--- title: "BIDU (Trans): Prioritizing Shareholder Returns and Better Mgmt. Execution" description: "Below is Dolphin Research's Trans of the $Baidu(BIDU.US) FY25Q4 earnings call. For our take on the print, cf. 'BIDU: Buybacks backstop, but the rerating still needs a narrative'.Key financials recap f" type: "topic" locale: "en" url: "https://longbridge.com/en/topics/38917013.md" published_at: "2026-02-26T14:53:41.000Z" author: "[Dolphin Research](https://longbridge.com/en/news/dolphin.md)" --- # BIDU (Trans): Prioritizing Shareholder Returns and Better Mgmt. Execution Below is Dolphin Research's compiled Trans of $Baidu(BIDU.US) FY25 Q4 earnings call. For the earnings read-through, cf. '[百度：回购托底，但翻盘还得靠 “叙事”](https://longbridge.cn/topics/38912994?channel=SH000001&invite-code=355628&app_id=longbridge&utm_source=longbridge_app_share&locale=zh-CN&share_track_id=386d5f2a-1326-46c9-9d53-d2ea65ddd2c0)'. **I. Key Financials Recap** **Core AI Biz. metrics (Q4 2025)** - Baidu General Business revenue: RMB 26.1bn. - Core AI-powered Business revenue: over RMB 11.0bn, 43% of Baidu General Business. - AI Cloud Infra – subscription revenue from AI accelerator infra: +143% YoY, accelerating vs. Q3's +128%. - Apollo Go fully driverless ops orders: 3.4mn delivered, total orders up over 200% YoY. **Core AI Biz. metrics (FY2025)** - AI Cloud Infra revenue: approx. RMB 20.0bn (+34% YoY), outpacing industry. - AI applications revenue: over RMB 10.0bn. - Apollo Go fully driverless ops orders: delivered over 10.0mn, with cumulative services to the public exceeding 20.0mn. - AI native marketing services (digital humans and agents) revenue: +110% YoY. **II. Earnings Call Details** **2.1 Management Highlights** **(1)** **Core AI capabilities and infrastructure** **Kunlunxin:** Built on a self-developed architecture, delivering stable, high-performance AI compute at scale with broad compatibility across models and frameworks. This lowers integration cost and speeds deployment for customers. It has been deployed at scale across leading enterprises in financial services, telecom, IT and internet. **AI Cloud:** Among the most advanced infra in China. Its full-stack, end-to-end AI architecture is increasingly differentiated, with sustained advantages in stability and cost efficiency by owning and optimizing all four layers. The revenue mix is shifting toward a more recurring and healthier structure. **Foundation models (ERNIE):** We adhere to an application-led approach. Recently, we reorganized the model development org into two dedicated teams: one pushing frontier foundation model capabilities for ERNIE, and another customizing models for specific business needs to reduce cost, improve latency, and optimize scale and efficiency. This ensures tech leadership while being easier to scale across businesses. **(2) AI application progress** **AI Search:** Ongoing AI-led search transformation improves result quality and expands user tasks that can be completed directly in results. Examples include AI-generated infographics and MCP-integrated actions for shopping and booking. By integrating the open-source agent framework OpenClaw into the Baidu App, nearly 700mn MAU can benefit directly. **ERNIE Assistant:** Broader multimodal capabilities pushed Dec. MAU above 200mn. Adoption of AI Search APIs accelerated, with calls up over 110% QoQ. This underpins stronger engagement. **Digital humans:** In Dec. 2025, the number of digital humans livestreaming on Baidu's platform rose nearly 200% YoY. Ultra-realistic digital human production cost fell to about one-third of Q3, delivering an industry-leading price-performance ratio. Partnerships include JD, Zuoyebang, and TikTok. **MeDo (no-code dev platform):** By early Feb., global users had created over 1mn AI apps on the platform. Adoption continues to rise. **Enterprise AI apps:** ETN (vision intelligence platform) is used by coffee chains and QSRs for automated ops compliance and safety checks. FM Agent (self-evolving agent) has been validated across manufacturing, energy, and financials. Use cases are scaling. **(3) Physical AI / Robotaxi (Apollo Go)** **Scale and growth:** Q4 delivered 3.4mn orders, with weekly peaks above 300k orders. Total orders were up over 200% YoY, and cumulative orders have exceeded 20.0mn, reinforcing global leadership. Momentum remains strong. **Global expansion: Global footprint now spans 26 cities.