Here is a polished English translation of the Chinese title: **Supermicro Unveils DCBBS Blueprint Powered by NVIDIA Vera Rubin NVL72 and NVIDIA HGX™ Rubin NVL8, Delivering End-to-End Complete Solutions from 5MW to 1GW**

• End-to-End Architecture Blueprint: Offers robust scalability, supporting power scales from 5MW to 1GW, and provides complete data center infrastructure for single-tenant or multi-tenant deployments.
• DLC-2 Direct Liquid Cooling Technology: Designed for nearly 100% heat absorption, exceptional energy efficiency, and low-noise operation. The complete integrated stack includes cold plates, coolant distribution units (CDUs), manifolds, rear door heat exchangers (RDHx), cooling towers, and SMC PG25-A ultra-high-resistance coolant.
• Management Software Suite: The end-to-end SuperCloud software platform provides unified infrastructure management, automated deployment, developer tools, and multi-tenant GPU cloud management capabilities.
• In-Rack, Row-Level, and Site-Level Infrastructure Solutions: Covers the entire deployment process, from rack integration, row-level CDUs, SuperCluster configurations, to site-level infrastructure construction solutions.
• Supermicro Expert Team: Provides service support across the entire project lifecycle, including site surveys, solution design, system integration, deployment implementation, and ongoing after-sales service.
• Support for NVIDIA’s Latest Reference Architecture: Integrates the NVIDIA Context Memory Storage Platform, NVIDIA Spectrum™-X Ethernet, and NVIDIA Quantum-X800 InfiniBand platform.

San Jose, California and TaipeiJune 8, 2026 /PRNewswire/ — Super Micro Computer, Inc. (NASDAQ: SMCI), a total solution provider for AI, enterprise computing, storage, and 5G/edge computing, today announced the launch of its Data Center Building Block Solutions (DCBBS) blueprint based on the NVIDIA Vera Rubin NVL72 and NVIDIA HGX™ Rubin NVL8 platforms. This blueprint is specifically designed for gigawatt-scale AI data center deployments, with a base module consisting of a single scalable unit equipped with 1,152 GPUs, expandable to larger scales based on business needs. The Supermicro DCBBS blueprint encompasses the design and delivery of end-to-end complete solutions, managed by a dedicated team throughout the project lifecycle. DCBBS provides the necessary computing, storage, networking, advanced liquid cooling technology, power distribution, and campus-level infrastructure to accelerate the construction and deployment of large-scale liquid-cooled AI data centers.

Charles Liang, President and CEO of Supermicro, stated: “The NVIDIA Vera Rubin NVL72 platform sets a new benchmark for AI factory performance, and our DCBBS blueprint provides customers with a proven deployment path scalable from 5MW to 1GW. We have already delivered some of the world’s earliest and largest liquid-cooled AI factories, and the practical experience from these projects is embedded in every blueprint, helping customers go from planning and design to full system deployment faster than ever before.”

Supermicro’s DCBBS blueprint is designed to address the real-world challenges of building advanced global AI infrastructure. The NVIDIA Vera Rubin platform significantly increases the performance density of AI factories, multiplying processing power across multiple computing domains. The latest NVIDIA reference architecture defines a scalable unit composed of 1,152 GPUs in detail, while Supermicro’s DCBBS blueprint outlines the specific path to successful deployment, drawing on its project experience in building the world’s largest liquid-cooled AI factories, deploying over 100,000 GPUs.

For more information on DCBBS, please visit: https://www.supermicro.com/en/solutions/dcbbs

PR_Image_NVIDIA_Vera_Rubin_CM_r1

Supermicro’s DCBBS Blueprint Precisely Addresses Real-World Challenges in AI Factory Deployment

When planning new or retrofitted AI factories, customers typically face a fixed constraint: available power. For the DCBBS blueprint targeting the NVIDIA Vera Rubin NVL72 platform, Supermicro provides optimized bill of materials (BOM) configurations tailored to different power load scales (from 5MW to 1GW). The blueprint achieves precise resource balancing among cooling capacity, power delivery, compute nodes, management nodes, high-performance storage nodes, context memory storage platform nodes, and network architecture, avoiding issues such as network oversubscription, power constraints, or thermal throttling that could impact system performance, thereby ensuring optimal performance.

