Brand: HP | Category: Servers
SKU: HP-P78420B21 | Part #: P78420-B21 | MPN: P78420-B21
Contact for Pricing — Request a Quote
The HPE Cray Supercomputing EX4000 (NVIDIA GB200) from HP is enterprise-grade Servers hardware built for data centers. HPE Cray EX4000 delivers liquid-cooled NVIDIA GB200 NVL72 blade-scale AI supercomputing density for the most demanding HPC and generative AI workloads. Available in brand new condition. HP servers products are trusted by enterprises and data centers worldwide. Available from Omnixon Global. Contact our sales team to request a quote.
| Brand | HP |
| Category | Servers |
| SKU | HP-P78420B21 |
| Part Number | P78420-B21 |
| Condition | New |
| Product Line | HPE Cray Supercomputing EX4000 |
| GPU Architecture | NVIDIA Blackwell (GB200) |
| NVL72 Configuration | 72x NVIDIA B200 Tensor Core GPUs + 36x NVIDIA Grace Arm CPUs per NVL72 rack domain |
| GPU Memory per B200 | 192 GB HBM3e |
| Total GPU Memory per NVL72 Rack | Up to 13,824 GB (13.5 TB) HBM3e unified via NVLink |
| NVLink Fabric Bandwidth (NVL72) | 130 TB/s bisectional NVLink Switch bandwidth within rack domain |
| CPU (Grace) | NVIDIA Grace CPU, 72-core Arm Neoverse V2, per CPU die; two Grace dies per GB200 Superchip |
| CPU Memory per Superchip | 480 GB LPDDR5X with 960 GB/s memory bandwidth (CPU + GPU coherent via NVLink-C2C) |
| Peak AI Compute (FP8, NVL72) | Up to 1.44 ExaFLOPS FP8 Tensor Core performance |
| Peak AI Compute (BF16, NVL72) | Up to 720 PetaFLOPS BF16 Tensor Core performance |
| System Interconnect Fabric | HPE Slingshot 11, 200 Gb/s per port, adaptive routing, low-latency congestion control |
| Chassis Form Factor | HPE Cray EX4000 blade chassis, rack-integrated, designed for NVL72 rack unit footprint |
| Cooling Technology | Full Direct Liquid Cooling (DLC); no air-cooled GPU thermal dependency; rear-door heat exchanger compatible |
| Rack-Level TDP | Up to 120 kW per NVL72 rack (liquid-cooled thermal envelope) |
| Storage Integration | HPE Cray ClusterStor E1000 compatible; NVMe-oF over Slingshot; Lustre and DAOS file system support |
| OS and Software Environment | SUSE Linux Enterprise HPC, HPE Cray Programming Environment (CPE), CUDA 12.x, NCCL, OpenMPI, UCX |
| Security Features | Secure boot, TPM 2.0, iLO 6 BMC out-of-band management, role-based access control, FIPS 140-2 module support |
| Management | HPE iLO 6, HPE Cray System Management (CSM), Redfish API, Prometheus/Grafana telemetry integration |
| Power Supply | Redundant high-efficiency AC/DC power shelves, 48V DC distribution to blades, IEC 60309 facility input |
| Scalability | Scales from single cabinet to multi-thousand node installations via Slingshot dragonfly topology |
| Launch Generation | 2025 (Q3 commercial availability) |
The HPE Cray Supercomputing EX4000 (NVIDIA GB200) is built for enterprise data-center workloads — virtualization (VMware, Proxmox, Nutanix), private cloud, database hosting, and AI/ML training. It fits standard EIA-310 server racks and supports redundant PSUs and hot-swap drives common in production environments.
Key specifications for the HPE Cray Supercomputing EX4000 (NVIDIA GB200): new condition; manufacturer HP (HPE); product line HPE Cray Supercomputing EX4000; gpu architecture NVIDIA Blackwell (GB200); nvl72 configuration 72x NVIDIA B200 Tensor Core GPUs + 36x NVIDIA Grace Arm CPUs per NVL72 rack domain; gpu memory per b200 192 GB HBM3e; total gpu memory per nvl72 rack Up to 13,824 GB (13.5 TB) HBM3e unified via NVLink. Manufacturer part number P78420-B21. For the full datasheet with electrical, environmental, and compliance details, contact our pre-sales engineering team.