AI Servers
Seraya supplies AI servers configured to your workload. Card count, chassis, power and cooling are agreed per order and stated in the quotation.
Open line
NVIDIA RTX PRO 6000 Blackwell
Three editions. One model name. Different machines.
96 GB of GPU memory in every edition, but memory bandwidth, power draw, cooling and the host platform change. Pick the edition before ordering.
Choose an edition
The three editions differ in memory bandwidth, power, cooling and the platform they fit. They are not interchangeable.
Pick an edition to see its specifications
Nothing is selected yet, so no SKU is attached to your enquiry. Choose an edition above, or choose “Not sure which edition” and describe the workload instead.
- Server Edition — Rack servers and multi-GPU nodes where the chassis supplies airflow and the system sets the power envelope.
- Workstation Edition — Tower workstations doing local model training, rendering and simulation at full performance.
- Max-Q Workstation Edition — Workstations and deskside units with limited power or cooling headroom that still need 96 GB of GPU memory.
How the editions differ
Everything below changes with the edition. Confirm the edition before you order — the same model name covers hardware that fits different machines.
| Comparison point | Server Edition | Workstation Edition | Max-Q Workstation Edition |
|---|---|---|---|
| Memory bandwidth | 1,597 GB/s | 1,792 GB/s | 1,792 GB/s |
| Power | 400-600 W (configurable) | 600 W | 300 W |
| Cooling | Passive — server chassis airflow required | Active — double flow-through | Active |
| Form factor | 4.4" (H) x 10.5" (L), dual slot | 5.4" (H) x 12" (L), dual slot | 4.4" (H) x 10.5" (L), dual slot |
| Interface | 512-bit | PCIe Gen 5 x16 | PCIe Gen 5 x16 |