The EdgeSites model aims to place high-density AI infrastructure inside already powered commercial and industrial properties, offering an alternative to building new data centers.
Infinium Edge has introduced EdgeSites, an infrastructure program designed to deploy AI computing capacity inside existing commercial and industrial properties rather than constructing conventional data centers from the ground up. The model pairs available electrical capacity at those sites with factory-built immersion-cooled computing modules, potentially shortening deployment timelines while avoiding the need for new grid interconnections and municipal cooling water.
The premise rests on an increasingly important constraint in the AI economy: computing demand can grow faster than the physical infrastructure needed to support it. Infinium points to commercial and industrial facilities whose electrical systems were designed around peak demand but are typically used below that level, arguing that some of this already energized capacity could instead support GPU infrastructure behind the meter.
At the center of EdgeSites is Vector ONE, a self-contained immersion cooling system designed to accommodate 1 MW of AI compute capacity. The units can be installed indoors or outdoors and arrive pre-integrated and commissioned, while a dry-cooler loop eliminates the need for municipal water. Multiple units can be combined at a location, with Infinium saying individual sites could support deployments of up to 10 MW depending on available power and demand.
The approach is particularly aimed at AI inference, where computing increasingly needs to be distributed closer to the places applications and users generate demand. Traditional data center projects can face lengthy utility interconnection queues, construction requirements and concerns over water consumption, making existing powered properties an intriguing alternative when their electrical and physical characteristics are suitable. Rather than treating warehouses, distribution centers and other commercial facilities solely as real estate, EdgeSites effectively treats spare electrical capacity as an infrastructure asset.
The model also divides responsibilities differently from a traditional data center development. Property owners can host the equipment without making a capital investment, according to Infinium, while AI companies and compute operators access capacity through agreements priced by the kilowatt-month. Infinium manages development, installation, operations and monitoring, creating a model whose broader significance will depend on how widely suitable host sites can be identified and converted into dependable computing infrastructure.