Baidu released its second-quarter 2026 financial results, posting total quarterly revenue of RMB 31.3 billion. Core business revenue reached RMB 25.2 billion, with AI-related businesses contributing 50% of that figure, marking the second consecutive quarter where AI has accounted for over half of core revenue. During the quarter, GPU cloud revenue within the AI cloud infrastructure segment surged 283% year-over-year, achieving triple-digit growth for the fourth straight quarter.

Following the earnings release, Chairman and CEO Robin Li, along with senior executives including Mobile Ecosystem Group President Luo Rong, Intelligent Cloud Group President Shen Dou, and CFO He Haijian, participated in the earnings call to discuss key results and field analyst questions. The following is a summary of the main Q&A session from the call.

Navigating the Competitive Landscape for Ernie

JPMorgan analyst Alex Yao asked management for their perspective on Ernie's competitive positioning, particularly as trillion-parameter models continue to emerge rapidly and push benchmark performance ceilings higher. He also inquired about the core technology and product focus for Ernie following the recent recruitment of senior foundational model talent, and what expectations investors should hold for its next phase of development.

Robin Li responded that from an industry perspective, foundational models are still iterating at a rapid pace, with different models achieving leadership in certain capabilities every few months. This indicates the field remains highly dynamic and the competitive landscape is far from settled. In such a market, long-term competitiveness often comes down to sustained technical investment, an application-driven approach, and patience. Baidu has always been a company that believes in technology and is willing to make long-term investments in it. Past experience has repeatedly proven that valuable technological innovation requires patience and persistence. Many of Baidu's important AI assets today, including Kunlun Chip and Apollo Go, are the result of over a decade of sustained investment. They have become core sources of differentiated competitiveness, with their performance and commercial value gaining increasingly broad recognition.

Similarly, Ernie has always been a vital part of Baidu's AI strategy and full-stack AI capabilities. Baidu was among the first companies in China to invest in foundational models. Along the way, there have been trials and missteps, but the determination to make Ernie competitive has never wavered. Looking ahead, Baidu will continue to invest the necessary resources to drive Ernie's ongoing development. As part of this effort, the organizational structure has been further optimized, and top AI talent has recently been brought on board. There is confidence in accelerating AI iteration and returning Ernie to the first tier of foundational models. The company will continue its application-driven approach, focusing on capabilities most important to Baidu's various applications, including AI search, digital humans, Miaoda, Farmou, and general-purpose agents like Doubao. These applications are key components driving Ernie's continuous iteration and improvement.

Using AI search as an example, when Baidu improves Ernie's ability to understand user intent and judge content quality, those improvements are directly implemented in search and feed businesses. This allows the company to quickly see results, identify areas for improvement, and feed relevant data back into model training, making the model better at understanding user intent and evaluating content quality. This closed loop serves as an important path for Ernie's development, directly converting technological progress into better product experiences, tangible user and commercial value, ultimately benefiting a wider range of users and enterprises.

Growth Drivers and Profitability Outlook for AI Cloud

Citi analyst Alicia Yap asked about the cloud business, noting the rapid revenue growth of Baidu's AI cloud infrastructure. She inquired about core growth drivers, revenue growth expectations for upcoming quarters, and how to view long-term margin potential as the business scales.

Shen Dou stated that AI cloud infrastructure revenue grew 50% year-over-year in Q2, which remains a strong growth rate above the industry average. Over a longer period, the AI cloud infrastructure has maintained rapid growth for multiple consecutive quarters, consistently outperforming the industry. The highlight this quarter was GPU cloud, with revenue up 283% year-over-year, marking the fourth consecutive quarter of triple-digit growth and accelerating from 184% in Q1. Looking ahead, multiple drivers support continued growth. Demand for AI computing power in China remains robust, and as AI becomes more deeply embedded in real-world applications and business workflows, particularly as inference workloads continue to scale rapidly, demand is expected to grow further. Meanwhile, the customer base is expanding quickly, with new customers of various sizes starting to use the AI cloud infrastructure, while existing core customers continue to increase usage and spending. Demand is also spreading across more industries and use cases, including internet, gaming, embodied intelligence, autonomous driving, smartphones, financial services, and other sectors.

