Cooling infrastructure and building automation are key support systems for AI factories
In 800VDC data center architectures, cooling infrastructure and building automation play the dual roles of "thermal management neural center" and "intelligent operations brain." Although they do not directly participate in power conversion, they are key support systems that determine whether an AI factory can operate stably, efficiently, and sustainably. The power density of AI factories has leaped from tens of kilowatts per rack in traditional data centers to 100kW or even megawatt levels, and traditional air-cooling solutions are approaching their physical limits. While the 800VDC architecture improves power delivery efficiency, it also introduces higher thermal density challenges.
The core role of cooling infrastructure and building automation has transitioned from "passive heat dissipation" to "active intelligent control." Cooling infrastructure serves as the "thermal management execution layer," responsible for efficiently removing the massive heat generated by key components such as GPUs, CPUs, SSTs, and busbars. Building automation serves as the "intelligent operations management layer," responsible for global monitoring, analysis, and optimization of energy usage, environmental parameters, and equipment status across the entire data center.
Under the 800VDC architecture, cooling technology is undergoing a full-chain innovation "from chip to cooling tower," with liquid cooling becoming standard. AI servers have shifted from air cooling to direct-to-chip liquid cooling (DLC) and even immersion liquid cooling. Busbars are also beginning to be liquid-cooled; as 800V DC busway distribution systems carry currents as high as 15,000A or even 25,000A in the future, their self-heating has become a non-negligible issue. In addition, full-chain collaborative optimization has become an important trend. Cooling systems are optimized across the entire chain from chip to cooling tower and deeply integrated with building automation systems, helping data centers achieve Power Usage Effectiveness (PUE) as low as below 1.1.
Building automation has evolved from "manual operations" to "AI-driven." AI technology is reshaping data center management methods, such as adopting digital twin simulation technology to build high-fidelity dynamic digital twins of data centers, integrating lighting, HVAC, sensors, and building automation to achieve up to 20% potential energy savings, as well as enabling predictive maintenance and real-time optimization. By using AI algorithms to analyze massive sensor data, it predicts equipment failures and dynamically adjusts cooling strategies, shifting from passive response to active prevention. Furthermore, AI-driven intelligent operations robots can now perform 24/7 autonomous equipment inspections, environmental monitoring, fault early warning, and other full-process tasks, with recognition accuracy exceeding 99% and annual labor cost reduction of 70%.
Cooling infrastructure and building automation: key support for intelligent, green, and sustainable operations
The 800VDC architecture brings not only improvements in power efficiency but also a sharp increase in thermal density - higher power means higher heat generation. Cooling and building automation have become decisive factors in whether AI factories can be deployed at scale. In cooling, liquid cooling technology is transitioning from an optional solution to a standard requirement. Not only is power transmitted through 800V busbars, but the busbars themselves are beginning to adopt liquid cooling. At the building automation level, AI digital twin technology is transforming data center operations. Cooling systems are also undergoing full-chain optimization from chip to cooling tower, deeply integrated with Building Management Systems (BMS) and supporting open standards such as Redfish.
Cooling infrastructure and building automation are the key supports that enable 800VDC data centers to move from the "physical challenge of high power density" toward "intelligent, green, and sustainable operations." Liquid cooling technology (from chip to busbar) addresses the thermal density crisis, while AI-driven building automation achieves global optimization of energy usage and intelligent operations maintenance. Together, they form the solid foundation for the stable operation and green transformation of AI factories. At the Arrow Electronics booth at PCIM Asia 2026, these solutions will demonstrate how power, cooling, and intelligent management can be integrated into a collaboratively operating whole.
In the 800VDC data center architecture, NXP's motor control solution does not directly handle 800V high voltage, but instead serves as the "intelligent brain" of the "cooling infrastructure," responsible for precisely driving and managing cooling fans, pumps, and other equipment based on real-time thermal conditions, ensuring stable system operation under ultra-high power density. The core logic of NXP motor control technology in intelligent thermal management is to use sensors and AI algorithms to achieve "on-demand cooling," transitioning from "passive heat dissipation" to "active intelligent control." It enables intelligent on-demand cooling, improves system reliability, and supports high-density deployment. It constitutes the intelligent brain of the cooling infrastructure and works in coordination with the building automation system to achieve global optimization of the data center's thermal environment.
Click here to view related information
In the ultra-large-scale, high-power-density complex environment of the 800VDC data center, NXP's UWB (Ultra-Wideband) Anchor and Tag technology plays the key role of the "digital spatiotemporal perception layer." It does not directly participate in power conversion, but provides centimeter-level, real-time precise location data that is critically important for intelligent operations maintenance, asset management, and security protection in data centers. It is one of the core technologies enabling building management to advance from "automation" to "intelligence." By precisely measuring the Time of Flight (ToF) of radio signals, UWB technology provides a new perception dimension for data centers, enabling high-precision asset tracking, robot positioning and navigation, intelligent access control and security, environmental monitoring, and process optimization, laying a solid foundation for efficient, safe, and intelligent management of AI data centers.
