The Zhitong Finance App learned that Great Wall Securities released a research report saying that with the improvement of model capabilities and the rapid decline in AI usage costs, the bank believes that the prosperity of the cloud computing and AI application industry is on an upward trend. Cloud Computing: The Jevins Paradox will drive total cloud demand into a stage of explosive growth, and cloud computing is expected to maintain a sharp upward trend in volume and price; AI applications: enterprise-level AI applications and C-side agent agents are in the accelerated implementation stage. Among them, enterprise AI budget release has gradually moved from pilot demonstration to large-scale implementation.
The main views of Great Wall Securities are as follows:
The ability of domestic open source models to quickly catch up with overseas closed source models is expected to accelerate the implementation of AI on the enterprise side
Since June, many leading domestic model makers have intensively released a new generation of large-scale models, showing trends of larger parameters, stronger agent (Agent) capabilities, more perfect open source ecology, and more in-depth multi-modal applications. As domestic model makers accelerate model iteration, domestic open source model capabilities are rapidly improving, and the gap with overseas leading closed source models (such as Anthropic, OpenAI, etc.) continues to narrow. Compared with the closed source model, the open source model (1) has a higher cost performance ratio; (2) the model can be optimized at a deeper level to promote autonomous and controllable model behavior; (3) the open source model can be deployed in public clouds, private clouds, local, edge and hybrid, and customers have a variety of choices. The bank believes that as open source models catch up and maintain high cost performance, the top model token price center is expected to continue to decline, thereby driving high-quality basic models from usable to easy to use, thereby speeding up enterprise-level AI deployment and commercialization, and driving the demand side to expand from cutting-edge model laboratories to various vertical fields such as industry, life sciences, finance, and the public sector.
Cloud Computing: The North American and domestic AI cloud computing industry has entered a phase of sharp rise in volume and price
Both North American/domestic CSP and Neocloud 26q2 results have been released. The 26Q2 Azure/AWS/Google/Coreweave/Nebius/Alibaba Cloud business revenue growth rates were 43%/37%/82%/112%/454%/45%, respectively, maintaining an accelerated growth trend. The bank believes that the cloud computing industry has entered a phase of sharp rise in volume and price, mainly due to (1) the rapid increase in open source model capabilities, cloud vendors can provide customers with corresponding models based on customer requirements in terms of quality, delay, cost, etc., and the model hierarchical strategy drives industry voice to the downstream CSP; (2) the actual cost of AI usage (Cost/Mtok) has dropped drastically. After there are economic benefits on the cost side, the Jevens paradox will drive total cloud demand to the explosive growth stage; (3) the number of open source model parameters continues to rise; (3) the number of open source model parameters continues to rise, such as KIMI Released on July 17th The number of K3 model parameters has reached 2.8 trillion. The resource threshold for a single deployment is rising simultaneously, and it is not economical for enterprises to build their own computing power, so they prefer to rent cloud vendor infrastructure, driving cloud factories' IaaS+MaaS revenue resonance. 2026 H1 Major domestic and foreign cloud vendors have successively raised the prices of computing power services, and the price war has become a thing of the past. The cloud business revenue growth rate of North American cloud vendors, NeoCloud and domestic cloud vendors continues to accelerate, and the logic of the sharp rise in volume and price was verified by the report.
AI applications: After Vibe Coding, focus on application-side opportunities such as AI4S, FDE, and AI office
Referring to the mobile internet era, AI applications will be the direct beneficiaries of falling token prices and improved model capabilities. The 26H1 downstream transformation scene is mainly focused on Vibe Coding. As model capabilities improve, AI is expected to empower more downstream application scenarios (1) AI4S: AI is used to accelerate the process of scientific research and discovery, and has now been implemented in various fields such as life science, physics, materials science, etc., to create a complete data closed loop of “AI prediction, automated verification, data backflow”; (2) FDE: model implementation enterprise business needs to overcome problems such as system connection, permission sorting, process transformation, compliance audit, and organizational adaptation. Facing customers, FDE transforms business experience and process rules scattered in employees' minds into knowledge bases, workflows, and application capabilities that can be used by AI; platform-oriented, FDE deposits industry scenarios, system interfaces, test methods and failure cases into reusable product components and continuously iterates products, thereby reducing the delivery cost and cycle of subsequent FDE projects; (3) AI office: enterprises and employees are highly motivated to pay for efficiency, AI improves work efficiency, and users are more willing to pay; in addition, the bank believes that AI can continuously obtain private enterprise data (such as approval flow, documents) to fine-tune the model to form a positive data flywheel cycle where users use it more - the more the model understands the enterprise - the more users are inseparable.
Related targets
Cloud Computing/Computing Power Rental: Alibaba, Jinshan Cloud, Guangdong-Gangwan Intelligent Computing, Litong Electronics, Jiechuang Intelligence; AI applications: Tencent Holdings, Kuaishou, Maifushi, Shenzhou Holdings, Huiliang Technology, Haizhi Technology Group, Fubo Group, Focus Technology, Chinese Online, Blue Cursor, Giant Network, Kaiying Network.
Risk warning: AI big model iteration falls short of expectations; commercialization of AI applications falls short of expectations; industry competition intensifies; regulatory policy impact