China’s cloud growth hits 26% for third straight quarter as AI moves beyond experiments

Something changed in how China’s cloud market grows, and the numbers from Q4 2025 offer the clearest signal yet. Omdia puts the market at $14.7 billion for the quarter, up 26 percent year on year, which sounds like a strong growth story on its own. The part worth paying closer attention to is that this marks three consecutive quarters above 20 percent, and Omdia is not calling it a recovery. The firm describes something structural, where AI has stopped sitting as a discrete workload and started pulling demand through storage, networking, databases, and enterprise services all at once.

That distinction is easy to gloss over but hard to overstate. The first wave of generative AI spending was relatively contained. Companies bought model access, ran training jobs, called APIs, and measured inference costs. What Omdia’s latest data reflects is different. Enterprises have moved past testing into actual deployment, and that transition changes the infrastructure equation considerably. Systems built around AI models need orchestration, data pipelines, and operational controls that keep running whether or not anyone actively queries a model. The surrounding infrastructure expands even when model activity holds steady.

Agent-based software pushes that dynamic further. Earlier AI applications could switch on and off. Agents run persistently, maintain context, interact with multiple external systems, and generate steady backend load rather than usage spikes. The infrastructure demand they create follows more predictable patterns but requires significantly more from storage and workflow coordination services simultaneously.

Alibaba Cloud leads the market at 37 percent and builds through vertical integration, combining models, platforms, and tooling into a single stack. Its Qwen model family and Model Studio platform drove a sixfold jump in average daily token consumption for public APIs over a three-month window, suggesting that developer activity scales in step with infrastructure spending rather than running ahead of it.

Huawei Cloud holds roughly 17 percent and targets industry-specific deployment through its Industry AI Foundry, focusing on sectors like healthcare where domain knowledge shapes how AI actually gets used. Tencent Cloud sits at around 10 percent and routes agent interaction through messaging platforms, embedding AI into existing communication channels rather than asking users to adopt new tools.

Partner revenue hit 25 percent of the total market during the quarter, and Omdia expects that share to keep climbing. Cloud providers can supply the infrastructure, but turning AI into business outcomes inside specific industries requires expertise that hyperscalers simply do not carry in-house.