Supermicro Provides Portfolio for Upcoming Stream of AI with NVIDIA Blackwell Extremely Answers, That includes NVIDIA HGX™ B300 NVL16 and GB300 NVL72
Uncategorized

Supermicro Provides Portfolio for Upcoming Stream of AI with NVIDIA Blackwell Extremely Answers, That includes NVIDIA HGX™ B300 NVL16 and GB300 NVL72


Breeze- and Liquid-Cooled Optimized Answers with Enhanced AI FLOPs and HBM3e Capability, with as much as 800 Gb/s Direct-to-GPU Networking Efficiency

SAN JOSE, Calif., March 18, 2025 /PRNewswire/ — GTC 2025 Convention Supermicro, Inc. (NASDAQ: SMCI), a General IT Answer Supplier for AI, Cloud, Bank, and 5G/Edge, is pronouncing unused methods and rack answers powered by means of the NVIDIA’s Blackwell Extremely platform, that includes the NVIDIA HGX B300 NVL16 and NVIDIA GB300 NVL72 platforms. Supermicro and NVIDIA’s unused AI answers reinforce management in AI by means of handing over leap forward functionality for essentially the most compute-intensive AI workloads, together with AI reasoning, agentic AI, and video inference programs.

“At Supermicro, we are excited to continue our long-standing partnership with NVIDIA to bring the latest AI technology to market with the NVIDIA Blackwell Ultra Platforms,” mentioned Charles Liang, president and CEO, Supermicro. “Our Data Center Building Block Solutions® approach has streamlined the development of new air and liquid-cooled systems, optimized to the thermals and internal topology of the NVIDIA HGX B300 NVL16 and GB300 NVL72. Our advanced liquid-cooling solution delivers exceptional thermal efficiency, operating with 40℃ warm water in our 8-node rack configuration, or 35℃ warm water in double-density 16-node rack configuration, leveraging our latest CDUs. This innovative solution reduces power consumption by up to 40% while conserving water resources, providing both environmental and operational cost benefits for enterprise data centers.”

For more info, please consult with https://www.supermicro.com/en/accelerators/nvidia 

NVIDIA’s Blackwell Extremely platform is constructed to triumph over essentially the most tough cluster-scale AI programs by means of overcoming functionality bottlenecks brought about by means of restricted GPU reminiscence capability and community bandwidth. NVIDIA Blackwell Extremely delivers an extraordinary 288GB HBM3e of reminiscence consistent with GPU, handing over drastic enhancements in AI FLOPS for AI coaching and inference for the most important AI fashions. The networking platform integration with NVIDIA Quantum-X800 InfiniBand and Spectrum-X™ Ethernet doubles the compute cloth bandwidth, as much as 800 Gb/s.

Supermicro integrates NVIDIA Blackwell Extremely with two sorts of answers: Supermicro NVIDIA HGX B300 NVL16 methods, designed for each and every information middle, and the NVIDIA GB300 NVL72, provided with NVIDIA’s next-generation Grace Blackwell structure.

Supermicro NVIDIA HGX B300 NVL16 device

Supermicro NVIDIA HGX methods are the industry-standard construction blocks for AI coaching clusters, with an 8-GPU NVIDIA NVLink™ area and 1:1 GPU-to-NIC ratio for high-performance clusters. Supermicro’s unused NVIDIA HGX B300 NVL16 device builds upon this confirmed structure with thermal design developments in each a liquid-cooled and air-cooled model.

For B300 NVL16, Supermicro introduces a emblem unused 8U platform to maximise the output of the NVIDIA HGX B300 NVL16 board. Each and every GPU is hooked up in a 1.8TB/s 16-GPU NVLink area offering a immense 2.3TB of HBM3e consistent with device. Supermicro NVIDIA HGX B300 NVL16 improves upon functionality within the community area by means of integrating 8 NVIDIA ConnectX®-8 NICs at once into the baseboard to help 800 Gb/s node-to-node speeds by means of NVIDIA Quantum-X800 InfiniBand or Spectrum-X™ Ethernet.

Supermicro NVIDIA GB300 NVL72

The NVIDIA GB300 NVL72 integrates 72 NVIDIA Blackwell Extremely GPUs and 36 NVIDIA Grace™ CPUs in one rack with exascale computing capability, that includes upgraded HBM3e reminiscence capability for over 20TB of HBM3e reminiscence interconnected in a 1.8TB/s 72-GPU NVLink area. NVIDIA ConnectX®-8 SuperNIC supplies 800Gb/s speeds for each GPU-to-NIC and NIC-to-network communique, significantly making improvements to cluster-level functionality of the AI compute cloth.

Liquid-Cooled AI Knowledge Middle Construction Prevent Answers

Experience in liquid cooling, information middle deployment, and construction oppose way positions Supermicro in order NVIDIA Blackwell Extremely with industry-leading time-to-deployment. Supermicro offer an entire liquid cooling portfolio, together with newly evolved direct-to-chip chilly plates, a 250kW in-rack CDU, and cooling tower.

Supermicro’s on-site rack deployment is helping enterprises develop information middle from the farmland up, together with the making plans, design, power-up, validation, checking out, set up and configuration of racks, servers, switches and alternative networking apparatus to fulfill the group’s particular wishes.

8U Supermicro NVIDIA HGX B300 NVL16 device – Designed for each and every information middle with a streamlined thermally-optimized chassis and a pair of.3TB HBM3e reminiscence consistent with device.

NVIDIA GB300 NVL72 – Exascale AI supercomputer in one rack with necessarily double the HBM3e reminiscence capability and networking speeds over its predecessor.

Supermicro at GTC 2025

GTC guests can to find Supermicro at San Jose, CA from March 17-21, 2025. Discuss with us at sales space #1115 to peer the X14/H14 B200, B300, and GB300 methods on show along side our rack-scaled liquid-cooled answers.

About Tremendous Micro Pc, Inc.

Supermicro (NASDAQ: SMCI) is a world chief in Utility-Optimized General IT Answers. Based and working in San Jose, California, Supermicro is dedicated to handing over first to marketplace innovation for Undertaking, Cloud, AI, and 5G Telco/Edge IT Infrastructure. We’re a General IT Answers supplier with server, AI, store, IoT, transfer methods, tool, and help services and products. Supermicro’s motherboard, capacity, and chassis design experience additional permits our building and manufacturing, enabling subsequent era innovation from cloud to edge for our international consumers. Our merchandise are designed and manufactured in-house (in the United States, Taiwan, and the Netherlands), leveraging international operations for scale and potency and optimized to enhance TCO and let go environmental have an effect on (Inexperienced Computing). The award-winning portfolio of Server Construction Prevent Answers® lets in consumers to optimize for his or her precise workload and alertness by means of deciding on from a wide public of methods constructed from our versatile and reusable construction blocks that help a complete eager of method elements, processors, reminiscence, GPUs, store, networking, capacity, and cooling answers (air-conditioned, distant breeze cooling or liquid cooling).  

Supermicro, Server Construction Prevent Answers, and We Hold IT Inexperienced are emblems and/or registered emblems of Tremendous Micro Pc, Inc.

All alternative manufacturers, names, and emblems are the quality in their respective house owners.

SOURCE Tremendous Micro Pc, Inc.



Source link