QumulusAI, an AI cloud provider, has secured $224 million in customer commitments. The company has signed two three-year contracts worth more than $220 million with “open access AI cloud” Hyperbolic and one other unnamed AI inference platform. QumulusAI is planning a data center in Denton, Texas.
The deals include $21.9 million in total upfront commitments. Under the agreement, QumulusAI will deploy 1,280 Nvidia Blackwell GPUs across 160 Lenovo and Supermicro bare-metal servers featuring B300 and B200 Blackwell GPUs, respectively. Cisco Nexus One will serve as the backbone for the cluster fabric in both deployments, providing secure and high-performance AI networking. “AI infrastructure can no longer rely on one-size-fits-all designs,” said Mike Maniscalco, CEO of QumulusAI.
Inference workloads have distinctly different performance and cost demands compared to model training setups. By aligning infrastructure with the specific workload, we can boost utilization, lower costs, and speed up deployment timelines for customers running at production scale. As AI adoption grows, organizations require infrastructure that can be rapidly deployed, scaled effectively, and optimized for the economics of production AI, added Maniscalco.
These agreements highlight the benefits of adopting a more adaptable strategy for AI infrastructure. QumulusAI has not revealed the deployment location of the cluster.
