ECL introduced its FlexGrid platform, a power-agnostic architecture designed to deploy high-density AI data centers in locations with limited grid capacity. The company positions FlexGrid as a way to scale inferencing infrastructure beyond large centralized training campuses and into metro, edge, and industrial markets where available grid power often falls below 50–100 MW. FlexGrid supports modular deployments starting with 2–10 MW grid connections and scaling to 20–25 MW per site by integrating additional local energy sources.
FlexGrid uses ECL’s proprietary power conditioning system to combine multiple energy inputs—including grid power, hydrogen, natural gas, renewables, and diesel—into a unified AC or DC feed for GPU-intensive workloads. Unlike traditional single-fuel data center designs, the platform enables operators to add or swap prime movers without redesigning the facility’s core electrical architecture. ECL says this approach allows operators to respond to regional energy constraints, policy changes, and fuel availability while maintaining consistent power quality to AI infrastructure.
The company framed the launch around the shift from AI model training to distributed inferencing, which requires compute capacity closer to users and enterprise data. ECL founder and CEO Yuval Bachar said the company designed FlexGrid to normalize whatever local energy source is available and deliver clean, reliable power to AI data halls in grid-constrained markets. Dean Nelson, CEO of CATO Digital and founder and chairman of Infrastructure Masons (iMasons), said ECL’s programmable energy layer aligns with the need to blend grid, gas, hydrogen, and renewables to scale AI infrastructure sustainably.
• Power-agnostic architecture combining grid, hydrogen, natural gas, renewables, and diesel
• Modular deployments starting at 2–10 MW, scaling to 20–25 MW per site
• Designed for AI inferencing expansion in metro and edge locations
• Unified AC or DC output via proprietary power conditioning system
• Backed by Molex Ventures and Hyperwise Ventures
“Inference has to live close to people, data and applications, in and around major cities, smaller metros and industrial hubs where there is rarely a spare 50 or 100 megawatts sitting on the grid,” said Yuval Bachar, founder and CEO of ECL. “FlexGrid was built exactly for the market conditions we face today.”
🌐 Analysis: ECL extends its hydrogen-centric foundation into a broader multi-fuel strategy as AI workloads push beyond hyperscale campuses and into distributed inferencing footprints. As competitors race to secure dedicated grid capacity or on-site generation for 50+ MW AI training clusters, ECL targets the 20–25 MW edge segment where energy aggregation and conditioning could determine deployment speed and site selection flexibility.
ECL was founded by Yuval Bachar, a veteran data center architect known for leading hyperscale infrastructure initiatives at companies including LinkedIn and Facebook. Bachar played a central role in advancing high-efficiency data center design and alternative energy strategies, including early exploration of hydrogen-powered facilities. At ECL, he has focused on rethinking the power layer as the core design constraint for AI infrastructure, developing patented approaches to power conditioning and modular deployment. The company has attracted backing from Molex Ventures and Hyperwise Ventures as it positions itself to address energy bottlenecks facing AI training and inferencing expansion.