Baidu, Inc. (BIDU) Earnings

Management highlights

- Full-Stack AI Infrastructure - Kunlunxin proprietary AI chips maintained robust, broadening demand across industries, with growing market recognition for its performance, efficiency, and scalability. The product added compatibility with new versions of major Chinese foundation models, improved inference throughput and compute efficiency, and maintains a clear product roadmap including the M100 for large-scale inference and upcoming M300. The listing process for Kunlunxin remains ongoing. - AI Cloud Infra delivered its fourth consecutive quarter of above-industry growth, driven by strong broad-based demand for AI computing across training and inference workloads, increasing spending from existing clients, rapid growth in new customers across internet, embodied AI, autonomous driving, smartphones, and financial services. Embodied AI revenue grew approximately 6-fold YoY, and profit and margins expanded YoY alongside revenue growth. - Qianfan MaaS platform enhanced inference optimization to deliver higher throughput, lower latency, and reduced inference costs for customers. - Foundation Model Development - Baidu reorganized its foundation model teams into focused groups for foundation models and applications, and recently added new top AI talent to accelerate innovation. The company follows an application-driven development loop where model improvements are tested and refined via integration with Baidu's core AI applications to deliver tangible user and commercial value. - AI Applications - Digital human technology achieved lower costs and broader enterprise adoption, with existing clients scaling usage after successful pilots. Overseas expansion of the digital human platform launched last quarter has shown encouraging early momentum. - MiaoDa 3.0 natural language coding platform saw 67% monthly active user growth between March and June 2026, with expanding adoption across multiple industry verticals and improving user stickiness. - New AI agents including PhamoAgent (enterprise operational optimization) and DouMate (general productivity, with a new enterprise version launched this quarter) gained early commercial traction. AI-powered features at Baidu Wenku and Baidu Drive saw 27.4% YoY growth in AI daily active user penetration in June. - AI search improved answer quality, reduced hallucination rates, and integrated with Ernie Assistant to enable interactive multi-turn conversations. Ernie Assistant grew 83% YoY in daily active users in June, with daily conversation volumes more than tripling YoY. The company deliberately delayed near-term monetization of AI search to prioritize product and user experience improvements, putting near-term pressure on online marketing revenue. - Autonomous Driving (ApolloGo) - ApolloGo delivered ~1 million fully driverless rides in Q2 2026, with cumulative public rides exceeding 23 million as of June 2026. Ride volume was temporarily impacted by regulatory-related operational adjustments in some domestic Chinese cities, with operations resuming in August 2026. The company recorded an industry-leading safety rate of one airbag deployment per 14.4 million kilometers driven. - Key global milestones: Secured Hong Kong's first fully driverless testing permit (first global robotaxi operator to test fully driverless in a right-hand drive, left-hand traffic market), launched open-road testing in London with Uber and Lyft, launched fully driverless commercial operations in Dubai (the largest-scale fully driverless service in the city), began testing in Switzerland, and signed an MOU for expansion in Kazakhstan.

Guidance

- AI Cloud Infra is expected to maintain strong growth in the second half of 2026, with potential for further acceleration of growth. - Online marketing revenue is expected to remain under pressure in the second half of 2026 due to ongoing AI search transformation and delayed monetization. - ApolloGo ride volume is expected to regain growth momentum in coming quarters as operations restart in affected domestic cities and expansion continues globally. Management's H2 2026 priorities for ApolloGo are improving safety standards, advancing global expansion, scaling fleet and ride volumes, and reaching unit economic break-even in more markets. - The voluntary conversion of Baidu's Hong Kong listing to dual primary status is expected to become effective in 2026, pending shareholder and Hong Kong Stock Exchange approval. - Long-term, AI Cloud Infra margins have meaningful room for expansion driven by mix shift to higher-margin GPU cloud, growing MaaS profitability, and end-to-end cost advantages from full-stack AI and self-developed chips.

