In response to the rapidly growing power demands driven by AI computing, a research team led by Distinguished Professor Ying-Yi Hong from the Department of Electrical Engineering at Chung Yuan Christian University (CYCU) has partnered with Delta Electronics, Inc. to advance next-generation power infrastructure technologies for AI data centers. Following the completion of a preliminary study on an 800V High Voltage Direct Current (HVDC) power supply system, the team has secured support from Taiwan\u2019s National Science and Technology Council (NSTC) and Delta Electronics to launch a two-year industry\u2013academia collaborative project. The initiative focuses on reliability analysis, power quality assessment, and digital twin technologies for high-power AI data centers, aiming to develop key technologies for Taiwan\u2019s next-generation AI power infrastructure.<\/p>\n\n\n\n
The rapid advancement of Generative AI and Large Language Models (LLMs) has led to a dramatic increase in data center energy consumption. NVIDIA\u2019s GB200 and DGX AI computing platforms, for example, require rack-level power capacities ranging from 30 to 45 kW, while large-scale AI clusters have already entered the megawatt (MW) era. Conventional data center power architectures are increasingly unable to meet the demands of high-density computing. As a result, global cloud service providers and equipment manufacturers are accelerating the adoption of 800V HVDC architectures to improve energy efficiency, reduce transmission losses, extend power delivery distances, and enhance system scalability.<\/p>\n\n\n\n
As a global leader in power management and data center infrastructure, Delta Electronics has long been committed to developing high-efficiency power conversion and energy management technologies and is a key supplier of power solutions for AI data centers worldwide. To address future AI infrastructure requirements, Delta and CYCU jointly conducted a six-month preliminary research project, establishing one of Taiwan\u2019s few analytical platforms dedicated to high-voltage DC power systems for AI data centers.<\/p>\n\n\n\n
Using MATLAB\/Simulink, Professor Hong\u2019s research team developed a detailed model of an 800V HVDC power system tailored to next-generation AI data center architectures. The study covered solid-state transformer (SST) power architectures, power flow analysis, load characteristics, and fault protection mechanisms. The resulting high-fidelity simulation platform enables the analysis of power supply behavior and system dynamics under various operating scenarios. It is expected to serve as a valuable tool for risk assessment, system planning, energy management, energy storage integration, and protection coordination in future AI data centers.<\/p>\n\n\n\n
Building upon these achievements, the research team has now launched a two-year collaborative project supported by the NSTC and Delta Electronics. During the first year, the team will focus on Delta\u2019s next-generation 800V HVDC data center architecture, conducting in-depth studies on system reliability and power resilience in high-power AI data centers. The project will also investigate power quality issues arising from high-load GPU operations and establish comprehensive evaluation methodologies to support future data center design and operation.<\/p>\n\n\n\n
In the second year, the team will integrate OPAL-RT real-time simulation, Power Hardware-in-the-Loop (PHIL), and Long Short-Term Memory (LSTM) neural networks to develop digital twin models capable of fault detection, anomaly prediction, and fault localization. These models will be combined with Unity and NVIDIA Omniverse 3D visualization platforms to provide real-time visualization of equipment status, fault locations, and energy flows, thereby enhancing intelligent operation, maintenance, and decision-making capabilities.<\/p>\n\n\n\n
Professor Hong, the project leader, emphasized that the competitiveness of future AI data centers will depend not only on computing performance but also on the availability of stable, efficient, and resilient power infrastructure. He noted that while 800V HVDC architectures are emerging as a global trend for AI data centers, analytical tools, reliability validation methods, and digital twin technologies remain limited. The team aims to establish a comprehensive analysis and verification platform that can help industry partners shorten development cycles, reduce testing costs, and strengthen Taiwan\u2019s technological capabilities in AI data center infrastructure.<\/p>\n\n\n\n
Professor Hong further highlighted Taiwan\u2019s strategic position within the global AI supply chain, spanning semiconductors, servers, power systems, and energy management technologies. Through this collaboration with Delta Electronics, the project is expected not only to accelerate the practical application of academic research but also to cultivate interdisciplinary talent with expertise in power engineering, artificial intelligence, and digital twin technologies. Such efforts will further enhance Taiwan\u2019s competitiveness in AI data centers, smart energy systems, and digital infrastructure.<\/p>\n\n\n\n
By combining CYCU\u2019s long-standing strengths in smart grids, microgrids, power electronics, and artificial intelligence with Delta Electronics\u2019 extensive experience in global data center power and energy management markets, the project is expected to establish critical technologies for the design and validation of high-power AI data centers. The outcomes may also be extended to applications in smart grids, energy storage systems, and large-scale critical infrastructure, providing new momentum for Taiwan\u2019s AI industry development and energy transition.<\/p>\n","protected":false},"excerpt":{"rendered":"
A research team led by Professor Ying-Yi Hong of the Department of Electrical Engineering at Chung Yuan Christian University has partnered with Delta Electronics to advance power infrastructure technologies for AI data centers. The two parties completed a preliminary study on an 800V High Voltage Direct Current (HVDC) power supply system and have subsequently secured support from the National Science and Technology Council and Delta Electronics to launch a two-year industry\u2013academia collaboration focused on developing next-generation power solutions for AI computing.<\/p>\n","protected":false},"author":7,"featured_media":34944,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[501],"tags":[],"class_list":["post-34943","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news"],"yoast_head":"\n