The ASUS AI POD solution has been expanded once again, now featuring the superior NVIDIA GB300 NVL72 system in the name of XA GB721-E2.
This enterprise-grade rack-scale platform leverages the aforementioned design, packing 6 NVIDIA Grace CPUs and 72 NVIDIA Blackwell Ultra GPUs into a single NVLink domain, offering ultra-low-latency and high-bandwidth GPU interconnectivity suited for trillion-parameter workloads. The platform is fully compatible with NVIDIA’s Quantum-X800 InfiniBand and Spectrum-X Ethernet networking, as well as ConnectX-8 SuperNIC, enabling cluster-scale scalability and enterprise AI factory deployment.
Housed within a 48U NVIDIA MGX-compliant rack, the XA GB721-E2 integrates nine NVLink switch trays and 18 compute trays, supported by manifold-based liquid cooling for predictable thermal performance and streamlined servicing. ASUS complements this with flexible storage configurations ranging from all-flash arrays for AI and HPC to hybrid and unified systems for enterprise data management, ensuring balanced performance across workloads.
For streamlined orchestration, the XA GB721-E2 incorporates ASUS Control Center (ACC) and ASUS Infrastructure Deployment Center (AIDC), enabling zero-touch onboarding, centralized management, and rapid deployment that brings full racks online in as little as 30 minutes. This integration accelerates the journey from installation to production-scale computing while maintaining operational efficiency.
To support deployment and lifecycle needs, ASUS also offers Professional Services covering validation, integration, and optimization, including assistance with cooling design, storage pipelines, and network configuration.
