QumulusAI (NASDAQ: QMLS) announced the completion of a GPU deployment at its Philadelphia colocation site, activating the full scope of NVIDIA Blackwell B300 capacity contracted under an $18 million, two-year take-or-pay agreement.
The deployment went live on September 1, 2026, marking the first operational milestone since the company first disclosed the deal on July 22, 2026. The hardware will serve a GPU cloud marketplace that provides compute access to AI teams across more than 100 regions, according to the company.
QumulusAI operates as a distributed AI cloud platform, supplying GPU compute infrastructure through an inference-first deployment model across a network of data center sites.
"Building a reliable GPU business starts with reliable supply partners, and that's the relationship this deployment represents," said Michael Maniscalco, CEO of QumulusAI. "The agreement we announced in July is now live in Philadelphia and fully in our customer's hands."
