QumulusAI Completes NVIDIA Blackwell B300 Deployment, Activating $18 Million Take-or-Pay Agreement

QumulusAI has completed the deployment of NVIDIA Blackwell B300 GPU capacity at its Philadelphia colocation site, reaching full customer handoff on Sept. 1, 2026, and activating an $18 million, two-year take-or-pay agreement originally announced on July 22, 2026. The deployment marks a significant milestone for the neocloud infrastructure provider as it moves contracted capacity into live production for a GPU cloud marketplace that serves AI teams across more than 100 regions worldwide.

The agreement represents the full scope of NVIDIA Blackwell B300 capacity contracted under the deal, which is now operational at the Philadelphia facility. The completion of this deployment underscores the growing demand for high-performance GPU compute infrastructure as AI workloads continue to scale globally, and it highlights the increasing role of specialized neocloud providers in delivering capacity that traditional hyperscale models have struggled to supply at speed.

“Building a reliable GPU business starts with reliable supply partners, and that’s the relationship this deployment represents. The agreement we announced in July is now live in Philadelphia and fully in our customer’s hands,” said Michael Maniscalco, CEO of QumulusAI.

The activation of the take-or-pay agreement signals confidence in QumulusAI’s inference-first, demand-led deployment model, which places compute capacity closer to customer demand through a distributed network of data center sites. For the broader AI infrastructure market, the deployment reflects the accelerating race among neocloud providers to secure and operationalize next-generation GPU hardware, particularly NVIDIA’s Blackwell architecture, as enterprises and AI teams seek faster, more flexible alternatives to centralized cloud capacity constraints.