Baidu, Inc. Q3 FY2025 earnings call

• AI Cloud: Continued strong growth with subscription-based revenue from AI accelerator infrastructure surging 128% yoy. Areas central to AI achieved fastest growth. Client portfolio improved with deeper collaboration across leading, mid-tier enterprises, and key verticals like embodied AI and automotive. • Apollo Go: Delivered over 3 million fully driverless operational rides in Q3, a 212% yoy growth. Global footprint expanded to 22 cities, with expansions in Europe, Middle East, and Hong Kong. • Mobile Ecosystem: Agents and digital humans drove strong growth. Combined revenue from agents and digital humans in Q3 was RMB 2.8 billion, up 262% yoy. Agents help advertisers get precise leads, and digital humans provide 24x7 AI-powered live streaming. • ERNIE 5.0: Unveiled at Baidu World, it drives capabilities in digital humans, FM agent, and AI search/cloud business, with focus on improving application performance in various scenarios.

Baidu Core reported total revenue of RMB 24.7 billion in Q3. AI Cloud revenue reached RMB 6.2 billion, increasing 21% year-over-year. AI Cloud Infra revenue in Q3 was RMB 4.2 billion, up 33% year-over-year. AI Applications generated revenue of RMB 2.6 billion. AI-native marketing services (agents and digital humans) reached RMB 2.8 billion in Q3, up 262% year-over-year, accounting for 18% of Baidu Core's online marketing revenue.

• Apollo Go to continue rapid global expansion, aiming for more markets with strong commercial potential while prioritizing safety. • AI-native marketing services expected to scale further, broadening adoption across verticals and deepening penetration with existing advertisers. • Expect margin improvement as AI business scales, with utilization increasing for AI Cloud Infra.

• Risks associated with forward-looking statements as actual results may differ from expectations. • Risks related to regulatory environments for autonomous driving operations in various global markets.

Q: About ERNIE 5.0 and how it drives digital humans and other applications, and future iterations and differentiation.

A: Robin Li stated ERNIE 5.0 is a native omni-model foundation model with world-class capabilities in omni-model understanding, creative writing, etc. It powers digital humans, FM agent, and AI search/cloud, with focus on improving application performance in areas where existing models fall short.

Q: About Cloud business growth moderation, demand changes, and outlook.

A: Dou Shen mentioned Cloud business grows well above industry average, with AI Cloud Infra revenue up 33% yoy. Subscription-based AI accelerator infrastructure revenue grew 128% yoy. Full-stack AI capabilities and strong demand for AI-centric cloud services support sustainable growth.

Q: About blurring line between search and chatbot, user behaviors, and competitive strategy.

A: Robin Li said Baidu leverages chatbot capabilities through ERNIE assistant (built-in with multi-round conversations) and ERNIE bot (stand-alone with experimental features). Future focus on multimodel real-time generative interactions.

Q: About growth outlook and profitability of AI-powered businesses.

A: Henry He explained AI-native views organize businesses into AI Cloud Infra, AI applications, and AI native marketing services. AI Cloud Infra sees growing demand, AI applications have broad portfolio with subscription models, and AI native marketing services drive second growth curve with increasing penetration.

Q: About Robotaxi business, global expansion, unit economics, and profitability.

A: Robin Li said Apollo Go has provided over 17 million cumulative rides, is market leader in China. Plans for global expansion, asset-light models, and expectation of improving unit economics as scale increases.

Q: About AI search monetization, advertiser/user feedback, and core advertising profitability.

A: Julius Rong Luo mentioned testing MCP in AI search, agents generating over RMB 25 million daily revenue, and testing digital human live streaming. Near-term focus on user experience improvement with long-term monetization potential.

Q: About asset impairment, CapEx plans, margin trajectory, and shareholder returns.

A: Henry He explained asset impairment due to outdated infrastructure, CapEx to continue high investment in AI, expectation of margin improvement as AI business scales, and ongoing review of shareholder return mechanisms including buybacks and dividends.