According to the latest data from Omdia, the cloud infrastructure services market in Mainland China reached $13.4 billion in the third quarter of 2025, representing a 24% year-over-year increase. This marks the second consecutive quarter of growth exceeding 20%. Sustained demand for artificial intelligence is not only accelerating the adoption of cloud services but also significantly driving the expansion of core cloud infrastructure, speeding up the shift of cloud resource consumption toward production-grade workloads. Against this backdrop, leading cloud providers are continuing to strengthen their AI capabilities, placing greater emphasis on productizing AI models and defining functional roles, while enhancing the underlying toolchains and platform capabilities that support AI agent platforms.
In the third quarter of 2025, Alibaba Cloud, Huawei Cloud, and Tencent Cloud held market shares of 36%, 16%, and 9%, respectively. As enterprises gradually transition from early-stage AI experimentation to large-scale implementation, AI is increasingly becoming the primary driver of new demand for core cloud infrastructure services, boosting consumption of computing, storage, and database resources.
Rachel Brindley, Senior Director at Omdia, commented, "As AI adoption deepens, leading cloud providers are not only continuously improving their foundational model capabilities but also embedding these models into more comprehensive platform architectures, making them functional modules within the platform ecosystem. Clear delineation of model functions and deep integration with platforms are emerging as key trends in the industry's evolution."
During the quarter, Alibaba Cloud expanded its multimodal model portfolio; Huawei Cloud accelerated AI adoption in key industry scenarios through jointly released industry models; and Tencent Cloud further clarified the positioning and division of labor for its models within the HY 2.0 Think and Instruct product lines, based on its Hunyuan 2.0 system.
As AI deployment enters the phase of large-scale production, enterprise focus is shifting toward platform-level reliability and operational stability. Yi Zhang, Senior Analyst at Omdia, noted, "Single-model performance alone can no longer meet enterprise needs in real-world business scenarios. The core challenge in scaling AI projects lies in orchestrating models, data, tools, and workflows within complex systems to achieve reusability, operational manageability, and commercial viability."
In this context, the second half of 2025 saw leading cloud providers converge more noticeably in their AI agent platform strategies. Alibaba Cloud, Huawei Cloud, and Tencent Cloud are all advancing their agent development platforms from a capability demonstration phase toward an engineering-oriented, operationally scalable stage, with systematic upgrades in areas such as Model Context Protocol (MCP), workflow orchestration, knowledge bases, and plugin frameworks, laying the groundwork for the commercialization of generative AI.
Additionally, partner-driven cloud revenue accounted for 25% of the market in the third quarter of 2025. As ecosystem collaboration plays an increasingly vital role in translating AI technical capabilities into tangible business outcomes, the contribution of partners is expected to continue growing.
Alibaba Cloud increased its share of the Chinese cloud infrastructure services market to 36% in the third quarter of 2025, maintaining its leading position. The company has reported triple-digit year-over-year growth in AI-related revenue for nine consecutive quarters, driven by the ongoing expansion of enterprise applications and strategic AI collaborations with key industry clients such as Marriott, GAC Group, L'Oréal China, and Haier.
At the product and platform level, Alibaba Cloud has continued to enhance application capabilities and platform usability around its Qwen model family. In October, it released nine Qwen3-VL multimodal models, improving video understanding and spatial perception capabilities. This was followed by the launch of the Wan2.6 video generation series, with Wan2.6-R2V being China’s first reference-based video generation model, significantly improving content consistency and controllability and accelerating the evolution of video generation toward production-grade scenarios. In December, the company introduced AgentRun, built on a function compute architecture, to support the deployment and operation of AI agents in production environments, further strengthening capabilities for engineering and scaling AI agents.
In terms of global infrastructure, Alibaba Cloud continued to expand its overseas presence, launching a second data center in Dubai in October to enhance regional service capabilities and global coverage.
Huawei Cloud maintained its position as the second-largest player in China’s cloud infrastructure services market in the third quarter of 2025, achieving 14% year-over-year revenue growth and holding a 16% market share. During the quarter, Huawei Cloud continued to advance AI implementation in industry scenarios, jointly releasing the "Sky Predictor Large Model 1.0" with China Southern Airlines. This model, built on Huawei Cloud Stack and AI-driven predictive capabilities, reflects Huawei Cloud’s efforts to promote AI adoption in key sectors such as aviation and manufacturing.
In October, Huawei Cloud upgraded its AI Agent development platform, systematically enhancing core capabilities including MCP services, workflow orchestration, knowledge base systems, and plugin frameworks. It also added over 80 official pre-configured MCP tools to improve agent development efficiency and collaboration. For its overseas expansion, Huawei Cloud announced that a third availability zone in its Ireland region will commence operations in early 2026 to support growing cloud computing and AI demand in the European market.
Tencent Cloud accounted for approximately 9% of China’s cloud infrastructure services market in the third quarter of 2025. Its growth remains somewhat constrained due to limited supply of advanced AI computing resources. Tencent Cloud is currently optimizing the allocation of its existing AI computing resources, balancing support for customer AI demands while upgrading its internal operational systems and proprietary AI products.
In November, Tencent Cloud launched HY 2.0 Think and HY 2.0 Instruct, enhancing complex reasoning and agent workflow capabilities to better support enterprise AI deployments requiring stability and scalability. In December, it upgraded its Agent Development Platform (ADP), strengthening core modules such as MCP plugins, workflow orchestration, knowledge management, and tool systems. Additionally, the company began commercial billing for RAG-based model capabilities within the ADP, marking a key step toward scalable, production-grade agent deployment.
Omdia defines cloud infrastructure services as a collection of services hosted by third-party providers and delivered over the internet, including Bare Metal as a Service (BMaaS), Infrastructure as a Service (IaaS), Platform as a Service (PaaS), Container as a Service (CaaS), and Serverless computing.