QumulusAI, an artificial intelligence cloud infrastructure company, has announced the purchase of 1,632 Nvidia Blackwell B300 graphics processing units (GPUs) as part of one of the largest capacity expansions in its history.
According to the company, the GPUs will be delivered across 204 Nvidia HGX B300 systems. The order also includes 192 Nvidia RTX PRO 6000 Blackwell GPUs, which will be used to support artificial intelligence training, inference and visualisation workloads.
QumulusAI said the acquisition was financed primarily through arrangements with Technology Finance Corporation and USD.ai.
The company noted that the expansion is intended to support more than US$124 million worth of three-year AI inference agreements secured earlier this year, strengthening its ability to meet growing customer demand for computing capacity.
QumulusAI sources its hardware from multiple original equipment manufacturers (OEMs), including Supermicro and Lenovo, and deploys the systems across a network of colocation facilities and company-owned data centres.
Commenting on the investment, QumulusAI Chief Executive Officer Mike Maniscalco said success in the AI infrastructure market depends on balancing availability, speed and flexibility.
”Meeting AI demand today is a test of access, speed and flexibility, and you can’t win on one without the other two,” he said.
The company also disclosed that its deployed GPU fleet grew by more than 450 per cent between June 2025 and June 2026, reflecting the rapid expansion of its AI cloud infrastructure business.