Segment performance

Baidu reported total Q2 2026 revenue of RMB 31.3 billion, a 2% decrease quarter-over-quarter (QoQ) and a 4% decrease year-over-year (YoY). - Baidu General Business: Total revenue of RMB 25.2 billion, down 3% QoQ and 4% YoY. Baidu Core AI-powered business contributed RMB 12.5 billion, accounting for 50% of total Baidu General Business revenue. - AI Cloud Infra: Revenue grew 50% YoY, outpacing broader industry growth. Within AI Cloud Infra, GPU cloud revenue grew 283% YoY, accelerating from 184% YoY growth in Q1 2026, and accounts for a growing share of AI Cloud Infra revenue. Qianfan MaaS platform revenue from external customer token usage grew more than 9-fold YoY. - iQiyi: Revenue of RMB 6.3 billion, up 1% QoQ and down 5% YoY. Other key financials: Operating income was RMB 3.0 billion with an operating margin of 10%; Non-GAAP operating income was RMB 3.8 billion with a non-GAAP operating margin of 12%. Net income attributable to Baidu was RMB 2.3 billion with a net margin of 7%; Non-GAAP net income attributable to Baidu was RMB 2.6 billion with a non-GAAP net margin of 8%. As of June 30, 2026, total cash and investments stood at RMB 283.1 billion.

Risks & headwinds

- Forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from current expectations, detailed in Baidu's prior regulatory filings. - Intense competition in AI search and evolving user information consumption patterns have increased competition for user attention and time, compounding near-term pressure on advertising revenue. - Global market supply of AI computing remains constrained, which could limit growth ability amid strong demand. - Autonomous ride-hailing expansion is subject to varying regulatory requirements across markets, and regulatory adjustments temporarily impacted Q2 2026 domestic ride volumes.

Analyst Q&A

• Q: How does Baidu position itself against rapidly evolving large foundation model competition, and what are its key priorities after adding new senior AI talent? /

  A: Robin Li noted that the foundation model space remains highly dynamic, with no player permanently leading across all capabilities. Long-term competitiveness relies on sustained, patient investment and an application-driven strategy, rather than chasing benchmark leadership alone. Baidu has over a decade of long-term AI investment that produced differentiated assets like Kunlunxin and ApolloGo, and remains committed to advancing Ernie. Baidu will focus development on capabilities that directly improve its core applications (such as AI search, digital humans, and agents), using a closed feedback loop to turn model improvements into better user experiences and commercial value.

• Q: What are the key growth drivers and long-term margin outlook for AI Cloud Infra? /

  A: Dou Shen stated that sustained growth will be supported by strong, growing structural demand for AI computing in China (particularly scaling inference workloads), a rapidly expanding customer base with increasing spending from existing clients, and broad-based demand across multiple industry verticals. Baidu's differentiated full-stack AI architecture across chips, infrastructure, models, and applications also positions it to capture more AI opportunities. Long-term margins will expand driven by the growing mix of higher-margin GPU cloud, improving profitability of fast-growing MaaS as it scales, and inherent cost advantages from self-developed chips and end-to-end optimization.

• Q: How should investors think about Baidu's operating margin trajectory, and how do you balance AI investment with profitability? /

  A: Henry He confirmed that while Baidu remains in an active AI investment cycle with unwavering commitment to long-term investment, the company invests with discipline and a strict focus on ROI. Most investments are driven by clear customer and internal demand, so they are put to work quickly and contribute to revenue relatively soon. Improving supply chain management also continues to boost capital efficiency. Long-term, AI Cloud margins will expand as it scales, and AI applications are expected to become increasingly meaningful contributors to overall profitability as adoption and monetization mature.

• Q: How does ApolloGo approach regulatory changes and the balance of domestic versus global expansion, and what is the outlook for overseas commercialization? /

  A: Robin Li noted that clearer, more systematic regulatory frameworks for autonomous driving around the world will ultimately support long-term industry growth by raising standards and building public trust, and ApolloGo has already contributed its expertise to China's first national mandatory automated driving safety standard. Baidu does not choose between domestic and global expansion, and adapts its expansion pace to each market's specific regulatory, demand, and market conditions. ApolloGo already has achieved meaningful early progress in global markets including Dubai, Hong Kong, and London, and the larger addressable market outside the U.S. and China, combined with stronger unit economics potential in higher-price overseas markets, creates substantial long-term opportunity.