Most importantly, Baidu has built and continues to strengthen a differentiated full-stack AI architecture covering chips, cloud infrastructure, large models, and applications, with competitive products at each layer. Especially at the application layer, Baidu moved early to build a portfolio of agents and AI application products. Products like Farmou, Doubao, and Miaoda have gained market recognition, strengthening the ability to capture increasingly diverse AI opportunities. Based on current demand trends, customer pipeline, and these differentiated advantages, there is confidence that AI cloud revenue can maintain strong growth in the second half of the year, with potential for further acceleration.

Regarding profitability, Shen Dou noted that profit and profit margins for AI cloud infrastructure both increased year-over-year in Q2. Looking forward, several factors will support further margin expansion. First, GPU cloud growth is significantly higher than AI cloud infrastructure overall and continues to hold a large share of the revenue mix. GPU cloud also has more attractive margin levels compared to traditional CPU cloud. With continued optimization of product and customer mix, improved resource utilization, and operational efficiency, there is room for further margin improvement as scale expands. Second, in the MaaS business, revenue from external customer token calls on the Qianfan platform is growing rapidly, though MaaS still represents a relatively small portion of total AI cloud infrastructure revenue. However, early momentum is promising. As usage scales and unit inference costs continue to decline, MaaS-related businesses are expected to unlock more profit potential and make increasingly important contributions to margins over the long term. Finally, full-stack AI capabilities and self-developed chips bring end-to-end cost advantages that support margin expansion. Overall, there remains significant room for long-term margin improvement in the AI cloud business.

Balancing AI Investment with Profitability Goals

BofA Merrill Lynch analyst Miranda Zhuang asked about the operating margin trajectory given that AI-driven businesses now account for half of revenue while capital expenditures continue to rise. She sought management's perspective on balancing ongoing AI investment with profitability targets.

He Haijian responded that AI-driven businesses continued to account for half of Baidu's core revenue this quarter, further highlighting AI's central role in the business. Within the AI-driven segment, AI cloud infrastructure is seeing rapid revenue growth with profits rising quickly and margins improving year-over-year. Within AI cloud infrastructure, GPU cloud typically has better margins and its share of total revenue continues to rise. With the ongoing business mix shift, robust market demand, and cost advantages from self-developed chips and full-stack AI capabilities, there is confidence that AI cloud infrastructure margins have considerable room for long-term expansion. As the business scales, improved resource utilization and operating leverage are also expected to further support margin expansion.

He also noted that AI applications have substantial long-term profit potential. Many of these products naturally have high user stickiness and are primarily subscription-based. As user adoption scales and commercialization advances, they have the opportunity to achieve increasingly attractive margins and make more important contributions to overall profitability. Meanwhile, Baidu remains firmly committed to its AI investment cycle. Investment decisions are driven by genuine, clear demand from customers and internal businesses, allowing most AI investments to generate revenue contributions in a relatively short period. Supply chain management capabilities are also being continuously strengthened, which should help improve capital efficiency as the business scales. While different AI investment projects have varying return timelines, with some taking longer to materialize, Baidu plans to invest decisively in areas most important to long-term competitive positioning while maintaining strict discipline on return on invested capital, operational efficiency, and cash flow. As AI businesses further scale and commercialization matures, these investments are expected to increasingly translate into sustainable profit growth.

Kunlun Chip IPO Progress and Growth Prospects

Goldman Sachs analyst Lincoln Kong asked for an update on the proposed listing of Kunlun Chip, including key milestones ahead, factors driving future growth, long-term commercial potential, and its strategic positioning within Baidu's AI ecosystem.

