The Chinese public cloud market is undergoing a structural shift driven by artificial intelligence. According to the latest "China Public Cloud Services Market Tracker, Second Half of 2025" report from International Data Corporation (IDC), the year-on-year growth rate for the public cloud Infrastructure-as-a-Service (IaaS) market in the second half of 2025 recovered to 20.0%. The total value of the overall public cloud market (IaaS+PaaS+SaaS) for the half-year period surpassed 200 billion yuan for the first time. This growth is not merely a market recovery; it reflects a fundamental reshaping of the cloud computing industry's underlying logic by AI demand. Market recognition is shifting from "resource scale" to "full-stack AI capabilities," market share is rapidly consolidating towards vendors strong in both "computing power and large models," overseas expansion is emerging as a second growth curve, and demand divergence across different industries is set to widen further. IDC believes that over the next three to five years, competition in the public cloud market will no longer revolve around price wars and scale battles, but will instead be a comprehensive contest involving computing power, models, industry solutions, ecosystems, and global capabilities. The vendors that can genuinely evolve from "cloud providers" to "AI service providers" during this AI dividend period will dominate the next phase of the market landscape.
AI capabilities are reconfiguring market share, with large models enabling cloud providers to regain pricing power. In the traditional IaaS market, cloud server leasing long dominated, making price wars the main theme of competition. However, with the explosion of generative AI and large model demand, the core criteria for customers selecting a cloud service provider are undergoing a fundamental change. Full-stack capabilities encompassing computing power, large models, chips, and platforms are replacing previous advantages in resource scale and price as the new competitive focus. Correspondingly, the pricing logic is shifting from resource-based billing to value-based payment. The most direct manifestation of this shift is that market growth momentum is moving from competition for existing customers to innovation-driven expansion. The traditional IaaS price war is gradually receding, replaced by a new wave of expansion driven by intelligent computing demand. Pricing models for large model products are increasingly leaning towards token-based billing, while demand for cloud infrastructure products, particularly high-performance computing power, has surged significantly. In this new competitive environment, whether a cloud service provider possesses AI-native capabilities and efficient computing power scheduling capabilities will directly determine its ability to secure a favorable position.
Full-stack capability is becoming a competitive barrier, with the market consolidating towards dual-strength providers in "computing power + large models." Areas such as AI computing clusters, liquid-cooled data centers, self-developed chips, and large model training represent high-investment tracks. Only providers with complete, closed-loop capabilities can achieve large-scale monetization of computing power and navigate the investment cycle successfully. According to IDC MarketScape evaluations, cloud providers with full-stack capabilities, such as Alibaba Cloud and Baidu AI Cloud, are seeing their market shares continue to rise. Leveraging comprehensive advantages in hardware chips, heterogeneous computing compatibility, cluster scheduling, Model-as-a-Service (MaaS) platforms, and industry ecosystems, these providers have accumulated extensive implementation cases across various sectors, including government enterprises and the internet. In contrast, providers relying solely on resource leasing are experiencing slowing growth due to a lack of technological and ecosystem support. The future market will further concentrate towards leading providers that possess full-stack AI capabilities, ecosystem synergy, and deep industry-specific adaptation capabilities.
Industry divergence is intensifying, with highly adaptable sectors leading the way. Differences in public cloud demand across various industries are widening. Sectors with high AI integration and ample capital resources, such as autonomous driving, e-commerce, gaming, internet finance, and collaborative office applications, are maintaining rapid growth in public cloud demand. IDC predicts that AI public cloud penetration in broad technology-related industries will continue to increase in 2025, further widening the growth rate gap between different sectors. Conversely, industries like government, manufacturing, and traditional finance, constrained by compliance requirements and system upgrade cycles, are adopting cloud services at a relatively slower pace. Customers in these industries place greater emphasis on multi-dimensional capabilities during vendor selection, including model training and inference performance, industry-specific fine-tuning and implementation, data sovereignty and compliance, and cost governance. This is also pushing cloud service providers to accelerate the development of industry-specific and scenario-based product portfolios.
Overseas resource deployment is accelerating, making international expansion a second growth curve. Cloud providers with international backgrounds, such as AWS and Microsoft Azure, maintain strong competitiveness in supporting the overseas expansion of cloud computing power, continuously serving the global deployment and cross-border AI business needs of Chinese companies. Simultaneously, Alibaba Cloud, Tencent Cloud, Huawei Cloud, and telecom operator-affiliated cloud services are actively targeting the overseas needs of Chinese enterprises, accelerating their presence in areas like cross-border e-commerce, game publishing, and AI application globalization. IDC tracking data indicates that the compound annual growth rate (CAGR) for the cloud market serving Chinese enterprises expanding overseas over a five-year period exceeds 30%, significantly higher than domestic market growth. Overseas expansion is becoming a crucial direction for Chinese cloud providers seeking a "second curve" of growth.
