en.Wedoany.com Reported - On June 2, Delta Electronics, a Taiwan, China-based power management and smart green solutions provider, launched a prefabricated AI modular data center solution at COMPUTEX 2026, targeting the construction needs of high-performance, megawatt-level, and high-density AI data centers. Through factory pre-assembly, pre-testing, and modular delivery, this solution reduces data center deployment time by up to 60%. It also showcases high-voltage DC power supply, liquid cooling, microgrid, and edge AI application solutions.
The core change of this solution is shifting the on-site integration work of traditional data center construction to the factory. The AI modular data center displayed by Delta integrates power, cooling, piping systems, and IT infrastructure into prefabricated units, offering a replicable delivery path for hyperscale cloud service providers, enterprise AI infrastructure, and high-density computing expansion projects. For customers building GPU clusters, AI training platforms, and enterprise private computing centers, construction time, power supply stability, cooling capacity, and on-site construction complexity have become key factors limiting the speed of computing deployment. Traditional data center projects typically require long-cycle design, civil engineering, electromechanical installation, commissioning, and load verification. Prefabricated delivery can transfer some engineering risks to standardized manufacturing and factory testing phases, enabling customers to complete computing deployment faster once site conditions, power access, and other prerequisites are met.
Delta also showcased an AI containerized data center, integrating UPS, IT equipment, and cooling infrastructure into a unified structure occupying about one parking space, claiming a PUE below 1.19.
AI data center construction is shifting from "server procurement" to a system-level competition in power, cooling, space, and delivery speed. As per-rack power continues to increase, traditional AC-DC power supply architectures, air cooling, and distributed server room construction models face efficiency bottlenecks. Delta's proposed 800VDC in-rack power supply solution places power supply and battery backup modules directly inside the cabinet, paired with a 2.4MW liquid-to-liquid cooling distribution unit, liquid cooling busbars, chip-level cold plates, and a platform cooling design for the NVIDIA Vera Rubin NVL72, aiming to reduce energy conversion losses from "grid to chip" and enhance system resilience. For AI cloud service providers and enterprise customers, simply increasing the number of servers is no longer sufficient as computing scale expands. Power supply and distribution efficiency, cooling redundancy, thermal management paths, and on-site maintenance capabilities directly impact equipment availability, energy costs, and long-term operational safety.
This launch also incorporates microgrids into the AI data center infrastructure portfolio. Delta's data center microgrid solution can integrate multiple power sources such as renewable energy, batteries, generators, and solid oxide fuel cells, reducing power conversion stages through DC coupling and solid-state transformers. As AI data centers place greater strain on power grids, companies must now consider regional power capacity, backup power, low-carbon energy access, and grid stability during site selection and construction, beyond just land, network, and tax factors. Combining microgrids with modular data centers allows some high-density computing facilities to achieve more flexible power supply architectures in areas with grid constraints, while also providing new engineering solutions for companies to control carbon emissions and operating costs.
Delta also demonstrated physical AI and edge AI applications based on the NVIDIA Omniverse library at the exhibition, including the DIATwin system for smart manufacturing, the AIoT platform Line Manager, and a predictive energy management solution for building automation. These applications have been implemented on Delta's AI server power supply production line in its Thailand factory. The building automation scenario uses virtual environments, IoT sensors, and physics simulation engines to proactively adjust shading, HVAC, and lighting. This part of the display indicates that the capability boundaries of AI infrastructure companies are extending from power supplies, cooling, and server room equipment to factory digital twins, building energy efficiency, and edge device thermal management.
Subsequent project variables focus on three aspects: first, whether the prefabricated AI modular data center can consistently achieve the 60% deployment time reduction in real customer projects; second, whether the 800VDC power supply, liquid cooling, and microgrid solutions can adapt to data center standards in different countries and regions; and third, whether Delta can form a more complete customer delivery loop across AI servers, data center infrastructure, and smart manufacturing scenarios. If these solutions are implemented in more projects, AI computing construction will no longer be just a competition of chips and servers, but also a comprehensive contest of modular engineering, energy systems, and thermal management capabilities.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