** - UK: Advancing partnerships with Uber and Lyft, with autonomous testing planned in London in H1 2026. - Switzerland: Testing launched in St. Gallen. - Middle East: Progress in Abu Dhabi (service launched via Auto Go) and Dubai (secured the first fully driverless testing permit, and announced service via Uber). - Asia: Entered South Korea (Seoul metro area); expanded Hong Kong testing (between Tsuen Wan, Airport Island, and Tung Chung). **Safety and business model:** Safety is paramount. A fully driverless vehicle experiences an airbag-deploying incident only after over 12mn km driven. With growing multi-market experience, we are confident about further scaling and achieving positive unit economics in more cities. **(4) Other updates** **Shareholder returns:** Announced a new $5bn share repurchase and adopted the first dividend policy, underscoring a sustained focus on shareholder value creation. Execution will be regular, disciplined, and transparent. **Org changes:** Established the Personal Super Intelligence Group (PSIG), integrating Baidu Wenku and Baidu Netdisk. Deeper collaboration aims to accelerate new app launches and cultivate an application-led growth curve. This enhances product velocity. **2.2 Q&A** **Q: Model iterations have been very active lately. How does management view the current competitive landscape? BIDU recently released an updated ERNIE 5.0 and made some organizational changes. Could you discuss the strategic rationale behind these moves, and how the company views the interplay between model evolution and applications within the overall AI strategy?** A: Model releases have indeed been active, with intense and fast-changing competition. Across all competition, we firmly believe applications matter more than models, because models ultimately create value through applications. That is why we have consistently taken an application-led approach. Model improvements are guided by the most valuable and promising use cases, and this has been consistent in every ERNIE iteration. We recently rolled out an updated ERNIE 5.0. In parallel, we proactively evolve the organization to stay agile in a rapidly changing market. We have reorganized the model team into distinct focus areas. One team continues to push frontier capabilities at the foundation model level to maintain technological leadership. ERNIE has clear advantages in several domains, including creative writing, multimodal understanding, and instruction following. We are confident ERNIE will keep improving in key application scenarios. Meanwhile, high-value applications provide real-world data and feedback to drive iteration, making the model increasingly better. Another team works closer to specific needs and applications, focused on lowering cost, increasing speed, boosting efficiency, or leveraging the best available model for targeted use cases, all to help enterprises utilize AI based on practical needs. We recognize model capabilities are broad and use cases highly diverse, and no single model leads across all fronts. We fully leverage ERNIE where it has clear strengths, while staying open to other models where they fit better. In short, we will continue an application-led approach, iterating and optimizing models driven by real application needs. We will also keep improving the applications themselves to deliver better outcomes and create tangible value for users and enterprises. **Q: BIDU's AI Cloud delivered strong growth in FY2025. Could you detail the key drivers behind this robust revenue growth? How should we think about AI Cloud growth in 2026?** A: In 2025, our AI Cloud revenue (including AI Cloud Infra and AI applications) reached RMB 30.0bn. AI Cloud Infra revenue rose 34% YoY, ahead of the market. Subscription revenue from AI accelerator infra grew 143% YoY in Q4 and has become a primary growth driver. Momentum is strong, and we remain highly confident in sustaining robust growth in 2026. Our growth is underpinned by accelerating enterprise AI adoption. Demand for both training and inference workloads is rising, and we expect AI compute needs to keep expanding, creating significant opportunities. BIDU's full-stack, end-to-end AI architecture is a key differentiator to capture this opportunity. Our infra balances performance, efficiency, and cost exceptionally well. It is powered by a diversified chip portfolio, and we have deep expertise in heterogeneous compute and unified scheduling, enabling efficient management of compute resources from multiple chip vendors and industry-leading performance and efficiency. Our self-developed chip capabilities also provide material competitive advantages. As Robin noted, our in-house Kunlunxin AI chips deliver strong performance, compatibility, and cost