Mitsubishi Electric Power Products, Inc. (MEPPI) has unveiled its new chip-to-grid reference designs, an integrated blueprint developed to support AI factories. The reference designs were built for hyperscale, neocloud, and colocation operators across North America with NVIDIA’s Vera Rubin NVL72 and future next-generation NVIDIA AI infrastructure.

The reference designs were built to provide scalable architecture built around 250 MW deployment blocks, support expansion to gigawatt-scale campuses, and enable rack densities to grow from approximately 200 kW today to more than 1 MW per rack.

As generative AI and accelerated computing drive unprecedented demand for AI factory capacity, access to power has emerged as one of the industry’s most significant constraints. To address this challenge, MEPPI’s chip-to-grid reference designs aim to integrate utility interconnection, on-site generation, battery energy storage, electrical distribution, advanced cooling, and facility controls into a unified framework. The goal is to give customers a repeatable and scalable approach to AI infrastructure deployment that reduces engineering complexity and accelerates time to capacity.

Each 250 MW deployment block was built to operate in an islanded configuration, using on-site generation and battery energy storage with a pathway to future utility interconnection, or grid connected. The design pairs scalable power delivery, including support for technologies such as NVIDIA 800 VDC architecture power distribution, with a dual-loop cooling architecture designed for high-density AI workloads, including elevated-temperature direct liquid cooling and lower-temperature chilled-water air-side cooling. Standardized power and cooling infrastructure were designed to enable incremental expansion while preserving flexibility for next-gen AI infrastructure and higher-density computing environments.

“The race to build the next generation of AI infrastructure is driving one of the most significant transformations of our time,” said Tricia Breeger, president and CEO of Mitsubishi Electric Power Products. “As AI demand accelerates, success depends on more than computing power. It requires resilient energy systems and scalable infrastructure that can grow from today’s deployments to tomorrow’s gigawatt-scale campuses. Our Chip-to-Grid Reference Designs help customers deploy AI infrastructure faster and scale with confidence.”

To work together as a single system, AI factories require compute, power and cooling, noted Vladimir Troy, vice president of AI infrastructure at NVIDIA. “Mitsubishi Electric’s Chip-to-Grid Reference Designs give operators a repeatable blueprint for deploying NVIDIA Vera Rubin infrastructure and scaling toward gigawatt-scale AI campuses.”