Gerchamp has launched a modular AI data center portfolio built to deliver factory-assembled power, liquid cooling, backup, and controls to sites that have available power but limited space. The lineup targets deployments at the edge, on-premises, behind the meter, and in remote locations where conventional data center builds may be impractical.
The portfolio ranges from 500 kW all-in-one systems up to 3.5 MW multi-container deployments. Gerchamp’s platform integrates power distribution, UPS, Nickel-Zinc battery backup, battery management system (BMS), data center infrastructure management (DCIM), and direct-to-chip liquid cooling into containerized systems designed to be delivered in months and powered in weeks.
At the small end, the 500 kW all-in-one system packages IT space and power distribution for either five OCP racks or four NVIDIA NVL72 racks, along with Gerchamp’s direct-to-chip liquid cooling, UPS, Nickel-Zinc battery backup, BMS, and DCIM. Servers and GPUs are installed separately.
For larger deployments, Gerchamp combines multiple containers into a coordinated 3.5 MW configuration with two IT containers, two cooling containers, and one power container. Gerchamp lists support for 24 OCP racks or 20 NVIDIA NVL72 racks, with power density up to 175 kW per rack.
Modular, factory-integrated infrastructure is a practical fit when the critical path is site readiness rather than new building construction. But the engineering tradeoffs are real: containerized systems shift a lot of complexity to integration, commissioning, and ongoing serviceability, especially once you add direct-to-chip liquid loops, battery backup, and controls into a tight footprint.
“AI infrastructure is no longer limited to traditional data center sites,” said Frank Yang, Chairman of Gerchamp. “Where power is available, there is an opportunity to deploy compute. The challenge is having infrastructure that can be deployed quickly and work within the site’s constraints.”
Gerchamp said each system is integrated and tested at the factory before shipment, with customer-selected servers and GPUs installed separately.
Source: Gerchamp