efficiency. They are deployed at scale among leading enterprise customers in financial services, telecom, energy and internet, with highly positive feedback. Kunlunxin is a critical part of our own cloud compute and plays an important role in our overall AI infra. As AI demand grows, our infra advantages will become increasingly pronounced. Beyond infra, we are continuously evolving our industry-leading agent infrastructure to help enterprises quickly build and scale AI agents. We keep introducing cutting-edge capabilities. For example, we recently launched simplified open-source Claw deployment on Baidu AI Cloud, streamlining setup so even users without coding experience can quickly deploy their own open-source Claw agents. Looking to 2026, as enterprise AI deployments deepen further, we are confident our cloud business will keep outgrowing the industry. We expect AI Cloud Infra to maintain strong momentum, with AI accelerator infra continuing as the core driver, steering the overall cloud business toward a more sustainable, higher-quality growth model. **Q: AI businesses continue to show steady growth. How does management view their current stage? When could they reach 50% of Baidu General Business? What are the key drivers of future growth?** A: Let me share how we think about our core AI businesses. This includes AI Cloud Infra, AI applications (e.g., Baidu Wenku and Baidu Netdisk), the Apollo Go Robotaxi business, and AI-native marketing services (including agents and digital humans). We organize AI businesses by the nature of their products and services, with AI enabling each business to deliver meaningful customer value and commercial impact. In Q4, AI business revenue exceeded RMB 11.0bn, accounting for 43% of Baidu General Business. This share has risen quickly in recent quarters, and AI is becoming the core driver of our overall revenue growth. Each AI business has a clear strategic position and competitive edge. **First, AI Cloud Infra.** As enterprises scale AI from pilots to production, our full-stack, end-to-end capabilities deliver strong performance at competitive cost. AI Cloud Infra revenue grew faster than the industry Avg. in 2025, and subscription-based AI accelerator infra revenue sharply accelerated in Q4. **Second, AI applications.** We have long believed AI's end value will be realized at the application layer, and we have built one of the most comprehensive AI app portfolios in China. With evolving AI capabilities and new use cases, we see substantial expansion potential. **Third, Apollo Go Robotaxi.** Apollo Go is scaling quickly while expanding internationally. We lead globally in ops scale, safety record, efficiency, and cost structure. Execution remains a priority. **Fourth, AI-native marketing services** such as agents and digital humans. They improve engagement and conversion, and we have seen strong market acceptance with significant future potential. We are investing to capture this growth. Looking mid-to-long term, as enterprise AI deployments deepen, monetization of AI applications improves, and physical AI use cases (e.g., autonomous driving) continue to expand, we are confident in the growth trajectory of our AI businesses. These AI businesses are not isolated; they reinforce each other through our full-stack capabilities. Based on current visibility, we believe our core AI businesses will become the majority of Baidu General Business in the foreseeable future. We will stay focused on disciplined scaling. **Q: What framework does management use to allocate capital across shareholder returns, organic investment, and potential strategic opportunities? Also, could you comment on Kunlunxin's long-term strategic position within BIDU?** A: Many of you may have noticed our recent series of actions. These include enhanced disclosure, improved operational efficiency, optimized org structure, progress on the proposed Kunlunxin spin-off and independent listing, as well as announcing a new share repurchase and the first dividend policy. We also formed PSIG, integrating Baidu Wenku and Baidu Netdisk. Overall, these actions reflect a coherent execution framework, improved managerial execution, and a sustained commitment to shareholder value. Take the new buyback plan as an example. We are focused on clear and sustainable returns for shareholders. In Feb., the BOD approved a new $5bn share repurchase plan, and we intend to execute it regularly in a disciplined and transparent manner. We also introduced BIDU's first dividend policy. We believe the dividend policy plus a large buyback will further