Meridian Cloud to Use Infinium Edge as Platform for Dedicated Inference Capacity

Multi-site deployment of Infinium Edge Thermal Vectoring™ at key low-latency metro locations supports Meridian's rapid time-to-live commitment to mid-market AI customers

DALLAS, June 24, 2026 /PRNewswire/ -- Meridian Cloud, the dedicated inference utility purpose-built for mid-market AI companies, today announced it has selected Infinium Edge™ as the platform underpinning its rollout of contracted GPU capacity across the United States. The agreement combines Meridian's reserved-instance model with Infinium Edge's next-generation cooling platform — delivering AI inference with less power, no water, and at a lower cost per token than legacy data centers.

The deployment will land at sites across Meridian's target Tier 1 metros chosen for their proximity to inference workloads and power availability.

Supporting the Inference Wave

Inference is the dominant workload of the next AI cycle, with 60–70% of all compute projected to shift to inference by 2027. In inference, success is measured by utilization and the competitive advantage will belong to those who build infrastructure capable of keeping GPUs operating at full capacity, every hour of every day. Air-cooled and direct-to-chip (DTC) platforms are challenged with this as their inability to support stable temperature profiles force silicon to throttle back under sustained load. Edge Thermal Vectoring™ overcomes this by enabling 100% utilization with no thermal throttling — translating directly into more tokens per GPU, per watt, per dollar.

"Mid-market AI companies need infrastructure that's purpose-built for inference. Infinium Edge gives us the density, efficiency, and time-to-live our customers demand. Edge Thermal Vectoring™ runs today's silicon at full utilization without GPU throttling, without freshwater use and it does it in a footprint that is denser than any air cooled or DTC data center."

— Vic Rose, CEO of Meridian Cloud

Infinium Edge Thermal Vectoring™

Infinium Edge's platform is engineered around the world's purest, most thermally stable dielectric fluids, designed together with tanks and cooling distribution infrastructure to capture heat at the source and support rack densities up to 250 kW. The platform targets a 1.05 PUE, can eliminate regular water use, and is delivered as factory-built, pre-integrated modules that ship to site and energize quickly. The platform scales from 1 MW Vector units suitable for industrial and commercial sites, to 10 MW distributed-edge systems and larger 100 MW+ campus designs.

"We designed Infinium Edge Thermal Vectoring™ for this moment where chip thermals are going exponential. Air cooling has passed its limit, direct-to-chip is approaching its own. Meridian saw what's coming and chose the platform built to serve today's and tomorrow's thermal needs. We're proud to be the data center layer behind their inference utility."

— Robert Schuetzle, CEO of Infinium Edge

Infinium Edge's platform runs across the full range of frontier compute, compatible with NVIDIA, AMD, Intel, and memory from Samsung, Micron, and others, preserving the silicon-agnosticism that Meridian's customers depend on. The platform's closed-loop cooling system eliminates routine water use while delivering industry-leading energy efficiency, addressing growing concerns from communities and policymakers about the resource demands of next-generation data centers. By reducing both power and water consumption, it supports the expansion of AI infrastructure while helping ensure local resources remain available for residents, businesses, and other critical community needs.

The agreement covers multiple projects at key low-latency inference sites, structured for rapid deployment. Infinium Edge will design, build, and operate the data center platform; Meridian Inference takes the full contracted capacity as offtaker, serving its customers through dedicated GPU clusters serving both commercial and sovereign demand.

About Meridian Cloud
Meridian Cloud is the dedicated inference utility for mid-market AI companies, delivering GPU clusters in a reserved-instance model with time-to-live in months, not years. Silicon-agnostic and built across Tier 1 U.S. metros, Meridian gives mid-market AI companies the dedicated infrastructure hyperscalers can't deliver and spot markets don't guarantee. Learn more at meridiancloud.com.

About Infinium Edge
Infinium Edge designs and operates next-generation data center platforms cooled by Edge Thermal Vectoring™ — engineered to run modern AI silicon at the edge of its thermal envelope, where density, throughput, and efficiency peak. Factory-built modules deliver up to 250 kW per rack at a 1.05 target PUE, with low to no water use, and energize in months rather than years. Infinium Edge is a portfolio company of Infinium Holdings, Inc. Learn more at infinium.ai and infiniumco.com.

SOURCE Infinium