Tokyo Electron, Nvidia expand collaboration for agentic AI, robotics

As part of the development of agentic AI and robotics solutions, Tokyo Electron (TEL) is collaborating with Nvidia, utilizing Nvidia Agent Toolkit including Nvidia NeMo, Nvidia NIM, Nvidia NemoClaw blueprints and the open robot development platform Nvidia Isaac within TEL’s “Epsira” concept for digital transformation (DX) solutions, Tokyo Electron said.

Using the Isaac platform allows TEL to feed designs into the digital twins integrated with Nvidia Omniverse libraries for/Simulation; often robotic maintenance must be done on various tool configurations and therefore utilizing a digital twin can allow much faster robot training and accuracy. Regarding AI Agents, NeMo helps TEL accelerate Agent training, tuning and evaluation; NemoClaw allows easier operation and more secure Agent deployment.

TEL’s Epsira focuses on enhancing customers’ productivity by improving field production efficiency, boosting yields, and increasing equipment availability by reducing both scheduled and unscheduled down time utilizing AI MTBWC (Mean Time Between Wet Cleaning) extension analysis, robotics maintenance and AI troubleshooting analysis, Tokyo Electron said.

“We are pleased that our collaboration with Nvidia, a world leader in the AI era, has moved to an even deeper state,” said Keiichi Matsushima, Vice President & General Manager, Corporate Development Division at TEL. “This deeper collaboration will allow us to further leverage TEL’s Epsira program to enhance our customers’ productivity by enhancing manufacturing utilizing Robotics maintenance and AI troubleshooting analysis.” 

“Semiconductor manufacturing is entering a new era of automation, where robotics can help fabs drive greater uptime, precision and resilience across increasingly complex production environments,” said Amit Goel, senior director of robotics and edge AI ecosystem at Nvidia. “TEL is bringing more intelligence to robotic maintenance, AI troubleshooting, and fab operations by leveraging Nvidia Agentic and Physical AI models and frameworks.”