ChronoScale had deployed 6,144 NVIDIA H100 GPUs across three locations by August 31, serving one dedicated GPU-capacity customer.

ChronoScale’s SEC filing provides a dated physical deployment count. As of August 31, 2026, the company had deployed 6,144 NVIDIA H100 graphics processing units, or GPUs, across three locations. One GPU-capacity customer had dedicated access to infrastructure at all three locations; the count refers to GPUs, not servers or racks.

The dedicated-capacity business uses GPU-hour billing. Customers pay a fixed per-GPU, per-hour rate that is subject to change, with fees billed and paid monthly. The excerpt does not disclose the actual rate, contract utilization, or the dedicated customer’s identity.

A separate activity served an additional customer. Beginning in the first quarter of fiscal 2027, ChronoScale also provided certain one-time resales of AI infrastructure hardware, together with ancillary support and maintenance services. Those arrangements served one additional customer as of August 31, which should not be counted as a second dedicated GPU-capacity customer.

Additional capacity and inference services are described separately from existing deployments. The filing says ChronoScale maintained additional deployable capacity at each location and intended to offer it through dedicated GPU infrastructure services and Token Factory, its managed inference offering billed on a usage-based, per-token basis. It does not quantify that additional capacity or establish that it was already contracted.