Capstone, MG4AI Sign Agreement on AI Data Center Infrastructure

Capstone Green Energy Holdings, Inc. and its subsidiary Capstone Green Energy, LLC have signed a Memorandum of Understanding with Microgrids 4 AI, Inc. (MG4AI) to collaborate on infrastructure solutions for the artificial intelligence (AI) data center market.

MG4AI designs modular, turnkey infrastructure for edge data centers under 20 MW, supporting enterprises and government customers in deploying GPU-intensive compute environments. By using localized, grid-independent microgrids with liquid cooling, the company aims to help customers avoid grid upgrade delays, lower operating costs, and increase resiliency.

The collaboration will pair MG4AI’s modular systems with Capstone’s microturbine technology, which is designed for low emissions, high efficiency, and reduced maintenance. The companies plan to integrate Capstone’s Combined Cooling and Power (CCP) systems with MG4AI’s containerized compute pods, enabling faster deployment and improved power and cooling performance.

“Capstone’s advanced distributed power platform is purpose-built for the next wave of digital infrastructure,” said Vince Canino, President and CEO of Capstone Green Energy. “Artificial intelligence is one of the most energy-intensive applications and disruptive workloads of our time.”

“The future of AI will not be built on yesterday’s data centers. It demands innovative new infrastructure – where power, liquid cooling, and compute converge as one, a complete kit,” said Ken Kajikawa, CEO of MG4AI. “Together with Capstone, MG4AI is creating a new blueprint for AI: sovereign, sustainable, and infinitely scalable (140kW today and scaling to 600kW per rack).”

Capstone’s systems are designed to provide on-site power and integrate with cooling equipment such as chillers, dry coolers, and pumping packages, as well as battery energy storage systems when required. Together with MG4AI’s edge compute expertise, the companies aim to deliver deployable solutions that meet the growing power and performance requirements of AI workloads.