Supermicro Launches NVIDIA Vera Rubin NVL72 Racks with Advanced Liquid Cooling Solutions
On September 23, 2026, Supermicro, a prominent player in the IT solutions realm, announced the shipment of its NVIDIA Vera Rubin NVL72 racks, which integrate the company's advanced Data Center Building Block Solutions (DCBBS) and a sophisticated direct liquid cooling stack, designated DLC-2. This cutting-edge technology is designed to meet the growing demands of AI-driven applications and workloads, marking a significant advancement in the deployment of supercomputing architectures.
Supermicro's developments in rack cooling technology have been pivotal for customers aiming to optimize their operations while managing heat loads generated by high-performance computing systems. The liquid cooling stack crafted by Supermicro has been meticulously engineered to handle the immense heat produced during operations, which air cooling methods often fail to manage adequately.
Charles Liang, the CEO of Supermicro, commented on the launch, noting that the company has spent years refining the liquid cooling systems alongside building a dedicated team for deployment. With this announcement, customers can anticipate receiving production-ready systems that are fully integrated and streamlined from the initial design phase right through to on-site deployment. This meticulous attention to detail allows for immediate operational readiness.
The NVIDIA Vera Rubin NVL72 racks are specifically designed with a full liquid-cooling stack, which includes a comprehensive suite of elements such as cold plates, cooling distribution units (CDUs), and tailored cooling towers. Every component within this setup has been optimized to serve the specific needs of the NVIDIA Vera Rubin platform, enhancing operational efficiency. The system can manage a power envelope ranging from 5 MW up to gigawatt scale, accommodating intricate computing clusters that can house 1,152 GPUs and 331 TB of HBM4 memory.
Key technical features of the NVIDIA Vera Rubin NVL72 racks include:
1. Integration of 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs: This configuration operates seamlessly within a single liquid-cooled rack, bringing together eighteen 1U compute trays, each equipped with four Rubin GPUs and two Vera CPUs, utilizing NVIDIA's advanced sixth-generation NVLink switch trays to achieve an astounding 216 TB/s of scale-up bandwidth.
2. In-row cooling distribution units (CDUs): Each rated at 1.8 MW, deployed with redundancy to ensure reliability during high-demand operations.
3. Optional Supermicro Rear Door Heat Exchangers: These units play a critical role in efficiently capturing excess heat, showcasing the innovative mindset of Supermicro towards thermal management.
4. Networking and cabling services: Tailored to meet the needs set out by NVIDIA's Reference Architecture, this ensures a robust AI compute fabric for smooth operation and management.
To facilitate the deployment of the NVL72 racks, Supermicro has developed a decisive DCBBS Blueprint, defining a balanced bill of materials that aligns perfectly with any given power capacity. This Blueprint emphasizes a holistic approach, balancing cooling, power delivery, high-performance storage, and networking demands. A dedicated Supermicro team oversees the project from inception to execution and ongoing support.
In conclusion, the launch of the NVIDIA Vera Rubin NVL72 racks represents a significant milestone for Supermicro as they continue to innovate at the intersection of high-performance computing and scalable AI infrastructure. As companies increasingly rely on AI technologies, Supermicro remains committed to delivering superior solutions, encouraging growth in sectors influenced by computational demand. By enhancing their cooling technology and overall system performance, Supermicro is poised to maintain its position at the forefront of the industry, satisfying the ever-evolving needs of global clients involved in AI, cloud computing, and 5G.
Learn more about how Supermicro's innovative solutions can help scale your next-generation AI infrastructure efficiently.
Advancements in Cooling Technology
Supermicro's developments in rack cooling technology have been pivotal for customers aiming to optimize their operations while managing heat loads generated by high-performance computing systems. The liquid cooling stack crafted by Supermicro has been meticulously engineered to handle the immense heat produced during operations, which air cooling methods often fail to manage adequately.
Charles Liang, the CEO of Supermicro, commented on the launch, noting that the company has spent years refining the liquid cooling systems alongside building a dedicated team for deployment. With this announcement, customers can anticipate receiving production-ready systems that are fully integrated and streamlined from the initial design phase right through to on-site deployment. This meticulous attention to detail allows for immediate operational readiness.
Features of the NVL72 Racks
The NVIDIA Vera Rubin NVL72 racks are specifically designed with a full liquid-cooling stack, which includes a comprehensive suite of elements such as cold plates, cooling distribution units (CDUs), and tailored cooling towers. Every component within this setup has been optimized to serve the specific needs of the NVIDIA Vera Rubin platform, enhancing operational efficiency. The system can manage a power envelope ranging from 5 MW up to gigawatt scale, accommodating intricate computing clusters that can house 1,152 GPUs and 331 TB of HBM4 memory.
Key technical features of the NVIDIA Vera Rubin NVL72 racks include:
1. Integration of 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs: This configuration operates seamlessly within a single liquid-cooled rack, bringing together eighteen 1U compute trays, each equipped with four Rubin GPUs and two Vera CPUs, utilizing NVIDIA's advanced sixth-generation NVLink switch trays to achieve an astounding 216 TB/s of scale-up bandwidth.
2. In-row cooling distribution units (CDUs): Each rated at 1.8 MW, deployed with redundancy to ensure reliability during high-demand operations.
3. Optional Supermicro Rear Door Heat Exchangers: These units play a critical role in efficiently capturing excess heat, showcasing the innovative mindset of Supermicro towards thermal management.
4. Networking and cabling services: Tailored to meet the needs set out by NVIDIA's Reference Architecture, this ensures a robust AI compute fabric for smooth operation and management.
Deployment Support and Future Prospects
To facilitate the deployment of the NVL72 racks, Supermicro has developed a decisive DCBBS Blueprint, defining a balanced bill of materials that aligns perfectly with any given power capacity. This Blueprint emphasizes a holistic approach, balancing cooling, power delivery, high-performance storage, and networking demands. A dedicated Supermicro team oversees the project from inception to execution and ongoing support.
In conclusion, the launch of the NVIDIA Vera Rubin NVL72 racks represents a significant milestone for Supermicro as they continue to innovate at the intersection of high-performance computing and scalable AI infrastructure. As companies increasingly rely on AI technologies, Supermicro remains committed to delivering superior solutions, encouraging growth in sectors influenced by computational demand. By enhancing their cooling technology and overall system performance, Supermicro is poised to maintain its position at the forefront of the industry, satisfying the ever-evolving needs of global clients involved in AI, cloud computing, and 5G.
Learn more about how Supermicro's innovative solutions can help scale your next-generation AI infrastructure efficiently.
Topics Consumer Technology）
【About Using Articles】
You can freely use the title and article content by linking to the page where the article is posted.
※ Images cannot be used.
【About Links】
Links are free to use.