Supermicro’s DCBBS: A Novel and Smart-Way to Keep AI Liquid and Data-Reaching Shelves Flat and Cool.
While artificial intelligence (AI), high-performance computing (HPC), and data-intense workloads are transforming industries, traditional data centers are facing challenges in keeping up with the speed, scale and sustainability requirements. Opening the door to this, Supermicro has designed arguably the most revolutionary aspect in modern infrastructure:
DCBBS (Data Center Building Blocks Solutions)—a liquid-cooled, modular, factory-integrated platform that transforms the way AI infrastructure is designed, delivered and scaled.
From edge AI clusters to hyperscale GPU-powered supercomputers, whether at lab scale, or attempting to add thousands of servers to an application on the fly, DCBBS® is the plug-and-play approach for an enterprise to scale intelligently and sustainably.
What is DCBBS?
DCBBS stands for Data Center Building Blocks Solutions, a flexible, customizable system of pre-integrated rack-scale servers, storage, networking, cooling and power distribution, designed and optimized for high performance workloads.
Every solution is designed to meet specific application requirements and arrives as a rack-ready system – reducing deployment times from weeks to hours.
Core Components of DCBBS
GPU Optimized Servers: Support NVIDIA HGX H100/H200 with NVLink and NVSwitch to deliver better inter-GPU bandwidth.
Storage Building Blocks: NVMe, hybrid and all flash storage provisions support built directly into the rack.
High-Speed Networking: 400Gbps+ InfiniBand and Ethernet options for GPU Direct RDMA and low-latency data movement.
Liquid Cooling Infrastructure: Designed for the data center, with direct-to-chip and immersion capable options to support overclocking while keeping thermals in check.
High-performance AI and ML workloads produce a ridiculous amount of heat — especially in a GPU-heavy environment. Supermicro pioneered DCBBS to optimize cooling efficiencies, which in turn allow these Cool IT enabled systems to run faster and cooler, enabling more work to get done and reducing energy waste.
Types of Cooling Supported:
Direct Liquid Cooling (DLC): Providing direct cooling of the processor, video cards, and memory, DLC shares benefits of increased performance with all three main heat producing components of a server with air cooled solutions, such as CPU, GPU, and memory up to 40% power reduction.
Rear-Door Heat Exchangers (RDHx): Captures the hot air then exits from the rack by cooled water loops.
Compatible With Immersion Cooling: Designed for the immersion of entire board into various dielectric liquids for more efficient cooling with less noise.
Results:
- Up to 95% heat dissipation efficiency
- Low PUE (Power Usage Effectiveness) even below 1.1
- Significant noise reduction
- Improved hardware lifespan
Designed from the Ground Up for AI, ML and Deep Learning at Scale
DCBBS is designed for multi-petaflop workloads and large model training scenarios:
- NVIDIA HGX Platforms with 8x H100 or H200 GPUs per node.
- NVLink™ and NVSwitch™ for up to 900 GB/sec GPU-to-GPU communication.
- 24 DDR5 DIMMs per node on more than 6TB RAM.
- PCIe 5.0 for ultra-fast transfer rates between GPUs, SSDs, and FPGAs.
- Pre-integrated AI SuperClusters for LLM, NLP, and generative AI tasks.
From silicon to system, no other DDR solution is architected to the same level as DCBBS for your most challenging AI applications.
Modular by Nature: Construct What You Require
DCBBS is the most customizable, where a firm can deploy a solution customized to a given workload.
Flexible Configurations:
- Training and inference GPU clusters
- Nodes with lots of storage for big data or analytics
- Compact edge AI racks
- Hybrid on-prem/cloud environments
This modular design means that customers can bolt on more compute, storage, cooling, and networking without fundamentally rethinking your data center.
Sustainable Infrastructure with Reduced TCO
Sustainability evolving into a strategic imperative for IT and data center leaders. DCBBS Tech That Is More Green Without Compromise — for once you really can have it all!
Sustainability Highlights:
- Lower power consumption thanks to efficient liquid cooling
- Eliminates need for HVAC in high density deployments
- Allows modular components to be reused and upgraded
- Lowers TCO due to enhanced energy efficiency and accelerated deployment cycles
Case Study:
A Tier-2 Japan-based data center deployed Supermicro’s DCBBS with direct liquid cooling and realized a 31% reduction in power consumption and 70% reduction in HVAC (heating, ventilation, and air conditioning) operational costs while increasing AI computing density per rack by 2X.
Rapid Deployment: Ready-to-Run Infrastructure
Legacy infrastructures can require weeks to months to construct, configure, and prove out. With DCBBS, Supermicro offers:
- Factory-integrated racks
- Pre-tested and validated configurations
- “Ready for use” wired systems with labeling and documentation
This plug-and-play process enables setup in hours – versus days, weeks or months – to get to production and to gain insights as quickly as possible.
Industries and Use Cases
The DCBBS is applicable in variety of domains:
- AI model training: LLMs, visual models, speech synthesis
- Independently driven robotics, controlled cars, UAVs (unmanned aerial vehicles)
- Life sciences: genomics, bioinformatics, medical AI
- Finance: quant modeling, fraud detection, real-time analytics
- Manufacturing: predictive maintenance, visual inspection
- Media: rendering, video analytics, CGI, animation etc
- Cloud providers: scalable, multi-tenant AI-as-a-service platforms
Why DCBBS is the Future of AI Infrastructure
Supermicro’s DCBBS is more than just hardware — it’s a reimagining of what scalable, sustainable, and high-performance infrastructure should look like in the AI-first era. Whether you’re building a massive GPU supercluster or deploying edge AI nodes, DCBBS offers the speed, flexibility, and efficiency your workloads demand. It’s time to go beyond traditional data center models — with DCBBS, you’re not just keeping up, you’re setting the pace.
To know more about Supermicro Services, click here on Supermicro