In the first quarter of 2026, Baidu delivered a financial report that appeared unremarkable at first glance but revealed significant structural changes.

Baidu's main business revenue reached 26 billion yuan, up 2% year-on-year and roughly flat quarter-on-quarter. Based solely on revenue figures, Baidu still resembles an internet company struggling to pivot from its old growth trajectory: advertising under pressure, search reaching its peak, and the mobile ecosystem unable to deliver high-growth stories to the capital markets.

However, the narrative changes entirely when viewed from a different perspective.

AI core business revenue hit 13.6 billion yuan, surging 49% year-on-year and accounting for 52% of the main business—the first time this share has exceeded 50% historically. Cloud infrastructure revenue reached 8.8 billion yuan, up 79% year-on-year, while GPU cloud business achieved an astonishing 184% growth rate.

In the fourth year of the large model wave, AI cloud infrastructure has quietly become Baidu's first "ballast stone" to break free from the search advertising cycle.

01 From 4.2 Billion to 8.8 Billion: The New Revenue Engine Accelerates

Baidu began disclosing cloud infrastructure revenue separately in Q3 2025. Combining Q1 financial data with the 2025 full-year total revenue of 19.8 billion yuan, the revenue trajectory over the past five quarters is as follows:

The 4.2 billion yuan in Q3 2025 marked a cyclical low, rebounding strongly to 5.8 billion yuan in Q4 and leaping to 8.8 billion yuan in Q1. The sequential increase expanded from 1.6 billion yuan to 3 billion yuan. For a business line approaching 10 billion yuan in a single quarter, a 3 billion yuan increment signals a comprehensive acceleration in growth momentum.

Two critical signals should not be overlooked.

First, sustained accelerating revenue growth.

The 1.6 billion yuan increase from Q3 to Q4 2025 indicates that enterprise AI computing demand has transitioned from early "trial exploration" to "essential procurement." The additional 3 billion yuan sequential growth in Q1 proves this is not a one-time spike but reflects continuously accelerating demand.

Second, GPU cloud has become the absolute primary growth engine.

GPU cloud revenue soared 184% year-on-year in Q1, far exceeding the overall 79% growth rate. This suggests a qualitative shift in cloud infrastructure revenue composition, with increasing emphasis on GPU cloud.

It should be noted that Baidu's cloud infrastructure growth stems from synergistic revenue generation across computing power, models, platforms, and applications—a fundamental distinction from pure model companies.

Pure model companies derive revenue elasticity from API call volumes, subscription users, and enterprise licensing. Baidu's AI revenue elasticity may come from a longer value chain: customers first purchase GPU cloud services, then access model platforms, develop industry applications, and finally integrate them into business systems.

If this chain operates smoothly, Baidu will no longer sell isolated AI capabilities but complete enterprise AI production systems.

02 Transition Between Old and New Growth Drivers: Traditional Businesses Remain Under Pressure

The most critical aspect of the Q1 financial report is not "AI revenue surge" but "AI growth offsetting traditional business contraction."

Advertising determines Baidu's revenue floor, while AI cloud infrastructure defines its ceiling. Correspondingly, Baidu's online marketing services (advertising) revenue reached only 12.6 billion yuan in Q1, down 22% year-on-year and 17% quarter-on-quarter.

To draw an analogy: Baidu's new engine is running at full throttle, but the old engine continues to decelerate, preventing the entire vehicle from achieving full acceleration.

This transition between old and new growth drivers signals an impending restructuring of Baidu's valuation logic.

Previously, capital markets evaluated Baidu based on "the stability of its search advertising foundation and whether iQIYI dragged down profits." Starting from Q1 2026, the core question in valuation models has shifted to: Can Baidu cultivate cloud infrastructure into a new revenue pillar supporting its entire business?

The financial report provides compelling preliminary data, but the final answer requires time to verify.

03 Market Share Analysis: Leading in Segments, Catching Up Overall

In China's AI cloud market, Baidu is not the dominant player.

Omdia reports that China's AI cloud market reached 56.7 billion yuan in 2025, with infrastructure accounting for 69% and model services 31%. Alibaba Cloud maintained its lead with a 38.1% share, followed by Volcano Engine (20.4%), Baidu Smart Cloud in third, and Tencent Cloud and China Telecom Cloud trailing behind.

Merely stating "Baidu also has AI cloud" fails to create sufficient differentiation.

In the underlying IaaS market, Baidu competes against Alibaba Cloud's scale, Huawei Cloud's government and enterprise foundation, Volcano Engine's recommendation systems and large model engineering capabilities, and Tencent Cloud's industrial internet resources... Baidu's real competitive edge lies not in selling cloud servers alone but in delivering full-stack AI solutions.

This advantage stems from several combined factors:

First, long-term accumulation of large model capabilities.

The ERNIE large model did not emerge overnight but builds on decades of experience in search, knowledge graphs, NLP, recommendation systems, and the PaddlePaddle ecosystem.

Second, enterprise AI engineering platforms.

The Qianfan platform's value extends beyond model invocation—it helps enterprises with model selection, fine-tuning, evaluation, deployment, and maintenance, serving as the critical bridge from "AI experimentation" to "production-grade AI."

Third, real-world industry scenarios.

Government, finance, energy, manufacturing, customer service, and marketing scenarios demonstrate that AI cloud is not abstract computing power but must integrate with business processes.

Fourth, search and advertising monetization experience.

Growth in AI-native marketing revenue essentially shows Baidu attempting to reintegrate AI into its core monetization framework. While overall advertising declines, AI's transformation of marketing efficiency may help stabilize traditional business revenue.