Click here to view related information
NXP's Matter solution is deeply integrated with its broad portfolio of MCUs, wireless connectivity, and security products, serving as an important cornerstone for building an building automation ecosystem. It provides a unified, secure, IP-based application-layer connectivity standard for a wide variety of edge devices within the data center (such as lighting, thermostats, sensors, access controls, etc.), enabling them to seamlessly access the Building Management System (BMS) and achieve campus-wide intelligent management. In data center operations, this means that cooling systems, lighting systems, security systems, and more can be collaboratively optimized based on a unified platform, integrating the vast and complex auxiliary systems of the data center into an efficient and synergistic whole. This capability enables the cooling infrastructure and the building management system to truly "communicate" and coordinate, ultimately supporting the entire AI factory's move toward green, intelligent, and sustainable operations.
Click here to view related information
In the 800VDC data center architecture, the 15kW three phase three-level T-type bidirectional PFC (Power Factor Correction) reference design jointly launched by Arrow Electronics and STMicroelectronics (ST) plays the key role of an "efficient, bidirectional intelligent interface between the grid and the DC busbar." It is not only a "rectifier" that efficiently converts grid AC power into 800V DC power, but also an "inverter" that can feed DC power back to the grid, making it a crucial link in enabling green energy dispatch and intelligent management for data centers. This 15kW PFC solution serves as the critical bridge between the power grid and the internal 800V DC busway in the AI factory's power supply chain, supporting efficient AC/DC conversion, bidirectional energy flow, and improved power quality. It provides solid power electronics technology support for AI factories to achieve green energy integration, smart grid interaction, and sustainable operations.
Click here to view related information
In the 800VDC data center architecture, Molex's liquid-cooled busbar plays the key role of an "integrated power backbone that combines power transmission and efficient heat dissipation." It directly addresses the "Power Path thermal management challenge" brought by the surge in AI computing power and is the infrastructure that supports rack power scaling toward the 1 megawatt (MW) era. Molex's liquid-cooled busbar is the "thermal-electrical integrated backbone" for AI high-power racks, breaking through the limits of air-cooling heat dissipation, supporting ultra-high current transmission, and enabling high-density rack deployment. Molex's liquid-cooled busbar demonstration kit integrates high-current power transmission with efficient liquid cooling heat dissipation, making the power system itself part of the cooling solution, thereby supporting high-density, high-power deployment of 1MW-class racks under the next-generation 800V DC architecture.
Click here to view related information
Analog Devices (ADI)'s Coolant Distribution Unit (CDU) solution is specifically designed for liquid cooling systems used in data centers, high-performance computing (HPC), and industrial computing environments. The solution features an efficient, modular, and plug-and-play intelligent control architecture. The core of MAX32651 serves as the central controller for collecting temperature, pressure, flow, and other sensor data, while executing overall cooling control strategies. The CDU connects to the control system through interfaces such as RS-485, USB, and UART, while an industrial PC or host system acts as the upper-level management platform for system monitoring, configuration, and operational management.
For data center applications, an optional Application Processor Board can be integrated to support open management interfaces such as Redfish. Through Ethernet or serial communications, it enables remote monitoring, out-of-band management, and seamless system integration. Compared with PLC-based architectures, the ADI CDU solution offers hot-swappable modules, redundancy support, online maintenance, and open protocol compatibility. It also features software-defined I/O (Programmable I/O) capabilities, allowing input and output functions to be rapidly reconfigured through software based on site-specific requirements, without hardware redesign. This flexibility helps customers improve system reliability, serviceability, and scalability while accelerating the development and deployment of liquid cooling systems.
The Cooling Distribution Unit (CDU) reference design based on the NXP main control chip serves as the "intelligent hub" for thermal management in AI factories. In the 800VDC architecture, this CDU plays the role of an intelligent heat exchange and distribution hub connecting the "heat source" and the "cold source." As AI cluster rack power moves toward the megawatt level, traditional air cooling can no longer handle the heat generated by GPU clusters. The value of this CDU lies in supporting efficient heat transfer, precise temperature and humidity control, and full-chain intelligent monitoring. Through high-performance real-time control, high-precision analog signal chains, and standardized communication interfaces, it achieves precise and reliable intelligent management of the liquid cooling system, and is one of the key technologies supporting AI factories in reaching megawatt-level rack power density and ensuring their stable operation.
Click here to view related information
Article tags
