The FlexGrid marks a strategic improvement for ECL, which has historically targeted on modular information centre improvements and hydrogen-based information centres. Whereas hydrogen stays a part of its method, the introduction of FlexGrid expands its portfolio towards modular energy flexibility.
ECL’s FlexGrid is designed to supply versatile energy options for AI information centres at power-constrained websites. It goals to help the deployment of GPU-based capability nearer to customers and information sources, significantly in areas the place grid energy and hydrogen availability are restricted.
FlexGrid utilises ECL’s proprietary energy conditioning system. The expertise integrates a number of power sources, together with grid energy, hydrogen, pure gasoline, renewables, and diesel, right into a single AC or DC energy feed. The system is designed to function with grid connections of two–10 MW and scale as much as 25 MW as extra power sources are integrated.
A key attribute of FlexGrid is its fuel-agnostic structure. Not like conventional amenities that depend upon a single main power supply, the design permits for the addition of various energy sources with out requiring a redesign. This method is positioned as related within the context of evolving power insurance policies and market circumstances.
With rising AI-related energy consumption, versatile energy infrastructure is more and more required to help AI implementations. The platform is meant to deal with energy constraints whereas supporting scalability.
Business forecasts point out rising electrical energy consumption by AI-optimised servers. FlexGrid is designed to decouple information centre energy structure from reliance on a single gasoline sort, enabling deployment throughout diversified areas to help AI inference workloads.
The FlexGrid displays a continuation of ECL’s give attention to modular energy programs and hydrogen-based information centres, whereas increasing into broader energy flexibility. The event concerned contributions from Yuval Bachar, with help from Molex Ventures and Hyperwise Ventures.
