Press Release

 **SAN JOSE, Calif. — March 11, 2026** — Inventec Corporation, a global leader in high-performance AI server solutions, today announced its participation as a Gold sponsor at NVIDIA GTC 2026, taking place from March 16 to March 19 in San Jose. As the AI industry enters a phase of unprecedented acceleration, Inventec is unveiling a comprehensive portfolio of next-generation infrastructure designed to help customers maintain a competitive edge through scalable, high-performance computing.

To address the evolving demands of modern data centers, Inventec will present an extensive range of AI servers that offer customers the flexibility to choose between air-cooling and liquid-cooling technologies based on their specific operational requirements, headlined by the “Artemis III” rack-scale solution. Built on the NVIDIA Vera Rubin NVL72 system, Artemis III features a fully liquid-cooled system integrating 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs, delivering exceptional performance for the most demanding workloads, including agentic AI systems and real-time video generation for AI factories. The showcase also includes a versatile range of AI servers tailored for diverse deployment environments, such as the P9000G7 10U air-cooled and P5800G7 5U liquid-cooled servers powered by the NVIDIA HGX B300 platform, along with the newly introduced P5880G7, which adopts the latest NVIDIA HGX Rubin NVL8 Platform with advanced liquid-cooling technology. Complementing these servers are the P4200G7 and the new P6000G7, both based on the NVIDIA MGX platform and equipped with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, delivering optimized performance in 4U air-cooled and 6U liquid-cooled form factors, respectively. Extending this high-performance architecture to the industrial edge, Inventec is also debuting the Edge IGX Thor 2U Server, the first standard rack-mount system built on NVIDIA IGX Thor. By integrating the NVIDIA IGX Thor T7000 board kit with optional NVIDIA RTX PRO Blackwell GPUs and ConnectX-7 networking, this 2U platform delivers enterprise-grade physical AI performance with the connectivity and control required for demanding edge workloads.

For the first time at GTC, Inventec is debuting its end-to-end robotics ecosystem, demonstrating a new era of Sim-to-Real industrial automation that bridges advanced AI simulation with physical deployment. Developed by Multidisciplinary capabilities, the portfolio first includes the iTwin dual-arm robot, achieving sub-0.05 mm precision through high-precision force-control software and an Isaac Sim–based Sim-to-Real verification pipeline, improving integration efficiency and reducing commissioning risks. In addition, the iUGV autonomous ground vehicle leverages NVIDIA Jetson Orin NX 16G Super mode and NVIDIA cuOpt for real-time fleet coordination in smart factory environments. Furthermore, the showcase features the iScout quadruped robot, developed using NVIDIA Isaac Sim, NVIDIA Warp, and a ROS 2–based control framework that enables rapid adaptation to diverse environments and terrain conditions, highlighting Inventec’s expertise in advanced bionic system design and end-to-end robotics software innovation. In terms of these robotic innovations are unified through the Inventec Fleet Management System (iFMS), which integrates NVIDIA Omniverse with NVIDIA MGX server platforms to enable a seamless, closed-loop automation framework for next-generation intelligent factories. Within NVIDIA’s three-computer framework, the Edge IGX Thor 2U Server acts as the critical execution platform, delivering real-time AI inference and deterministic control alongside industrial robots to provide a scalable, safety-focused foundation for the next generation of intelligent factories.

Inventec warmly welcomes all attendees to visit Booth #1130 at the San Jose Convention Center. This presentation offers a unique opportunity to experience firsthand the future of AI factories and AI infrastructure, where high-performance server infrastructure meets the next wave of intelligent, autonomous robotics.