Who will benefit from the AI dividend: Three types of providers charting distinct paths. In this AI-driven market restructuring, different types of cloud service providers are following distinct growth trajectories. Full-stack providers, represented by Alibaba Cloud and Baidu AI Cloud, are building a competitive moat in the public cloud space through a closed-loop product line formed by "models + chips + platform + public cloud infrastructure." Their public cloud IaaS year-on-year growth has increased from single-digit percentages in 2024 to over 25% in 2025, with market share continuously rising and increased investment in large model platforms, MaaS, and industry fine-tuning. Scenario-driven providers, represented by Tencent Cloud and Volcano Engine, are more focused on the commercial implementation of AI, pushing to make AI truly "usable." Tencent Cloud achieved scaled profitability for the first time in 2025, while Volcano Engine, leveraging cost-effective intelligent computing solutions and flexible billing models, again achieved over 100% year-on-year growth in 2025, rapidly increasing its market share. Computing power operator-type providers, represented by China Telecom's eCloud, China Mobile Cloud, and Huawei Cloud, rely on a combination of self-built infrastructure and cross-platform scheduling capabilities to flexibly adapt third-party solutions, providing assurance for AI applications in high-security industries like government enterprises and finance. Benefiting from years of accumulated data center resources, localized services, and their state-owned enterprise background, the enterprise service advantages of these providers are gradually becoming apparent, with their market rankings consistently within the top five.
From computing power to models: A four-stage evolution path is taking shape. In the first stage, during 2026, investment in computing power will remain the core direction for AI development. Capital expenditure will continue to climb, driven by AI computing clusters, self-developed chips, and data center construction. IDC predicts that by 2027, over 85% of Chinese organizations will transition their traditional cloud environments to new platforms adapted for AI workloads. In the second stage, leading up to 2027, commercialization will see a breakthrough. The revenue model for cloud service providers will shift from pure computing power leasing to a dual-revenue structure of "Tokens + computing power." The billing model will migrate from the resource side to the scenario side, subsequently reshaping the profit landscape for cloud providers. A market structure for large model-driven AI cloud services is forming, with new business models like MaaS, industry fine-tuning, and Agent platforms accelerating their implementation. Entering the third stage in 2028, the competitive focus will shift from the computing power foundation to the training, fine-tuning, and inference optimization of large model application scenarios. The "usability of models on the cloud" will become the decisive factor for enterprise customers choosing a cloud provider. Enterprise buyers will prioritize evaluating capabilities in multimodal model coverage, inference accuracy, industry adaptation, and the completeness of the ecosystem toolchain. Looking ahead over the next five years, industry divergence will become a long-term characteristic. Broad technology sectors will continue to lead, while traditional industries will gradually but steadily advance their cloud adoption processes, driven by policy guidance and domestic adaptation. The market will concentrate towards comprehensive AI service providers with global capabilities, industry solution expertise, and ecosystem synergy. Leading providers will build long-term barriers through full-stack capabilities and deep industry expertise.
Analyst Conclusion: The Upgrade Battle from "Cloud Provider" to "AI Service Provider". The return to high growth in the Chinese public cloud market in the second half of 2025 is essentially a dividend from the public cloud infrastructure boom driven by the AI industry's explosion. An IDC China research manager stated that the IaaS growth rate rebounding above 20% signifies that China's public cloud industry has transitioned from competing for existing market share to innovation-driven expansion. Over the next three to five years, competition in AI capabilities will enter an intense phase. The market will no longer be a simple contest of price and scale, but a comprehensive competition involving computing power, models, industry solutions, ecosystems, and global capabilities. Cloud computing, especially public cloud services, as the core carrier of AI competition, will directly reflect the development status of the AI landscape through the breadth of its resource deployment, capability resilience, security, revenue growth, and dynamically improving profit conversion rates, serving as a barometer for AI competitive dynamics. In this upgrade battle, the providers that can more rapidly translate AI capabilities into tangible business value perceived by customers are poised to gain a leading position in the upcoming market reshuffle, truly completing the leap from "cloud provider" to "AI service provider."