PEGATRON Unveils High-Density GPU Rack Featuring NVIDIA GB300 NVL72 and NVIDIA MGX Servers with RTX PRO 6000 at GTC 2025

San Jose, CA — Mar. 18th, 2025 – PEGATRON, a globally recognized Design, Manufacturing, and Service (DMS) provider, proudly announces its latest high-density GPU rack solution, built on the NVIDIA GB300 NVL72 platform, AI servers with NVIDIA HGX™ B300 NVL16, and NVIDIA MGX servers with NVIDIA RTX PRO™ 6000 Blackwell Server Edition. Driving digital transformation, Pegatron exhibits a simulation-first AI-driven agent and Virtual Standard Operating Procedure (VSOP). The new solutions, showcased at GTC 2025 booth 232, deliver unprecedented performance, scalability, and efficiency for AI, machine learning, and data center operations.

Cutting-Edge Performance with NVIDIA GB300 NVL72 Solution

The Pegatron RA4402-72N2, equipped with the NVIDIA Blackwell Ultra GB300 NVL72, heralds a new era in AI with enhanced computational power and expanded memory. Boasting increased AI FLOPS, over 20TB of HBM3e memory, and integrated networking platforms like NVIDIA Quantum-X800 InfiniBand and NVIDIA Spectrum-X Ethernet, NVIDIA Blackwell Ultra offers exceptional performance for AI reasoning, agentic AI, and video inference tasks.

Unparalleled Efficiency with NVIDIA HGX Platform

The Pegatron MS303-2A1G liquid-cooled server, with NVIDIA HGX H200 4-GPU and dual AMD EPYC™ 9005 series processors, creates a high-density OCP server that meets ORv3 specifications. Combined with the waterless, two-phase direct-to-chip liquid cooling solution from ZutaCore®, the MS303-2A1G server reshapes end-to-end cooling architecture in the data center, delivering performance and energy savings. Pegatron is also preparing to support the upcoming NVIDIA HGX B300 NVL16 system, offering enterprise-ready servers for AI reasoning, agentic AI, and video inference applications for every data center. Further details will be announced soon.

Modular Flexibility with NVIDIA MGX Architecture and RTX PRO 6000 GPU

Pegatron’s servers AS400-2A1 and AS205-2T1 leverage NVIDIA MGX architecture, offering a flexible and modular design that accelerates time-to-market and ensures multi-generation compatibility. Featuring the NVIDIA H200 NVL and supporting RTX PRO 6000 Blackwell Server Edition, which is a powerful data center GPU for AI and visual computing. These servers accelerate demanding enterprise workloads, including AI, scientific computing, graphics, and video applications.

Revolutionary Simulation-First AI-Driven Agent

Pegatron's JEDI is a cutting-edge AI agent system, built on the PEGAVERSE platform with NVIDIA Omniverse and NVIDIA Isaac Sim. This self-optimizing robotic system simulates dispensing results, leveraging machine learning for accuracy and consistency. JEDI's adaptable architecture applies to various industrial tasks, ensuring high-quality production and reliability. With its scalable design, JEDI drives growth and innovation in manufacturing, empowering manufacturers to thrive in an era of rapid change and unlocking new possibilities for competitiveness and success in the industry with consistent and precise operations, enhancing overall efficiency and productivity levels significantly.

Transforming Corporate Efficiency with VSOP

Pegatron's YODA is a groundbreaking Virtual Standard Operating Procedure (VSOP) platform that enhances corporate efficiency through technique cultivation, feedback, and supervision. Leveraging NVIDIA Omniverse and Isaac Sim, YODA enables virtual instruction for industrial applications, streamlining assembly processes and reducing the learning curve for assemblers, ultimately improving yield rates. As a designated NVIDIA solution provider, Pegatron demonstrates its technical expertise in driving digital transformation, consistently pursuing excellence and innovation in the industry, and solidifying its position as a leader in revolutionary technology solutions, with a focus on efficiency and productivity.

Driving Innovation at GTC 2025

“We are thrilled to introduce our latest high-density data center solutions at GTC 2025, designed to meet the growing demands of high-performance AI workloads,” said James Shue, the CTO of Pegatron. “With NVIDIA’s cutting-edge platforms, our advanced cooling technologies, and cloud-based PEGAVERSE platform, we are setting a new standard for data center performance and sustainability.”

Visit Pegatron at GTC 2025 booth 232 to experience the future of high-performance computing and learn more about our groundbreaking solutions.