DeepCtrls and Chindata Launch Joint Lab to Advance Computing Infrastructure with Physical AI
Published: 2026-9-18
On September 17, DeepCtrls and Chindata officially inaugurated their joint laboratory. Building on their extensive industry experience and technical expertise, the laboratory will focus on intelligent cooling system diagnostics, predictive operations and maintenance, and coordinated optimization of computing and cooling. It aims to integrate Physical AI into the core operations of computing infrastructure and develop new approaches to intelligent data center operations.
As large AI models continue to advance and AI computing clusters expand, computing infrastructure is becoming denser and more complex. Coordinating cooling systems with computing workloads and energy infrastructure is increasingly challenging. Moving from manual operations and preventive management toward system-wide sensing, autonomous analysis, closed-loop control, and predictive management has become a key priority for intelligent data center upgrades.
The joint laboratory will pursue research in three core areas:
1. Intelligent Diagnostics and Performance Optimization for Critical Cooling and Supporting Equipment
Combining Deepctrls’ proprietary PhyAI technology with Chindata’s technical expertise, the laboratory will develop performance models, operating-state simulations, anomaly detection, and intelligent diagnostics for critical cooling equipment. These capabilities will support condition monitoring, performance simulation, and problem identification.
Addressing challenges such as the lack of dynamic baselines, reliance on manual assessment, and inefficient diagnostics, the research will enable performance forecasting, early anomaly detection, and simulation-based validation of operating strategies to improve operational stability and diagnostic efficiency.
2. Predictive Maintenance and Equipment Health Management
Component-level fault knowledge bases, fault trees, and equipment health assessment frameworks will help shift maintenance from reactive responses after failures to proactive prediction based on equipment condition. The laboratory will also connect diagnostics, decision-making, and maintenance execution in a closed loop.
3. Coordinated Optimization of Computing and Cooling
To address fluctuating computing demand, the laboratory will develop integrated forecasting models covering IT workloads, cooling loads, and ambient temperatures. These models will enable cooling systems to adjust dynamically to computing tasks, coordinating and optimizing computing workloads and cooling operations.
The partners have already conducted extensive research into intelligent computing infrastructure operations, covering performance simulation, computing–cooling coordination, and intelligent operations and maintenance. Their work spans water-cooled systems, air-cooling terminal units, indirect evaporative cooling, and liquid cooling.
The joint laboratory will draw on Chindata’s years of experience operating and managing hyperscale data centers to guide research, technology validation, product development, and deployment. This will help Physical AI evolve to serve larger and more complex AI computing environments.
The laboratory marks another step in Deepctrls’ efforts to use Physical AI to transform global computing and energy infrastructure.
In April this year, Deepctrls was selected for the China Academy of Information and Communications Technology’s (CAICT) “Ten Technology Testbeds for Computing–Power Coordination.” Alongside Huawei, ByteDance, Baidu, and other leading companies, it participated in research addressing direct green power supply and the integration of generation, grids, demand, and energy storage.
In September, the Research Report on Artificial Intelligence Applications for Green Computing was officially released. Led by Deepctrls and CAICT’s Cloud Computing and Big Data Research Institute, the report was jointly developed with industry partners including CATL, Foxconn Industrial Internet, Shanghai Unicom, and Chindata. It examines how Physical AI can support infrastructure planning and design, operational optimization, energy coordination, and maintenance management.
From computing–power coordination to predictive maintenance, and from individual equipment optimization to integrated computing, cooling, and energy management, Physical AI is increasingly becoming part of the core operations of computing infrastructure.
The joint laboratory will connect research, real-world validation, practical deployment, and adoption at scale. Its goal is to advance infrastructure operations from manual control and intelligent assistance toward autonomous decision-making, operation, and optimization, opening a new chapter in unattended operation for large-scale AI data centers.
