This earnings report marks a key turning point in Baidu’s business development. Baidu’s business structure is now undergoing a significant transformation. In the first quarter, revenue from Baidu’s core AI-related businesses — including AI cloud infrastructure, autonomous driving, and various AI applications — reached RMB 13.6 billion. For the first time, AI-related revenue accounted for more than 52% of Baidu’s core operating revenue, representing a sharp 9 percentage-point increase from the previous quarter. Management described this milestone as a critical turning point and stated that AI is expected to create significantly greater value for Baidu over the coming quarters. In sharp contrast, Baidu’s traditional online marketing services — once the company’s core cash-generating business — continued to decline. Revenue from online marketing fell 22% year-on-year to RMB 12.6 billion, with its contribution to overall revenue dropping to just 48%.

For Baidu, the first quarter represents a defining moment in the company’s evolution. AI-driven businesses now account for the majority of Baidu’s core revenue for the first time, officially surpassing traditional businesses to become the company’s largest revenue source. This means AI has now become the dominant driver within Baidu’s overall revenue structure.

Baidu’s AI business currently consists of three major segments: AI cloud infrastructure, AI applications, and autonomous driving.Breaking down the numbers further, AI cloud infrastructure remains the primary growth engine. In the first quarter, Baidu’s AI cloud infrastructure revenue reached RMB 8.8 billion

, representing year-on-year growth of 79%. Within that segment, high-margin GPU cloud revenue surged 184% year-on-year.

This explosive growth has been driven by a concentrated wave of industry-wide AI inference demand. Enterprise demand for AI infrastructure has strengthened significantly, particularly on the inference side, signaling that customers are moving beyond pure model-training phases and rapidly deploying AI across broader real-world business applications.

Baidu also revealed that its Kunlun P800 AI chip has completed large-scale validation and has already been deployed across multiple 10,000-GPU clusters. Its fully domestic AI cluster has also completed training for major versions of the ERNIE 5.1 large language model. In addition, Baidu’s Tianchi 256-GPU supernode powered by Kunlun chips has officially gone live and is scheduled for commercial release in June. Compared with the previous generation, throughput performance has improved by 25%, while inference efficiency has increased by 50%. The platform is compatible with major AI models including ERNIE, DeepSeek, GLM, and MiniMax. Baidu also appears increasingly interested in accelerating an independent listing of its AI chip business, allowing markets to separately evaluate and unlock the value of the segment.

Compared with the strong momentum in AI cloud infrastructure, Baidu’s AI application business remains relatively subdued. The earnings report showed that AI application revenue totaled approximately RMB 2.5 billion during the first quarter, largely flat compared with the same period last year. AI applications targeting enterprise clients are widely viewed as Baidu’s key strategic opportunity in the broader AI competition landscape.

Market Analysis：

Global AI commercialization remains in a very early stage, with business models still evolving rapidly. Currently, token-based pricing remains the dominant industry model, where users effectively pay for access to foundational model capabilities. In the future, however, Baidu believes AI applications and AI agents will increasingly be able to complete far more complex real-world tasks. Ultimately, users may begin paying directly for AI agents or AI applications themselves, rather than simply paying for token consumption. Baidu believes this future market could become significantly larger than today’s token-based AI economy.