Super Micro Computer, , a global leader in AI, cloud, storage, and 5G/Edge total solutions, announced a new class of data center architectures natively optimized for Arm® AGI processors. This hardware release is custom-engineered to manage the unique computational workflows of modern agentic AI, maximizing node processing density and performance-per-watt within standard data center footprints.
Optimizing Rack Economics for Agentic AI Orchestration
With the shift from traditional AI to autonomous agent systems, there is a significant value being put into efficient parallel processing, memory bandwidth, and microservices latency instead of just the computational speed in today’s loads. Current hardware is unable to scale such demanding tasks without surpassing the average power budgets in the data center.
This issue is resolved by leveraging the unique approach to computing with DCBBS technology offered by Supermicro. Through the use of custom-designed hardware modules, Supermicro reduces data center deployment complexity and increases time-to-online capabilities to give an enterprise solution that is more economically feasible overall.
“Supermicro continues to lead the industry when it comes to deploying new and innovative rack-scale solutions that maximize performance and efficiency,” said Charles Liang, president and CEO of Supermicro. “Our DCBBS technology stack delivers end-to-end data center solutions of any size, which combined with the new density and efficient performance optimized Arm AGI CPU microarchitecture, helps enterprises realize significant TCO savings on their agentic AI infrastructure investments.”
Also Read: NVIDIA Launches Alpamayo 2 Super Open Reasoning VLA Model to Accelerate Robotaxis
“Agentic AI is driving a fundamental shift in infrastructure requirements, where efficiency, scalability, and orchestration performance are becoming just as critical as raw compute,” said Mohamed Awad, Executive Vice President, Cloud AI Business Unit, Arm. “By combining Arm AGI CPUs with Supermicro’s rack-scale system expertise, we’re enabling infrastructure designed to deliver higher AI throughput, maximum compute density, and improved data center economics at scale.”
A Specialized Portfolio of Air- and Liquid-Cooled Form Factors
The new server lineup introduces multiple configuration options built on the foundation of the energy-efficient Arm Neoverse® CSS V3-based architecture:
2U Compute-Optimized Servers: Air-cooled, dual-socket systems tailored for high-volume token handling and high-throughput application orchestration.
5U GPU-Optimized Servers: Heavy-duty, air-cooled clusters engineered to balance heavy acceleration with high-density CPU throughput.
Multi-Node Liquid-Cooled Environments: High-density, direct-to-chip liquid cooling circuits built explicitly to sustain peak rack performance while cutting cooling power demands.
According to Arm’s architectural estimates, deploying the Arm AGI CPU within Supermicro’s dense systems can deliver over 2x the performance per rack compared to traditional legacy x86 setups. At hyper-scale, this optimization wave can potentially save enterprises up to $10 billion in CAPEX per gigawatt of global AI data center capacity by reducing unnecessary physical site footprints and electrical routing layers.
Inside the Arm AGI Microarchitecture
The Arm AGI CPU features a highly dense, native 136-core design built specifically to minimize legacy x86 instruction overhead and execute more concurrent instructions per clock cycle. To sustain massive multi-agent parallel processing, each chip provides a massive 6 GB/s memory bandwidth per individual core, alongside latency-optimized memory caches to enable clean linear performance scaling. Backed by expansive memory allocations and versatile I/O lanes, the hardware platform easily orchestrates thousands of concurrent, automated tasks across vast distributed infrastructures without experiencing thermal throttling.
