Hewlett Packard Enterprise Ships First Nvidia Grace Blackwell System For Large AI Deployments

Hewlett Packard Enterprise announced that it has shipped its first Nvidia Blackwell family-based solution, the GB200 NVL72. This rack-scale system is designed to help service providers and enterprises quickly deploy large, complex artificial intelligence clusters with advanced, direct liquid cooling solutions to optimise efficiency and performance. 

The GB200 NVL72 features shared-memory, low-latency architecture with the latest GPU technology designed for extremely large AI models of over a trillion parameters, in one memory space. The system offers integration of Nvidia CPUs, GPUs, compute and switch trays, networking, and software, addressing heavy workloads like generative AI model training and inferencing, along with Nvidia software applications. 

“AI service providers and large enterprise model builders are under tremendous pressure to offer scalability, extreme performance, and fast time-to-deployment,” said Trish Damkroger, senior vice president and general manager of HPC & AI Infrastructure Solutions, HPE. “HPE offers customers lower cost per token training and best-in-class performance with industry-leading services expertise.” 

The system is equipped with 72 Nvidia Blackwell GPUs and 36 Nvidia Grace CPUs interconnected via high-speed Nvidia NVLink, and packs up to 13.5 TB total HBM3e memory with 576 TB/sec bandwidth. 

It also features HPE direct liquid cooling technology, which is better than air-cooled systems, to keep up with increasing power and compute requirements.

HPE supports massive, custom AI clusters with services including expert on-site support, customised services, sustainability services and more. The company provides enhanced incident management with proactive support through dedicated remote engineers, ensuring rapid installation and faster time-to-value. 

According to HPE, its services include:

Onsite Engineering Resources: On-site support through trained resident engineers who work closely with a customer’s IT teams to ensure optimal system performance and availability. 

Performance And Benchmarking Engagements: Team of experts to fine tune solutions throughout the life of a system.  

Sustainability Services: Energy and emissions reporting, sustainability workshops, and resource monitoring to reduce environmental impact.

“Building on continued collaboration between HPE and Nvidia, HPE’s first shipment of Nvidia GB200 NVL72 will help service providers and large enterprises efficiently build, deploy and scale large AI clusters,” said Bob Pette, vice president of enterprise platforms at Nvidia.