Executive Summary / Key Takeaways

• Baidu is undergoing a profound AI-first transformation, strategically pivoting from a traditional internet services provider to a leader in generative AI, cloud computing, and autonomous driving.
• The AI Cloud segment is demonstrating robust growth, with revenue increasing 27% year-over-year in Q2 2025 and 42% in Q1 2025, driven by strong demand for its cost-effective, full-stack AI solutions.
• Apollo Go, Baidu's autonomous ride-hailing service, is rapidly scaling globally through strategic partnerships with Uber and Lyft , and its RT6 vehicle boasts industry-leading cost efficiency, positioning it for significant future profitability.
• While heavy investments in AI transformation are creating near-term pressure on online marketing revenue and overall margins, particularly in Q3 2025, management is confident these are necessary trade-offs for substantial long-term monetization opportunities in AI-powered search and new AI-native applications.
• Baidu's proprietary ERNIE models and unique 4-layer AI architecture provide a significant technological moat, enabling superior performance, cost reduction, and competitive differentiation in China's evolving AI ecosystem.

The AI-First Transformation: A New Era for Baidu

Baidu, Inc., a company with a two-decade history as a public entity, stands at a pivotal juncture, aggressively embracing the artificial intelligence mega-trend to redefine its core businesses and unlock new growth frontiers. The company's strategic shift is rooted in an "application-driven approach" to AI innovation, aiming to translate cutting-edge research into tangible, real-world value across its diverse portfolio. This transformation is not merely incremental; it represents a fundamental re-architecture of Baidu's offerings, from its foundational search engine to its burgeoning AI cloud and autonomous driving segments.

The broader industry landscape in China is rapidly evolving towards AI-centric computing, creating immense opportunities but also intensifying competition. Baidu's long-standing commitment to technology and innovation, coupled with substantial R&D investments, positions it as a front-runner in this dynamic environment. The company's overarching strategy is to leverage its proprietary AI capabilities to enhance user experience, drive operational efficiency, and establish new monetization pathways, even if it entails near-term financial pressures.

Technological Differentiators: Baidu's AI Moat

Baidu's competitive edge is deeply embedded in its advanced AI technology, particularly its ERNIE foundation models and unique 4-layer end-to-end AI architecture. These technological differentiators provide quantifiable benefits that underpin the company's strategic initiatives and future growth.

The ERNIE models are at the heart of Baidu's AI ecosystem. The company has rapidly iterated these models, with ERNIE 4.5 and ERNIE X1, its first multimodal and reasoning models respectively, launched in March 2025. These were swiftly followed by upgraded Turbo versions in April 2025, which delivered enhanced performance at dramatically lower pricing. Specifically, ERNIE 4.5 Turbo is priced 80% lower than its predecessor, ERNIE 4.1, and ERNIE X1 Turbo is half the price of ERNIE X1. This aggressive cost reduction is a direct result of Baidu's full-stack AI capabilities. The models' prowess is evident in applications like digital human technology, which has achieved new levels of realism and capabilities, matching or even exceeding human performance in certain scenarios. A notable example is a 7-hour live stream powered by ERNIE 4.5 Turbo, which generated tens of millions in Gross Merchandise Value (GMV) by creating complete scripts, dialogue, tone, and action cues. Furthermore, ERNIE X1.1 has demonstrated performance on par with top-tier global models like GPT-5 and Gemini 2.5 Pro, while surpassing Chinese competitor DeepSeek R1-0528 in overall performance. Baidu's decision to open-source its ERNIE 4.5 series by June 30, 2025, aims to further expand market awareness and adoption, democratizing AI development.

The 4-layer end-to-end AI architecture is a core competitive advantage, enabling holistic optimization across infrastructure, framework, models, and applications. In Q2 2025, Baidu achieved a critical system engineering breakthrough by completing the large-scale stable deployment of a prefiled detailed separation architecture. This significantly improves inference concurrency and resource utilization while substantially reducing inference costs. The company has also demonstrated remarkable efficiency in its GPU cluster management, achieving a 99.5% valid LLM training time in Q3 2024 and minimizing throughput loss to within 5% when combining GPUs from different vendors. These efficiencies allow Baidu to deliver better performance at lower cost, reinforcing its position as a top-tier cloud provider in the AI era.

