# PEGATRON Corporation to show AI and HPC platforms at SC26

URL: https://decisioninsights.ai/pegatron-corporation-to-show-ai-and-hpc-platforms-at-sc26/

PEGATRON Corporation plans to present AI and HPC infrastructure at SC26 in Chicago, showing how its platforms respond to evolving demands from agentic AI and HPC.Agentic AI relies on CPUs to manage concurrent agent environments, tool executions, sandboxed code, reinforcement learning evaluations, and data processing. As AI moves beyond training and inference toward autonomous agents that execute complex tasks, infrastructure requirements are changing. Meeting these demands calls for high-density CPU infrastructure that operates alongside GPU-accelerated systems at AI factory scale.NVIDIA Vera CPU Rack, built on the NVIDIA MGX modular reference architecture, integrates 256 liquid-cooled NVIDIA Vera CPUs, delivering more than 22,500 cores and 45,000 threads, with up to 300 TB/s of aggregate LPDDR5X memory bandwidth. It supports 22,500 sandboxes and delivers 80% faster agentic CPU performance, helping AI factories scale agent execution, reinforcement learning evaluations, and data-processing pipelines.NVIDIA Spectrum-X Ethernet and NVIDIA Quantum-X800 InfiniBand networking enable AI factories to scale CPU-intensive workloads alongside accelerated compute. The RA4803-72N3 is a fully liquid-cooled NVIDIA Vera Rubin NVL72 rack integrating 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs for frontier-scale AI training, reinforcement learning, and large-scale agentic inference. The AS210-2T1-8H3 is a fully liquid-cooled 2U NVIDIA HGX Rubin NVL8 system with an eight-GPU Rubin baseboard.PEGATRON also plans to introduce the MS310-2A2E, a compact 2OU, 2-node server with AMD EPYC 9006 Series processors, designed to increase compute density within limited rack space. The GS203-2A2 and GS205-2A2E dual-socket servers, along with the SS201-1A1 storage platform supporting up to 32 E3.S NVMe SSDs, address HPC, memory-intensive applications, and high-throughput AI data pipelines.“GPUs remain essential for AI acceleration, while CPUs are taking on a more critical role in agent execution, reinforcement learning, and data processing.” said Gary Cheng, President and Co-CEO of PEGATRON. SC26 runs November 16–19 at McCormick Place in Chicago, and PEGATRON plans to host visitors at Booth #5416.
