Samsung Electronics said its collaboration with Nvidia is expanding to include "artificial intelligence (AI) agents" and "digital twins." It said that beyond the existing memory partnership centered on graphics processing units (GPUs), it is improving Samsung Electronics' manufacturing efficiency through Nvidia's solutions. A digital twin refers to a concept that precisely recreates the real world in a digital space to run simulations.
Song Yong-ho, head of the Samsung Electronics AI Center, delivered a semiconductor engineering strategy presentation on the 17th (local time) at Nvidia's annual developer conference (GTC 2026) at the San Jose Convention Center in California. Song said Samsung Electronics is optimizing the entire process from design to manufacturing using agentic AI, rather than focusing on individual stages. In design, Samsung Electronics' agents are organically linked to partners' tools to shorten development time and improve design efficiency and quality. In manufacturing, the company is also using AI to strengthen analytics capabilities and maintain a high-quality management system even in complex process environments.
Samsung Electronics also introduced cases where this strategy has been implemented in practice through a joint presentation with Synopsys. Song said the Pyeongtaek Plant 1 was recreated in a virtual space using Nvidia's digital twin solution Omniverse. The plan is to efficiently find ways to achieve high-quality chip production. The company also presented what results could be achieved by introducing a Humanoid Robot to semiconductor fabs.
"This event shows that the collaboration between Nvidia and Samsung Electronics is expanding beyond simple product supply into semiconductor engineering innovation centered on agentic AI and digital twins," Song said. "Based on this, Samsung Electronics will continue to support the implementation of Nvidia's next-generation AI chips, including 'Vera Rubin,' 'Vera Rubin Ultra,' and 'Feynman.'"