strengthen our shareholder return profile and attract a broader investor base, making it more diversified. As you noted, the proposed spin-off and independent listing of Kunlunxin is another good example. We have made solid progress on the listing process. Kunlunxin is the result of over a decade of investment and a key infra component within our four-layer AI architecture. We believe the spin-off and listing will be well-received by the market and unlock substantial value for BIDU. We are committed to long-term value creation. Looking ahead, we are confident the company has significant intrinsic value and will continue to unlock it through various actions. We remain committed to sustainable and consistent returns for shareholders. More initiatives will be introduced in due course, so please stay tuned. **Q: On Robotaxi, congrats on expanding to more countries, especially my hometown Hong Kong. Could you share the overseas strategy for 2026 and your key competitive advantages there?** **Also, given Waymo's recent valuation, how is management thinking about unlocking value for Apollo Go? Would you consider a spin-off?** A: As I mentioned last quarter, I believe Robotaxi has reached an inflection point globally. By consistently delivering safe, positive miles, we are seeing more countries and regions create favorable environments for Robotaxi operations. We expect the industry to accelerate in 2026. Apollo is a clear global leader. We have completed over 20.0mn cumulative orders, with weekly fully driverless orders exceeding 300k at peak. To date, the Apollo Go fleet has driven over 300mn km, including more than 190mn km fully driverless, with an excellent safety record. We continue advancing industry-leading tech to make rides safer and more comfortable. At the same time, we are accelerating international expansion to capture global opportunities. Our footprint now spans 26 cities across continents, covering both left-hand and right-hand drive markets. Our autonomous system operates reliably across different traffic patterns and urban environments. Notably, there are very few players entering the Robotaxi market, while we have already built operations and are scaling rapidly. This positions us well. We also have fundamental cost advantages. RT6 is the world's first purpose-built mass-production vehicle designed from scratch for L4 autonomy. With per-vehicle cost below $30k, RT6 offers the best-in-class cost structure coupled with leading operating efficiency, enabling the world's lowest cost per mile while maintaining exceptional safety. We were the first to achieve UE break-even in Wuhan by end-2024, and as you know, ride-hailing prices in most major cities are higher than Wuhan. To accelerate global expansion, we leverage diverse strategic partnerships. For example, we are working with Uber and Lyft to launch services in London this year, and we are also collaborating with Uber in Dubai. These partnerships enable faster and more efficient market rollout. We view Apollo Go as a strategic growth engine with significant long-term potential. Many large cities face a shortage of human drivers. The capacity provided by our Robotaxi service not only offers safer mobility, it also stimulates ride-hailing demand, increasing tax revenue for governments. It frees up valuable land used for parking and creates additional monetization opportunities for these real-estate assets. We are focused on three areas: first, actively scaling safe and comfortable operations by deploying more vehicles. Second, continuously improving unit economics, with a goal to achieve UE break-even in more cities this year. Third, expanding domestically and internationally with flexible business models. As for strategic options, we will stay flexible and evaluate the best path to maximize long-term shareholder returns. Of course, execution and sustainable growth remain our focus. We believe the autonomous ride-hailing industry is still undervalued. Over time, we expect industry valuations to better reflect the transformative potential of this technology, creating meaningful upside for Apollo Go. We will act prudently. **Q: Competition in consumer-facing AI products has intensified recently, especially during the Spring Festival. How do you assess the current competitive dynamics? What differentiates BIDU's AI-to-C products like ERNIE Assistant, and how do you think about monetization?