HPE Announces Shipment of Its First NVIDIA Grace Blackwell System
Hewlett Packard Enterprise (NYSE:HPE) has announced the shipment of its first NVIDIA Blackwell family-based solution, the NVIDIA GB200 NVL72. This rack-scale system features direct liquid cooling and is designed for service providers and enterprises deploying large AI clusters.
The GB200 NVL72 includes 72 NVIDIA Blackwell GPUs and 36 NVIDIA Grace CPUs, offering up to 13.5 TB total HBM3e memory with 576 TB/sec bandwidth. The system is specifically designed for AI models exceeding a trillion parameters in one memory space.
HPE, which has built seven of the top 10 world's fastest supercomputers, brings five decades of liquid cooling expertise to this solution. The company offers comprehensive support services, including on-site engineering resources, performance optimization, and sustainability services to help customers maximize system efficiency and performance.
Hewlett Packard Enterprise (NYSE:HPE) ha annunciato la spedizione della sua prima soluzione basata sulla famiglia NVIDIA Blackwell, il NVIDIA GB200 NVL72. Questo sistema a rack presenta un raffreddamento a liquido diretto ed è progettato per fornitori di servizi e aziende che implementano grandi cluster di intelligenza artificiale.
Il GB200 NVL72 include 72 GPU NVIDIA Blackwell e 36 CPU NVIDIA Grace, offrendo fino a 13,5 TB di memoria totale HBM3e con una larghezza di banda di 576 TB/sec. Il sistema è specificamente progettato per modelli di intelligenza artificiale che superano un trilione di parametri in uno spazio di memoria.
HPE, che ha costruito sette dei 10 supercomputer più veloci al mondo, porta con sé cinque decenni di esperienza nel raffreddamento a liquido per questa soluzione. L'azienda offre servizi di supporto completi, inclusi risorse ingegneristiche in loco, ottimizzazione delle prestazioni e servizi di sostenibilità per aiutare i clienti a massimizzare l'efficienza e le prestazioni del sistema.
Hewlett Packard Enterprise (NYSE:HPE) ha anunciado el envío de su primera solución basada en la familia NVIDIA Blackwell, el NVIDIA GB200 NVL72. Este sistema a nivel de rack cuenta con refrigeración líquida directa y está diseñado para proveedores de servicios y empresas que implementan grandes clústeres de inteligencia artificial.
El GB200 NVL72 incluye 72 GPU NVIDIA Blackwell y 36 CPU NVIDIA Grace, ofreciendo hasta 13,5 TB de memoria HBM3e total con un ancho de banda de 576 TB/sec. El sistema está diseñado específicamente para modelos de inteligencia artificial que superan un billón de parámetros en un solo espacio de memoria.
HPE, que ha construido siete de los 10 supercomputadores más rápidos del mundo, aporta cinco décadas de experiencia en refrigeración líquida a esta solución. La empresa ofrece servicios de soporte integral, incluidos recursos de ingeniería en el sitio, optimización del rendimiento y servicios de sostenibilidad para ayudar a los clientes a maximizar la eficiencia y el rendimiento del sistema.
휴렛팩커드 엔터프라이즈 (NYSE:HPE)는 첫 번째 NVIDIA 블랙웰 패밀리 기반 솔루션인 NVIDIA GB200 NVL72의 배송을 발표했습니다. 이 랙 규모 시스템은 직접 액체 냉각 기능을 갖추고 있으며 대규모 AI 클러스터를 배포하는 서비스 제공업체 및 기업을 위해 설계되었습니다.
GB200 NVL72에는 72개의 NVIDIA 블랙웰 GPU와 36개의 NVIDIA 그레이스 CPU가 포함되어 있으며, 총 13.5 TB의 HBM3e 메모리와 576 TB/sec의 대역폭을 제공합니다. 이 시스템은 한 메모리 공간에서 1조 개 이상의 매개변수를 초과하는 AI 모델을 위해 특별히 설계되었습니다.
