一度飙升 488%。分歧本质是美国投资者焦虑于丰厚利润池面临竞争威胁，而中国关注的仍是市场渗透。 | [Link](https://longbridge.com/en/news/276531940.md) |

---

> **Disclaimer**: This article is for reference only and does not constitute any investment advice.