Hewlett Packard Enterprise has officially shipped its first NVIDIA Blackwell family-based solution, the NVIDIA GB200 NVL72, marking a significant advancement in AI infrastructure. This high-performance rack-scale system is built to help service providers and enterprises rapidly deploy large, complex AI clusters while optimizing efficiency with industry-leading direct liquid cooling technology.
The NVIDIA GB200 NVL72 introduces shared-memory, low-latency architecture, enabling AI models with over a trillion parameters to operate in a unified memory space. The system integrates NVIDIA CPUs, GPUs, networking, and software, delivering cutting-edge performance for AI training, inference, and generative AI (GenAI) applications.
“Building on our strong collaboration with NVIDIA, the shipment of the first NVIDIA GB200 NVL72 by HPE provides AI leaders with the tools to efficiently build, deploy, and scale large AI clusters,”
Bob Pette, Vice President of Enterprise Platforms at NVIDIA.
Key Features of NVIDIA GB200 NVL72 by HPE
- 72 NVIDIA Blackwell GPUs and 36 NVIDIA Grace CPUs interconnected via high-speed NVIDIA NVLink
- 13.5 TB HBM3e memory with 576 TB/sec bandwidth
- HPE direct liquid cooling technology for enhanced energy efficiency
“AI service providers and large enterprises need scalable, high-performance solutions that can be deployed quickly.”
Trish Damkroger, SVP & GM, HPC & AI Infrastructure Solutions, HPE
Comprehensive AI Services & Support
HPE offers global AI infrastructure solutions, including:
- Onsite engineering support with expert resident engineers
- Performance tuning and benchmarking services
- Sustainability solutions for energy efficiency and carbon footprint reduction
With a 50-year legacy in liquid cooling, HPE has delivered eight of the top 15 energy-efficient supercomputers on the Green500 list. The NVIDIA GB200 NVL72 by HPE is part of an extensive portfolio supporting AI, scientific computing, and high-performance computing (HPC) workloads.