HPE는 세계에서 가장 빠른 10대 슈퍼컴퓨터 중 7대를 구축한 경력을 가지고 있으며, 이 솔루션에 액체 냉각 전문성을 50년 이상 제공합니다. 이 회사는 현장 엔지니어링 리소스, 성능 최적화 및 지속 가능성 서비스를 포함한 종합적인 지원 서비스를 제공하여 고객이 시스템의 효율성과 성능을 극대화할 수 있도록 돕습니다.
Hewlett Packard Enterprise (NYSE:HPE) a annoncé l'expédition de sa première solution basée sur la famille NVIDIA Blackwell, le NVIDIA GB200 NVL72. Ce système de type rack dispose d'un refroidissement liquide direct et est conçu pour les fournisseurs de services et les entreprises déployant de grands clusters d'intelligence artificielle.
Le GB200 NVL72 comprend 72 GPU NVIDIA Blackwell et 36 CPU NVIDIA Grace, offrant jusqu'à 13,5 To de mémoire HBM3e totale avec une bande passante de 576 To/s. Le système est spécifiquement conçu pour des modèles d'IA dépassant un trillion de paramètres dans un seul espace mémoire.
HPE, qui a construit sept des dix superordinateurs les plus rapides au monde, apporte cinq décennies d'expertise en refroidissement liquide à cette solution. L'entreprise propose des services de support complets, y compris des ressources d'ingénierie sur site, l'optimisation des performances et des services de durabilité pour aider les clients à maximiser l'efficacité et les performances du système.
Hewlett Packard Enterprise (NYSE:HPE) hat den Versand seiner ersten auf der NVIDIA Blackwell-Familie basierenden Lösung, dem NVIDIA GB200 NVL72, angekündigt. Dieses Rack-System verfügt über eine direkte Flüssigkeitskühlung und ist für Dienstanbieter und Unternehmen konzipiert, die große KI-Cluster bereitstellen.
Das GB200 NVL72 umfasst 72 NVIDIA Blackwell GPUs und 36 NVIDIA Grace CPUs und bietet bis zu 13,5 TB Gesamtspeicher HBM3e mit einer Bandbreite von 576 TB/s. Das System ist speziell für KI-Modelle konzipiert, die mehr als eine Billion Parameter in einem Speicherplatz überschreiten.
HPE, das sieben der zehn schnellsten Supercomputer der Welt gebaut hat, bringt fünf Jahrzehnte Expertise in der Flüssigkeitskühlung in diese Lösung ein. Das Unternehmen bietet umfassende Unterstützungsdienste an, einschließlich Ingenieure vor Ort, Leistungsoptimierung und Nachhaltigkeitsdienste, um den Kunden zu helfen, die Effizienz und Leistung des Systems zu maximieren.
- First-to-market with NVIDIA's latest Blackwell GPU technology
- Industry-leading position in liquid cooling technology with 7 of top 10 fastest supercomputers
- Significant memory capabilities with 13.5 TB total HBM3e memory and 576 TB/sec bandwidth
- Comprehensive service offering including on-site support and performance optimization
- None.
Insights
HPE's first shipment of the NVIDIA GB200 NVL72 system marks a strategic advancement in the $165 billion AI infrastructure market. The timing is particularly significant as enterprises rapidly scale their AI capabilities, with the system's specifications directly addressing the computational demands of trillion-parameter AI models that are becoming industry standard.
The integration of HPE's direct liquid cooling technology represents a important competitive advantage. With data center power consumption reaching critical limits, this cooling solution enables higher compute density and lower operating costs - key differentiators that could help HPE capture a larger share of the enterprise AI infrastructure market. Their proven track record in building seven of the world's top 10 fastest supercomputers demonstrates unmatched expertise in deploying and managing complex computing systems.