04 Business Model Evolution: From "Selling Computing Power" to "Selling Subscriptions"?

The 184% year-on-year growth in GPU cloud represents Baidu's most eye-catching Q1 figure.

If GPU cloud growth primarily comes from short-term project deliveries, high growth rates may prove volatile. However, if it stems from annual subscriptions, long-term computing commitments, and enterprise production system bindings, Baidu's cloud infrastructure business model is undergoing a fundamental shift.

Capital markets focus not just on revenue scale but on three key metrics:

First, revenue predictability.

Higher subscription revenue improves visibility into future quarterly revenues, making valuation models easier to switch from project-based to recurring revenue.

Second, customer retention rates.

The core challenge for AI cloud is not acquiring customers but whether they continue using services after pilot projects. Limited value exists if customers make one-time purchases for model training. However, if customers deploy core business systems on Baidu's GPU cloud, the value proposition changes completely.

Third, gross margin improvement potential.

GPU cloud requires heavy upfront investment, but marginal returns improve significantly once utilization rates rise. Conversely, insufficient utilization quickly erodes profits through depreciation, bandwidth, power, and maintenance costs.

Compared to "184% year-on-year growth," these metrics better determine whether Baidu's GPU cloud represents a temporary surge or sustainable long-term growth.

05 Hidden Concerns Behind Prosperity: Free Cash Flow Still "Bleeding"

The paradox of AI cloud infrastructure is that faster growth often increases short-term cash flow pressure.

The reason is straightforward.

GPUs must be purchased first, smart computing centers built, networks and storage expanded, and engineering teams assembled. Especially as large model competition enters production stages, cloud providers must pre-lock computing resources, leading all global cloud giants into a new capital expenditure expansion cycle.

For example, Alibaba's cloud intelligence group revenue reached 41.6 billion yuan in Q1 2026, up 38% year-on-year. Prior to this, Alibaba announced "380 billion yuan in cloud and AI infrastructure investment over three years."

Baidu finds itself in a similar cycle.

Q1 operating cash flow remained positive (approximately 2.7 billion yuan), but free cash flow remained negative (approximately -3.2 billion yuan) due to heavy capital expenditures. Full-year 2025 free cash flow was -15.1 billion yuan (consolidated), with capital expenditures of 12.1 billion yuan—even excluding iQIYI's impact, Baidu remains in a growth-prioritized, upfront-investment phase for cloud infrastructure.

In the coming quarters, the following three dimensions will prove more critical than mere revenue figures:

1. Growth resilience: Can Q2 and Q3 maintain over 20% sequential growth? Did Q1's surge partially exhaust concentrated delivery demands?

2. Revenue quality: Can subscription customer retention rates and GPU cloud's revenue share continue rising?

3. Cash generation capability: Can capital expenditure pace stabilize? When will free cash flow truly turn positive?

Answering these questions determines whether Baidu's AI can evolve from a "high-growth story" to a "high-quality growth story."

06 Kunlun Core: Not Just a Valuation Bonus, But Baidu's "Strategic Trump Card"

The spin-off and potential IPO of Kunlun Core represent another major catalyst in Baidu's valuation model.

If Baidu's cloud infrastructure relies heavily on external GPUs, faster growth increases dependence on upstream chip supply. Premium GPU pricing, supply cycles, export controls, and cluster construction costs all affect Baidu's future profit elasticity.

Should Kunlun Core achieve scaling deployment (large-scale deployment) in training, inference, search, recommendation, and industry model scenarios, Baidu could establish a closed loop: self-developed chips reduce computing costs, cloud infrastructure improves chip utilization, large models provide Adaptation scenario (scenario adaptation), the Qianfan platform handles enterprise delivery, and industry applications feed back data and demands.

This represents Baidu's ideal AI infrastructure flywheel.

However, for AI chips to meaningfully contribute to profitability, at least four hurdles must be cleared:

First, software ecosystem.

Enterprises don't migrate to domestic chips solely for nationalistic reasons. Development tools, operator libraries, framework compatibility, and operational experience determine migration decisions.

Second, model adaptation.

Chips must serve models, not the other way around. The adaptation efficiency between ERNIE, PaddlePaddle, Qianfan, and Kunlun Core determines whether Baidu's full-stack closed loop holds.

Third, comprehensive cost-effectiveness.

Single-card performance alone is insufficient. Cluster efficiency, energy consumption, stability, failure rates, and total cost of ownership matter.

Fourth, supply capacity.

AI cloud requires scalable supply, not laboratory prototypes. Stable mass production and continuous iteration capabilities are industrialization keys.

Kunlun Core's current accurate positioning: Not Baidu's already-realized second growth curve, but a potential lever for transitioning Baidu's AI from "business growth" to "profit optimization."

In the short term, Baidu Smart Cloud relies on market demand and GPU cloud growth. In the medium term, success depends on model platforms and industry delivery enhancing customer stickiness. In the long term, Kunlun Core's ability to reduce costs and strengthen supply chain security will determine how long Baidu can sustain its full-stack AI infrastructure narrative.

07 From Search Company to AI Infrastructure Provider

Baidu's historical valuation discount essentially stemmed from two issues:

First, search advertising growth had peaked, and markets were unwilling to assign high multiples to traditional advertising businesses.

Second, the AI narrative was too grand but lacked clear short-term financial validation.

The Q1 financial report begins to partially correct the second issue.

Over the next 4-6 quarters, Baidu's strategic objectives may become more focused: converting numerical growth rates into absolute market share advantages; transforming high revenue volumes into healthy cash flow and gross profits; and converting full-stack technology stacks into real switching costs that enterprise clients cannot easily abandon.

If achieved, Baidu will shed its label as "a search company struggling to transform" and emerge as China's indispensable, must-revalue superplayer in AI infrastructure.