These blueprints cover Supermicro’s complete end-to-end implementation process for delivering large-scale AI projects at industry-leading speed:

• Site Facility Survey: Supermicro’s expert team assesses the actual site based on deployment requirements. Evaluations include: unloading areas and logistics flow, data center room dimensions and ceiling height measurements, floor plan layout, and floor load-bearing capacity assessment. The team also conducts a comprehensive evaluation of existing and planned power and cooling infrastructure to tailor the most suitable construction plan for the customer.
• Project Design and Solution Planning: Develops a customized construction plan containing all key technical details based on the customer’s business needs and site conditions. Supermicro will determine the optimal combination of DCBBS components, including cooling solutions (row-level coolant distribution units with cooling capacity up to 1.8MW for data centers fully supporting direct liquid cooling deployment; liquid-cooled auxiliary cabinet solutions for infrastructure without water supply systems; currently developing in-rack CDU solutions based on 52U rack configurations; optional rear door heat exchangers as auxiliary cooling for high ambient temperature deployment scenarios). The customer will ultimately receive a complete solution with a transparent, detailed bill of materials and a clear deployment timeline.
• Solution Integration and Comprehensive On-Site Services: Supermicro’s integration work begins before equipment is delivered to the site, with most complex assembly and testing completed at its U.S. manufacturing facilities, including in-rack equipment installation, system integration, and cable deployment. Supermicro employs a verification process exceeding industry standards, covering system-level (L10) and cluster-level (L11) multi-node verification, conducting comprehensive system testing. The Supermicro team also manages logistics for key equipment such as CDUs, cooling towers, and power infrastructure, and coordinates with customer-designated third-party partners to advance project implementation. Delivery and on-site integration services include rack positioning, power and liquid cooling system connection, network cabling, system commissioning, software deployment, and overall solution acceptance verification.
• Support Services and Software Platform: Supermicro provides a service support system covering the long-term operational lifecycle of the project, helping customers ensure the stable operation of critical businesses. For mission-critical environments, on-site technical support services with a response time as fast as 4 hours are available. On the software front, Supermicro integrates a complete infrastructure management software suite, including Supermicro SuperCloud Composer® and SuperCloud Director, enabling unified operations management from bare metal management to multi-tenant workload orchestration. It also supports the complete NVIDIA AI software ecosystem, including NVIDIA AI Enterprise and the NVIDIA Run:ai platform. Additionally, asset tracking capabilities ensure real-time access to asset information and sensor data for each CDU and other components.

Supermicro’s DCBBS Blueprint Fully Compatible with NVIDIA Vera Rubin NVL72 Reference Architecture

The NVIDIA Vera Rubin platform promises a leap in AI computing performance, but realizing its full potential requires a replicable, scalable, and reliable deployment solution. Supermicro ensures its solutions are fully compatible with the latest NVIDIA reference architecture, helping customers build AI infrastructure that aligns with the NVIDIA cloud partner ecosystem.

As a core component of Supermicro’s DCBBS blueprint, each scalable unit is equipped with 1,152 NVIDIA Rubin GPUs and provides a total of 331TB of HBM4 high-bandwidth memory. Compared to the NVIDIA Blackwell architecture, Vera Rubin doubles GPU memory bandwidth, GPU-to-GPU NVLink interconnect bandwidth, and per-GPU network bandwidth, providing a solid infrastructure foundation for training and inferencing cutting-edge AI models with trillions of parameters.