Qianfan, Baidu's MaaS (Model as a Service) platform, further extends this technological lead. It features a comprehensive model library, including the newly open-sourced ERNIE 4.5 series and various third-party models, offering superior price performance due to enhanced stability, higher concurrency, and lower inference costs. Qianfan's toolchains are among the most comprehensive, with industry-leading reinforcement learning and post-training tools, and new features like one-click distillation. These enhancements simplify AI application development, making it easier for enterprises to adopt advanced AI technology.

In autonomous driving, Baidu's Apollo Go leverages its AI expertise. The RT6, Apollo Go's sixth-generation autonomous vehicle, is the world's first and only purpose-built mass-produced Level 4 autonomous vehicle, boasting a unit cost below $30,000. This cost advantage, coupled with over 200 million autonomous kilometers accumulated by its fleets and an outstanding safety record, provides Apollo Go with a significant differentiation in the global robotaxi market.

For investors, these technological advancements are critical. They form a robust competitive moat, driving cost efficiencies, enabling superior product performance, and positioning Baidu for sustained market leadership and long-term growth across its key segments. The ability to continuously reduce inference costs and enhance model capabilities directly translates into a stronger value proposition for both internal products and external enterprise clients.

Financial Performance and Operational Momentum

Baidu's financial performance in recent quarters reflects a strategic pivot, with robust growth in its AI-driven segments offsetting headwinds in its traditional online marketing business. In Q2 2025, Baidu Core's total revenue was RMB 26.3 billion, a 2% year-over-year decrease. However, a significant shift is underway, as Baidu Core's non-online marketing revenue exceeded RMB 10 billion for the first time, growing 34% year-over-year. This performance was primarily driven by the accelerating AI Cloud business.

The AI Cloud segment has been a standout performer, with revenue reaching RMB 6.5 billion in Q2 2025, a 27% year-over-year increase. For the first half of 2025, AI Cloud revenue surged 34% year-over-year, accelerating from the low teens growth in the first half of 2024. This growth is fueled by strong demand for Baidu's cost-effective, end-to-end AI products and solutions. Notably, subscription-based revenue within the enterprise cloud, which constitutes the vast majority of AI Cloud revenue, is growing at a solid pace, indicating a healthier and more sustainable revenue structure. The segment also achieved year-over-year growth in non-GAAP operating profit in Q2 2025, maintaining a healthy margin, which was in the "teens" in Q1 2025.

In contrast, Baidu Core's online marketing revenue faced "near-term headwinds," decreasing 15% year-over-year to RMB 16.2 billion in Q2 2025. This decline is largely attributable to the ongoing AI transformation of Baidu Search and a challenging macroeconomic environment. However, early monetization efforts within AI Search are showing promise. Revenue generated by agents for advertisers grew 50% quarter-over-quarter in Q2 2025, contributing 13% of Baidu Core's online marketing revenue, up from 9% in Q1 2025. Digital human-generated revenue also increased by 55% quarter-over-quarter, contributing 3% of online marketing revenue.

Apollo Go, Baidu's autonomous driving service, is demonstrating remarkable operational scaling. In Q2 2025, Apollo Go provided over 2.2 million fully driverless rides to the public, a substantial 148% year-over-year increase, marking its strongest quarterly growth in two years. As of August 2025, cumulative rides provided to the public surpassed 14 million. The service's global footprint now spans 16 cities as of June 2025, with fleets accumulating over 200 million autonomous kilometers.

Liquidity remains robust, with cash, cash equivalents, restricted cash, and short-term investments totaling RMB 124.2 billion as of June 30, 2025. The net cash position for Baidu was RMB 155.1 billion. However, free cash flow was negative RMB 4.7 billion in Q2 2025, primarily due to increased investment in AI business. This reflects the company's aggressive capital allocation towards its AI-first strategy, including a CNY4.4 billion offering of CNY-denominated senior notes in September 2025 and a $2 billion bond issuance in March 2025.

Strategic Initiatives and Future Outlook

Baidu's strategic initiatives are tightly integrated with its AI-first vision, aiming to capitalize on its technological leadership and expand its market reach.

The AI Search transformation is a top priority. Baidu is fundamentally revolutionizing search by replacing static hyperlinks with intelligent, structured, and multi-modal AI-generated responses. By July 2025, AI-generated content reached over 64% of mobile search result pages, up from 35% in April. This transformation has reached over 90% of Baidu App's monthly active users, with over 60% of such pages featuring rich media elements. While this aggressive transformation is expected to "weigh on our revenue" in the short term, particularly in Q3 2025, management believes it will "unlock exciting commercial possibilities and upside is substantial" over time, potentially shifting monetization from CPC to CPS.

