Advantech to demo Nvidia’s NemoClaw edge‑AI stack at Computex Taipei
- Advantech said April 27 it will show an Nvidia NemoClaw-based edge artificial intelligence system at Computex 2026 as it pushes industrial deployments. - The company said its next integrated edge AI platforms will deliver at least 100 TOPS and package hardware in embedded, appliance, and server tiers. - NemoClaw arrived in March as Nvidia’s secured OpenClaw stack for autonomous agents, giving Advantech a fresh software layer. (nvidia.com)
Advantech said on April 27 it will demo an integrated edge artificial intelligence system built on Nvidia’s NemoClaw stack at Computex 2026 in Taipei. (taipeitimes.com) The company tied the demo to a broader push into edge AI, where models run near factories, hospitals or robots instead of sending every task back to a distant cloud. Miller Chang, president of Advantech’s embedded Internet of Things group, said that shift is being driven by low-latency, real-time and data-privacy requirements. (taipeitimes.com) NemoClaw is Nvidia’s new software stack for OpenClaw, an open-source agent platform that lets AI systems carry out longer chains of actions. Nvidia introduced NemoClaw on March 16 at GTC, saying it bundles Nemotron models and the OpenShell runtime in one command and adds policy-based privacy, network and security controls. (nvidia.com 1) (nvidia.com 2) In plain terms, Advantech is trying to turn an AI agent from a cloud demo into an industrial box: a local computer that can see data, follow rules and act inside a plant or device. Advantech’s own April 16 product launch said its MIC-AI systems combine OpenClaw, Nemotron and Nvidia OpenShell inside NemoClaw on Jetson Thor hardware for “autonomous decision-making at the edge.” (advantech.com) That matters because industrial customers usually care less about chatbot novelty than about whether a system can stay on-site, respond fast and keep sensitive production data local. Chang told the Taipei Times that the biggest inference demand is likely to come from healthcare, manufacturing and semiconductors, where privacy and real-time response are harder to outsource to the cloud. (taipeitimes.com) Advantech said its next integrated edge AI platforms will ship this year with at least 100 trillion operations per second, or TOPS, of compute. Chang also described a three-tier hardware plan: Edge AI Embedded modules, Edge AI Appliance systems and Edge AI Server products for on-site processing where cloud deployment does not fit. (taipeitimes.com) The company has already been building out the hardware side of that pitch with Nvidia. In March and April, Advantech promoted Jetson Thor-based systems for robotics, medical devices and industrial AI, and said its Computex 2026 pavilion would focus on edge computing and AI-powered deployments. (advantech.com 1) (advantech.com 2) Computex 2026 runs from June 2 to June 5 in Taipei, and Nvidia chief executive Jensen Huang is already scheduled to keynote the show. That gives Advantech a high-profile stage to test whether NemoClaw can move from a March software launch into an industrial product story by early June. (computextaipei.com.tw) (taipeitimes.com)