• Advanced Direct Liquid Cooling Stack (DLC-2): Includes a 5MW cooling tower, 4 row-level coolant distribution units (each with a cooling capacity of up to 1.8MW), 16 vertically mounted coolant distribution manifolds, and 576 D2C direct-to-chip copper cold plates (1 per compute node). Supermicro SMC PG25-A coolant offers excellent chemical and thermal stability. For deployment environments without liquid cooling infrastructure, Supermicro also offers hybrid air-liquid cooling solutions, including configurations supporting up to 200kW cooling capacity per rack and configurations supporting up to 500kW cooling capacity for dual racks.
• Power Distribution Infrastructure: Covers the complete power supply chain from medium-voltage transformers, low-voltage distribution systems, rack-level power shelves, to battery backup units (BBUs). Each Vera Rubin NVL72 rack is equipped with 4 110kW power shelves, utilizing redundant 18.3kW power modules. The DCBBS solution also supports mission-critical data center deployments, with an optional Supermicro battery energy storage system (BESS) for instantaneous backup power switching.
• Rack Configuration Options: Offers 48U and 52U rack specifications, both optimized for high-density direct liquid cooling scenarios.
• 16 Compute Racks: Optimized for the NVIDIA Vera Rubin NVL72 and NVIDIA HGX™ Rubin NVL8 platforms.
• 6 Network Racks (4 Compute Network Racks, 2 Aggregation Network Racks): Supports NVIDIA Spectrum-X Ethernet or NVIDIA Quantum-X800 InfiniBand network architectures, with compute network bandwidth up to 1.6TB/s. Also offers silicon photonic network solutions supporting co-packaged optics (CPO) technology, further optimizing total cost of ownership (TCO), energy efficiency, and system reliability without pluggable optical modules.
• 4 High-Performance Storage Racks: Built on the Supermicro Petascale server platform, meeting high-performance storage needs such as NVMe tiered application storage and model training checkpoints.
• 2 Context Storage Platform Racks: Designed for long-context inference, agent working memory management, and retrieval-augmented workloads, meeting the large-scale context data access requirements of next-generation AI applications.

For more information, please visit: https://www.supermicro.com/en/accelerators/nvidia/vera-rubin

Supermicro’s DCBBS Blueprint Offers a Single-Vendor Responsibility System

Traditional AI infrastructure construction projects typically require coordinating over a dozen different vendors, covering computing, storage, networking, racks, coolant distribution units, cooling towers, power infrastructure, battery backup systems, cabling, transceivers, and various professional services. When projects involve multiple vendors, each handoff between them can introduce project schedule risks and unclear responsibility delineation, not only potentially impacting deployment timelines but also increasing the complexity of subsequent operations and troubleshooting.

The DCBBS blueprint for the NVIDIA Vera Rubin NVL72 and NVIDIA HGX™ Rubin NVL8 platforms is now open for project inquiries. Related solutions are expected to begin deployment in the second half of 2026, coinciding with the official launch of the NVIDIA Vera Rubin platform.

About Super Micro Computer, Inc.

Supermicro (NASDAQ: SMCI) is a global leader in application-optimized total IT solutions. Headquartered and operating from San Jose, California, Supermicro is committed to delivering market-leading innovative technologies for enterprise, cloud, AI, and 5G telecom/edge IT infrastructure. We are a total IT solution manufacturer providing servers, AI, storage, IoT, switch systems, software, and support services. Supermicro’s expertise in motherboard, power, and chassis design further optimizes our development and production, enabling next-generation innovation from cloud to edge for our global customers. Our products are designed and manufactured in-house (in the US, Asia, and the Netherlands), reducing total cost of ownership (TCO) through product design optimization, minimizing environmental impact through green computing technology, and achieving exceptional manufacturing scale and efficiency through global operations. The award-winning Server Building Block Solutions® portfolio allows customers to select the right model from a highly diverse system product line to optimize performance for their workloads and applications. This diverse system product line is built from highly flexible, reusable building blocks that support various hardware form factors, processors, memory, GPUs, storage, networking, power, and cooling solutions (air conditioning, natural air cooling, or liquid cooling).

Supermicro, Server Building Block Solutions, and We Keep IT Green are trademarks and/or registered trademarks of Super Micro Computer, Inc.

All other brands, names, and trademarks are the property of their respective owners.