Apollo Go's global expansion is accelerating. Following successful validation of its business model in China's low-fare markets, Baidu is strategically entering higher-fare international markets. Key partnerships with Uber and Lyft will deploy thousands of Apollo Go vehicles across Asia, the Middle East, and Europe, leveraging partners' market presence for efficient scaling. Apollo Go's leadership in the right-hand drive robotaxi market, demonstrated by its progress in Hong Kong, provides a significant competitive advantage. Management anticipates "accelerating growth in ride volumes with our global operational fleet size multiplying" and foresees a "clear path to profitability" as hardware and labor costs decline.

In the AI Cloud business, Baidu is focused on strengthening its MaaS platform, Qianfan, and expanding its customer base. The company is seeing strong demand from diverse sectors, including internet companies, utilities, financial services, and emerging embodied AI industries. Baidu's full-stack AI capabilities and cost-effective solutions are driving increased spending from both top-tier and mid-tier enterprise clients. The company is also driving internal productivity through Comate, its AI coding assistant, which contributed to generating over 45% of new code in July 2025. Miaoda, Baidu's no-code platform, has already seen approximately 200,000 applications created by users, democratizing AI development.

Baidu is committed to increasing its AI investments in 2025 to solidify its AI foundation and prepare for future growth. While this will exert "considerable pressure" on near-term revenue and margins, the company expects a "gradual recovery in profitability over time" as the core advertising business stabilizes and non-advertising businesses expand their revenue share and improve profitability.

Competitive Positioning

Baidu operates in a highly competitive landscape, facing formidable rivals across its segments. In AI and cloud services, Baidu competes with Chinese tech giants like Alibaba Group Holding Limited and Tencent Holdings Limited , as well as emerging AI startups like DeepSeek. Baidu's unique 4-layer AI architecture and ERNIE models provide a distinct advantage, enabling superior performance and cost-effectiveness in AI infrastructure and MaaS platforms. While Alibaba (BABA) and Tencent (TCEHY) boast broader ecosystems and larger user bases, Baidu's specialized AI expertise, particularly in search and content generation, allows it to differentiate through notably faster data analysis for targeted advertising and more efficient content personalization. Baidu's ERNIE X1.1, performing on par with global leaders and surpassing DeepSeek R1-0528, underscores its strong model capabilities in this intense environment.

In autonomous driving, Apollo Go is a global leader, particularly in the challenging right-hand drive markets. Its RT6 vehicle, with a unit cost below $30,000, offers a significant cost advantage over competitors. The strategic partnerships with global mobility platforms like Uber (UBER) and Lyft (LYFT) are crucial for rapidly scaling operations internationally, allowing Baidu to compete effectively against other autonomous driving players such as Waymo and Tesla (TSLA), who are also pursuing global deployments.

Baidu's online marketing business faces competition from other digital advertising platforms and new content ecosystems like ByteDance's Douyin and Tencent's WeChat. The AI transformation of Baidu Search is a direct response to these competitive pressures, aiming to redefine user engagement and unlock new monetization avenues that go beyond traditional search.

A key vulnerability for Baidu is the geopolitical environment, particularly GPU chip restrictions. However, the company is strategically addressing this by building flexible AI architectures that maximize GPU utilization and support a variety of chips, including domestic ones. This focus on a self-sufficient supply chain and homegrown software stacks is critical for sustainable innovation in China's AI ecosystem.

Conclusion

Baidu is in the midst of an ambitious AI-first revolution, strategically reorienting its business towards high-growth areas of artificial intelligence, cloud computing, and autonomous driving. While the transformation of its core online marketing business is creating near-term revenue and margin pressures, particularly in Q3 2025, this is a calculated trade-off for substantial long-term value creation. The company's proprietary ERNIE models and unique 4-layer AI architecture provide a formidable technological moat, driving cost efficiencies and superior performance that differentiate it from competitors.

The robust growth of its AI Cloud segment and the rapid global expansion of Apollo Go, bolstered by strategic partnerships and cost-effective technology, are clear indicators of Baidu's future potential. Investors should recognize that the current period of heavy AI investment and early monetization is foundational to unlocking new revenue streams and strengthening Baidu's market leadership. With a strong balance sheet and a clear technological roadmap, Baidu is positioning itself to be a dominant force in China's AI-powered future, offering a compelling investment thesis for those with a long-term perspective.