** A: Competition in AI-to-C is intense. We observed some competitors employing very aggressive tactics during Spring Festival to rapidly expand user bases. However, with rapid advances in tech and products, we remain convinced our core strategy should be grounded in real user needs. We are committed to continuously enhancing our existing products and services through AI innovation to better serve users. Within our flagship consumer product, the Baidu App, we have embedded ERNIE Assistant to strengthen full-journey service from information sync to solutions and task completion. On the information sync side, ERNIE Assistant significantly improves how users obtain information. For example, we improve answer accuracy and relevance via RAG, and ERNIE Assistant maintains low error rates with minimal hallucinations, providing highly trustworthy content. We also integrated multilingual AI Search API capabilities, allowing users to access a broad set of sources during conversations, improving richness and usability. This is very helpful for scenarios such as travel planning. In Dec., ERNIE Assistant MAU exceeded 200mn, with rapid growth in dialog turns and interactions. For task completion, we are integrating MCP agents to connect users with tools and real-world services. In Q4 alone, we added nearly 100 service capabilities, especially across healthcare, travel, education, and e-commerce. For example, through Baidu Health MCP integrated in ERNIE Assistant, users can access a range of healthcare capabilities with online-to-offline services. In e-commerce, our MCP modules delivered very strong QoQ GMV growth. Meanwhile, we take a different approach for the standalone ERNIE App, positioning it as a platform for innovation and experimentation. Our early multimodal AI features have resonated with younger users. Recently, we added AI capabilities focused on workplace productivity, tapping into its potential to handle complex tasks in professional scenarios. We are seeing early positive signals in these productivity use cases. We adopt a prudent stance on monetization for AI-to-C products, prioritizing product excellence and user experience. Monetization will come naturally as products mature. **Q: On AI investment. How do you view AI-related capex over the next 12–24 months? How should we think about returns on these investments and the impact on margins? Broadly, what efficiency opportunities can support margin and cash flow improvement?** A: First, on capex and AI investment, since our launch in Mar. 2023, we have invested over RMB 100bn. Looking ahead, we will maintain this investment intensity. Second, we are highly focused on returns and understand investors' attention on ROIC. That is why we have worked to improve financial performance and delivered good outcomes across key metrics in recent quarters. For example, in Q4 BIDU's GP achieved double-digit QoQ growth, and BIDU's non-GAAP OP rose approx. 30% QoQ. Margin performance also improved. Importantly, BIDU's operating cash flow turned positive in Q3 and remained positive in Q4, with nearly RMB 4.0bn in H2 operating cash flow. Free cash flow also turned positive in Q4. Third, we have identified and explored alternative financing methods to support our financial needs, including operating leases and finance leases, and we have access to low-interest bank loans. Some bank loans and leasing facilities carry rates below 2%. These approaches help us maintain a healthy long-term financing structure while sustaining AI investment and supporting business growth. In summary, we will maintain AI investment intensity while balancing investor focus on profitability and return timelines. We believe operating cash flow will remain positive going forward, even with significant AI investment. **Risk disclosure and statement:**[**Dolphin Research Disclaimer and General Disclosure**](https://support.longbridge.global/topics/misc/dolphin-disclaimer) ### Related Stocks - [09888.HK - BIDU-SW](https://longbridge.com/en/quote/09888.HK.md) - [BIDU.US - Baidu](https://longbridge.com/en/quote/BIDU.US.md) - [89888.HK - BIDU-SWR](https://longbridge.com/en/quote/89888.HK.md) - [K3SD.SG - BIDU ADR US](https://longbridge.com/en/quote/K3SD.SG.md) - [GO.US - Grocery Outlet](https://longbridge.com/en/quote/GO.US.md) - [HBUD.SG - Baidu HK SDR 10to1](https://longbridge.com/en/quote/HBUD.SG.md) - [UBER.US - Uber Tech](https://longbridge.com/en/quote/UBER.US.md) - [09618.HK - JD-SW](https://longbridge.com/en/quote/09618.HK.md) - [89618.HK - JD-SWR](https://longbridge.com/en/quote/89618.HK.md) - [HJDD.SG - JD HK SDR 10to1](https://longbridge.com/en/quote/HJDD.SG.md) --- > **Disclaimer**: This article is for reference only and does not constitute any investment advice.