As industries and enterprises find new applications for AI, the deployment of AI infrastructure focused on inference has expanded from the data center to the edge. Edge AI aims to locate the AI inference hardware closer to the user to reduce the communication delay in obtaining the inference results. This builds upon the enterprise data center infrastructure which uses the company’s proprietary data to train and fine-tune the enterprise-specific AI models.
Once the models are developed in the data center, they can be deployed using power and space-efficient hardware suitable for the location such as a factory floor, a fast-food restaurant drive-through, a self-service order kiosk, or an intelligent store. These locations often are an IT equipment closet or even mounted on a wall next to the cashier. Each deployment requires environment-specific solutions such as fan-less operation due to airborne contaminants and noise limitations, low power or high temperature operations, small-form factor, or mounting location limitations.
Supermicro has long been known for offering customers the most comprehensive selection of optimized server systems and we’ve just announced a number of new high-performance and compact edge systems that deliver performance right where you need it — at the edge and on your desk.
Let’s look at a few of Supermicro’s latest edge server and client systems that are powered by the latest CPUs:
This edge system features the new Intel® Core™ Ultra Series 3 processors (codenamed Panther Lake) with up to 12 Xe3 GPUs and NPU 5.0, enabling a performance of 180 TOPS (Trillions of Operations Per Second), making it an ideal solution for robotics and AI at the edge deployments.
Optimized for retail, security, manufacturing, healthcare, and more. The new compact edge systems are powered by AMD EPYC™ 4005 processors to provide high-performance cores for smaller focused operations. The short 1U form factor, up to 16 cores, delivers maximum performance-per-watt, and with features like a compact build format and IPMI with dedicated port, it is ideal for virtualized workloads at the edge.
Supermicro’s portfolio also includes a versatile lineup of edge AI and networking servers that are NEBS compliant, designed for networking and telco applications, and support the latest AMD EPYC™ 8000 series CPUs to deliver more energy efficiency and higher-core performance.
A 2U PCIe GPU system built around a single Intel® Xeon® 6700 series processor that can pack up to 86 high-performance cores. This server is a powerhouse for applications like scientific research, model analysis, high-performance computing (HPC), edge AI inference, and even financial workloads — it’s basically the perfect platform when you need serious computing power. It has sixteen DIMM slots ready for DDR5-6400 RDIMMs or DDR5-8000 MRDIMMs, so it can handle memory-hungry tasks with ease. Its efficient heat dissipation design enables support for up to four double width 600W GPUs (with NVIDIA NVLink™ support) reliably at 35°C ambient temperature.
This new platform has been specifically designed to handle demanding AI applications, including large-scale local inference and generative AI projects, delivering substantial computational capabilities directly to your desk.
At the heart of this system is the Intel® Core™ Ultra Series 2 CPU, which features a built-in Neural Processing Unit (NPU) for efficient AI acceleration. Paired with NVIDIA’s cutting-edge GeForce RTX™ 50 series discrete GPUs (based on the Blackwell architecture), this design delivers exceptional performance for running complex models locally.
This system is highly expandable too, with support for large DDR5 memory capacities — up to 256GB across four DIMM slots — to handle the huge datasets common in AI inference workloads. On the storage front, it includes dedicated M.2 PCIe 5.0 NVMe slots, matching the massive data throughput needs of modern AI workflows.
The new client systems feature the AMD® Ryzen 9000 Processor in an AM5 socket, up to 8 cores, 120W TDP. The new slim designed systems come preinstalled with Windows 11 Home in a compact case with a sleek and features a compact, sleek, minimalist design, optimized for AI applications in both office and personal use.
Powered by an Intel® Xeon® 6 SoC, this new workstation can deliver agentic AI performance with support for a wide range of GPUs and large memory capacity. Built-in Media Transcoding Acceleration for content delivery networks (CDNs) and dual 100GbE networking provide a platform that can be used for a variety of use cases, including software-defined broadcasting, Video-On-Demand, CDN, live streaming, and virtual desktop interface (VDI) services.
Complementing the edge-optimized servers are new Supermicro workstations which have been designed for AI model development and testing. The latest Supermicro workstations and rackmount workstations also feature a lineup of AMD’s Ryzen Threadripper Pro 9000 CPUs (up to 96 cores) with overclocking features and supports dual NVIDIA RTX PRO 6000 GPUs. Suited towards ML engineers and technical computing professionals focused on multi-threaded AI data preprocessing, mid-sized model training, and inference workloads. The AS-531AW-TC combines extreme CPU core counts with GPU acceleration for hybrid AI pipelines in design and simulation environments.
As AI and HPC increasingly integrate into every facet of daily life, the need for workload optimized edge servers and high-performance consumer gear is bigger than ever. From individual AI developers to start-ups, retail shops to researchers and professors, these new machines are designed just for you.
To read more about how Supermicro brings enterprise-class AI performance to the client, edge, and consumer markets, check out our recent press release.