Model XA GB721-E2
ASUS rack-scale AI infrastructure built on NVIDIA Grace Blackwell Ultra, with liquid cooling and deployment services.
The ASUS AI POD (XA GB721-E2) is ASUS's full-rack AI supercomputer built on NVIDIA GB300 NVL72, combining 72 Blackwell Ultra GPUs and 36 Grace CPUs in a 48U NVIDIA MGX-compliant rack.
It supports both liquid-to-air and liquid-to-liquid cooling and comes with ASUS software, management and deployment services, targeting enterprise AI, cloud providers, research institutions and national AI clouds.
Each with 288 GB of HBM3e and 1.5x more AI compute than Blackwell.
Per rack, for trillion-parameter models.
1.8 TB/s GPU-to-GPU bandwidth across the rack.
Quantum-X800 InfiniBand or Spectrum-X Ethernet.
Liquid-to-air or liquid-to-liquid, manifold-based.
Software, management and services to streamline roll-out.
| Rack | 48U NVIDIA MGX-compliant |
|---|---|
| Compute trays | 18 |
| NVLink switch trays | 9 |
| Cooling | Liquid-to-air or liquid-to-liquid, manifold-based |
| GPUs | 72× NVIDIA Blackwell Ultra, 288 GB HBM3e each |
|---|---|
| CPUs | 36× NVIDIA Grace |
| Fast memory | Up to 40 TB per rack |
| GPU interconnect | Fifth-generation NVLink, 1.8 TB/s |
| SuperNICs | NVIDIA ConnectX-8 |
|---|---|
| Fabric | NVIDIA Quantum-X800 InfiniBand or Spectrum-X Ethernet |
Next-generation rack-scale platform for agentic AI and reasoning, unifying 72 Rubin GPUs and 36 Vera CPUs in a single NVLink domain.
Turnkey, ready-to-deploy enterprise AI infrastructure built on the Vera Rubin platform and managed with NVIDIA Mission Control.
Blackwell Ultra rack-scale system built for AI reasoning, training and data processing at scale.
Tell us your workload and timeline for the ASUS AI POD (NVIDIA GB300 NVL72). We reply within one business day with a configuration, pricing and lead time.