Shen Dou confirmed that the listing process is still underway, and the company will update the market as soon as more information becomes available. From a business perspective, there is high confidence in Kunlun Chip's long-term growth and commercial potential for several reasons. First, across the industry, AI computing demand continues to rise for both training and inference scenarios. Model capabilities keep improving, and more applications are being deployed in real-world scenarios, especially as agent technology advances into broader use cases. Inference computing demand is growing particularly fast, and this trend is expected to continue, creating long-term structural growth opportunities for the AI chip industry. Second, the domestic market has enormous growth potential, while computing supply is likely to remain tight for some time. In this environment, customers increasingly seek high-performance, high-reliability, cost-effective domestic AI chips. This creates significant opportunities for chip companies with strong technical capabilities and large-scale delivery capacity. After more than a decade of investment, Kunlun Chip has built solid capabilities in chip performance, software-hardware integration, adaptation to mainstream models and frameworks, and large-scale deployment, gaining recognition from a growing number of customers. Third, within Baidu's AI ecosystem, Kunlun Chip is an important part of the infrastructure layer of the full-stack AI architecture covering chips, cloud infrastructure, large models, and applications. Close collaboration across layers enables end-to-end optimization, delivering superior performance, reliability, and cost efficiency. This supports long-term deployment of AI cloud infrastructure and other AI businesses while further strengthening the competitiveness of Baidu's full-stack AI capabilities. Looking ahead, Kunlun Chip is expected to continue playing an important role in Baidu's AI infrastructure, capturing more commercial opportunities and serving broader market demand.

Timeline for Hong Kong Dual Primary Listing

UBS analyst Wei Xiong asked about the expected timeline for the Hong Kong dual primary listing conversion, the possibility of inclusion in Stock Connect, the strategic rationale behind the move, and its long-term impact on Baidu's investor base, stock liquidity, and valuation.

He Haijian provided the timeline, noting that the board approved the conversion to dual primary listing back in July. An application has been submitted to the Hong Kong Stock Exchange, and the receipt acknowledgement has been received. The next step is the extraordinary general meeting scheduled for August 26, where shareholders will be asked to approve certain matters required to complete the conversion. Following that, subject to HKEX approval and satisfaction of other applicable conditions, the conversion is expected to take effect within this year. Regarding Stock Connect, Baidu is actively preparing for inclusion after completing the dual primary listing, hoping the stock can be included as soon as possible. This remains subject to meeting applicable eligibility requirements and completing the review process, depending on final decisions by the relevant exchanges.

As for the strategic logic, the core purpose of the dual primary listing is to broaden the investor base, enhance stock liquidity, and provide greater flexibility in accessing both Hong Kong and US capital markets. It also allows more investors, particularly those in Asia, to better understand and participate in the value of an AI-focused Baidu. Looking further ahead, if successfully included in Stock Connect, this is expected to significantly expand participation from mainland Chinese investors and support further diversification of the shareholder base over the long term. Further updates will be provided as progress is made.

AI Search Progress and Advertising Outlook

Jefferies analyst Thomas Chong asked for updates on AI search progress across product capabilities, user experience, and commercialization. He also noted continued pressure on online marketing revenue in Q2 and asked about key factors and expectations for the second half.

Luo Rong responded that over the past few quarters, the focus of the AI search transformation has been improving the quality of AI answers to significantly enhance user experience. Accuracy and authority have always been core strengths, and these advantages have been consistently reinforced throughout the AI transition. Today, AI search can better understand what users are looking for, with more reliable answers, clearer structure, and diverse presentation formats. Hallucination rates remain low, and the model's ability to distinguish content quality has improved. As a result, more high-quality answers are being delivered while low-quality answers are decreasing. User feedback has been positive, with steady improvements in user satisfaction, search intent, and retention.

This quarter, Baidu further integrated AI search with the Ernie Assistant, transforming single search responses into more coherent, interactive multi-turn conversations that better handle follow-up questions and broader user needs. Tool calling, multi-step planning, and complex task execution capabilities have also been strengthened to help users accomplish more. Recently, the Ernie Assistant's task agent achieved top rankings in two influential third-party agent benchmark evaluations: the global benchmark PinchBench V2, which focuses on complex real-world tasks, and the domestic Super CLUE-X Claw evaluation of mainstream agent products. These results validate Ernie Assistant's leading capabilities in tool calling, multi-step planning, and task execution.

That said, industry competition remains intense. As new product formats such as robotaxis gain market acceptance and the way users acquire and consume information continues to evolve, competition for user time and attention has intensified. Meanwhile, Baidu continues to advance its AI search transformation and has deliberately restrained AI search monetization. Both factors have created short-term headwinds for the advertising business. Given that these market conditions are likely to persist, advertising revenue is expected to face continued pressure in the second half of the year. On the commercialization front, the current priority remains refining the product and delivering excellent user experience. As model capabilities, user experience, and task complexity continue to improve, more commercialization opportunities naturally suited to AI interactive experiences are expected to emerge.