The comprehensive service offering accompanying the GB200 NVL72 creates multiple revenue streams beyond the initial hardware sale. The inclusion of sustainability services, performance optimization, and dedicated engineering support positions HPE to benefit from long-term service contracts, potentially improving profit margins and revenue predictability. This is particularly valuable as enterprises typically require ongoing support for these complex systems.
HPE's partnership with NVIDIA, especially in being among the first to ship Blackwell-based systems, strengthens their position in the enterprise AI market. The focus on extreme scalability and performance optimization directly addresses the needs of AI service providers and large enterprises, two segments experiencing rapid growth in AI infrastructure spending. This strategic alignment with market demands could accelerate HPE's revenue growth in their high-performance computing segment.
GB200 NVL72 featuring industry-leading direct liquid cooling now available
“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. “As builders of the world’s top three fastest systems with direct liquid cooling, HPE offers customers lower cost per token training and best-in-class performance with industry-leading services expertise.”
The NVIDIA 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. GB200 NVL72 offers seamless integration of NVIDIA CPUs, GPUs, compute and switch trays, networking, and software, bringing together extreme performance to address heavily parallelizable workloads, like generative AI (GenAI) model training and inferencing, along with NVIDIA software applications.
“Engineers, scientists and researchers need cutting-edge liquid cooling technology to keep up with increasing power and compute requirements,” said Bob Pette, vice president of enterprise platforms at NVIDIA. “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.”
With escalating power requirements and data center density dynamics, HPE has five decades of liquid cooling expertise that uniquely positions the company to help customers bring fast deployment and an extensive infrastructure support system for complex liquid-cooled environments. This experience has enabled HPE to deliver eight of the top 15 supercomputers on the Green500 list, which ranks the world’s most energy-efficient supercomputers. HPE is recognized as a leader in direct liquid cooling technology, having built seven of the top 10 world’s fastest supercomputers.
Features of NVIDIA GB200 NVL72 by HPE:
- 72 NVIDIA Blackwell GPUs and 36 NVIDIA Grace CPUs interconnected via high-speed NVIDIA NVLink
- Up to 13.5 TB total HBM3e memory with 576 TB/sec bandwidth
- HPE direct liquid cooling technology
Industry Leading Services and Support:
HPE is able to deliver AI solutions at a global scale, with proven ability to support massive, custom AI clusters with superior serviceability including expert on-site support, customized services, sustainability services and more. HPC & AI Custom Support Services are tailored to meet customer needs. With several levels of SLA coverage, HPE provides enhanced incident management with proactive support through dedicated remote engineers, ensuring rapid installation and faster time-to-value. Available services include:
- Onsite engineering resources: Comprehensive on-site support through highly trained resident engineers who work closely with a customer’s IT teams to ensure optimal system performance and availability.
- Performance and benchmarking engagements: Industry-leading 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.
The newly shipped NVIDIA GB200 NVL72 by HPE is one of a wide array of high-performance computing and supercomputing systems that address every use case for GenAI, scientific discovery, and other compute-intensive workloads. Learn more about our compute and supercomputing systems and other solutions in the NVIDIA AI Computing by HPE portfolio.
About Hewlett Packard Enterprise
Hewlett Packard Enterprise (NYSE: HPE) is a global technology leader focused on developing intelligent solutions that allow customers to capture, analyze, and act upon data seamlessly. The company innovates across networking, hybrid cloud, and AI to help customers develop new business models, engage in new ways, and increase operational performance. For more information, visit: www.hpe.com.
View source version on businesswire.com: https://www.businesswire.com/news/home/20250213323241/en/
Media Contact:
Cristina Thai
cristina.thai@hpe.com
Source: Hewlett Packard Enterprise
FAQ
What are the key specifications of HPE's new NVIDIA GB200 NVL72 system?
How does HPE's liquid cooling expertise benefit the GB200 NVL72 system?
What support services does HPE offer for the GB200 NVL72?
What is the target market for HPE's GB200 NVL72 system?
What is the memory capacity of HPE's GB200 NVL72 system?
