Super Micro Computer, Inc. announced an expanded portfolio of Open Compute Project Accepted systems, racks, and rear door heat exchangers (RDHx). Supported by DCBBS, the solutions portfolio can be incorporated into complete end-to-end rack- and cluster-scale deployments.
By leveraging OCP community best practices and DCBBS design and services, data center operators can increase compute density and energy efficiency for AI and HPC workloads based on an open and validated rack-scale foundation—from systems and busbar to rack-level and data center cooling—to bring high density solutions online more efficiently to accelerate Time-to-Market (TTM) and TTO.
The ORv3 and ORW racks replace per-server power supplies and cables with a centralized 48V DC busbar, allowing customers to optimize their CPU and GPU systems for higher density compute racks. This density is further supported by Supermicro’s DLC-2 Direct-to-Chip liquid cooling, which removes heat more efficiently from CPUs, GPUs, and other critical components, while residual heat can be removed from the rack by purpose-built ORv3 RDHx’s without requiring additional retrofit of data hall air conditioning systems.
Supermicro RDHx units can be mounted to the rear of ORv3 and ORW racks and remove residual heat from server exhaust before it enters the data hall. The portfolio spans 10 kW to 120 kW of cooling capacity, giving operators a practical way to increase rack density in both new AI facilities and existing data centers. DC-powered models connect to the rack’s busbar to reduce cabling, while designs include N+1 redundant fans to ensure uptime and a smart anti-condensation system to prevent moisture damage to other components inside the rack.
Operators can monitor and manage the entire portfolio of DCBBS components, through SuperCloud Composer using industry-standard protocols. Further, Supermicro’s software, such as SuperCloud Director, SuperCloud Automation Assistant, and SuperCloud Developer Console, enables organizations to manage and operationalize all aspects of modern AI factories, enterprise data centers, and NeoCloud infrastructure, through a unified software control solution.
As part of DCBBS portfolio, Supermicro also delivers a comprehensive portfolio of AI networking solutions, ranging from air and liquid-cooled 800G and 1.6T Ethernet switches including next-generation Co-Packaged Optics (CPO) platforms built on Broadcom Tomahawk 6 switching silicon, optimized for EIA or ORv3 racks. The Supermicro Enterprise Advanced SONiC Networking Operating System gives customers the flexibility to select the networking architecture that best meets their performance, power, scalability, and deployment requirements as AI and large-scale data center environments continue to evolve.
Supermicro DCBBS will also include NVIDIA Spectrum-X Ethernet Photonics switch systems, the first co-packaged Ethernet switches with 200Gb/s SerDes. Built on the 102.4 Tb/s NVIDIA Spectrum-6 Ethernet ASIC, Spectrum-X Ethernet Photonics delivers 5x better power efficiency, 10x greater MTBI, 5x longer uptime for AI applications and 1.3x faster time to deployment, compared to pluggable transceiver-based networks. Spectrum-X Ethernet Photonics is in production now.
Portfolio overview
Supermicro’s broad solution portfolio incorporates many community-based standards, which enable faster and more optimized data centers to be brought online faster.
These include:
1 OU FlexTwin solution: A purpose-built rack-scale HPC solution featuring 1 OU 2-node systems powered by the latest Intel Xeon 6 or AMD EPYC CPUs.
NVIDIA Vera Rubin NVL72: Supermicro’s OCP contributed NVIDIA MGX 1.2 Rack Series is engineered exclusively for NVIDIA Vera Rubin NVL72 and NVIDIA Vera CPU platforms with Supermicro’s DCBBS DLC-2 liquid-cooling stack support to streamline rack integration to accelerate deployment.
2 OU NVIDIA HGX B300: Built to the OCP ORv3 specification with advanced liquid cooling technology, each compact 8-GPU node fits into 21-inch racks, enabling up to 18 nodes and 144 total GPUs per rack.
2 OU 4-Node rack-scale solution: Optimized architectures unlocking breakthrough performance, efficiency, and rack density for 24/7 agentic AI, featuring up to 45,696 Arm AGI CPUs per rack.
250 kW 4U in-rack CDU: Features a high cooling capacity and a compact design with up to 250 kW cooling capacity in a compact 4U form factor. It supports high-density AI racks with 45°C warm-water cooling and energy-efficient data center operation.
OCP ORv3 rack series: 44OU and 48OU 21-inch racks with integrated busbars and a 5,000 lb. (2,268 kg) certified static load capacity, liquid cooling ready architecture, optimized mechanical design, including MGX 1.2 (Gen 3) racks.
AMD Helios platform solution: ORW architecture that is fully integrated and purpose-built for high-volume inference, frontier-model training, fine-tuning, as well as sovereign and agentic AI.
ORv3/ chiller door: Compatible with ORv3 rack and features 20-100 kW cooling capacity, and Redfish/SNMP/WebUI/Modus control.
ORW chiller door: Compatible with ORW rack and features 60 kW cooling capacity and Redfish/SNMP/WebUI/Modus control.
DCBBS coolants: Built for an OCP-Style liquid-cooling stack customers have two ready-to-use PG coolants for the TCS loop — DLC-1 for standard liquid-cooled racks, and ultra-low-conductivity DLC-2 for specialty cold-plate designs. Both are designed to align/compatible with the OCP PG25 Base Spec, so racks, in-rack CDUs, RDHx, and cold plates can run as one cooling stack.
Rear door heat exchanger (RDHx): Six rack-mounted cooling solutions that remove heat from liquid-cooled racks. Cooling capacities range from 10 kW to 120 kW, with AC and DC power options available.
Power shelves (1 OU) and Battery Backup Units (2 OU) easily fit into OCP racks.
2 OU CPO Switch: Designed for OCP ORv3 racks with 40-60 VDC busbar power and blind-mate QD08 liquid cooling. Delivers 102.4 Tbps across 128 × 800G ports with Supermicro Enterprise Advanced SONiC support.
System management software: Supports Redfish APIs and OpenBMC based architecture for remote management. Availability of attestation and PQC algorithms for security compliance with OCP standards.
DCBBS is Supermicro’s integrated approach to building and deploying production-ready AI infrastructure—from rack-scale systems to complete AI factories. DCBBS brings together accelerated compute, high-performance networking, storage, power, advanced liquid cooling, system management software, and deployment services in validated solutions, optimized for, and co-designed by, each customer. By combining Supermicro’s engineering and manufacturing capabilities with leading technology ecosystems, DCBBS helps organizations reduce deployment complexity and total cost of ownership, while optimizing performance and efficiency.
For more information, visit supermicro.com.
The post Supermicro expands OCP portfolio for AI data center deployments appeared first on Engineering.com.