“Helping optimize processes and predict risks through a digital twin–based platform”
Samsung Electronics and SK hynix announced that they are enhancing semiconductor design and manufacturing capabilities by using NVIDIA’s “digital twin” platform, which replicates real-world spaces identically in virtual environments.
At NVIDIA’s annual developer conference “GTC 2026,” held in San Jose, California, on the 17th (local time), Song Yong-ho, Head of the Samsung Electronics AI Center, presented the company’s semiconductor engineering strategy and made this announcement. Song introduced cases where Samsung’s AI is integrated with partner companies’ semiconductor design and manufacturing tools to shorten development time and improve production efficiency and quality.
In particular, Song unveiled the implementation results of Samsung Electronics’ Pyeongtaek Plant 1 based on NVIDIA’s digital twin platform “Omniverse,” explaining that “it has become possible to conduct preemptive responses by optimizing processes and predicting risks through simulation.” On the same day, Samsung also unveiled for the first time a video titled “Humanoid Manufacturing Innovation,” showing how the application of humanoid robots to semiconductor processes might look.
Do Seung-yong, Executive Vice President and Head of the DT Division at SK hynix, speaks during a panel discussion at NVIDIA GTC 2026. Provided by SK hynix
On the same day, Do Seung-yong, Head of the DT (Digital Transformation) Division at SK hynix, stated during a panel discussion at GTC 2026 that the company “is working to build an autonomous fab (semiconductor plant) by 2030.” He explained, “The goal is to enable the plant to learn and make decisions on its own, dramatically shortening the transition time from design to mass production.”
Do also said that SK hynix is leveraging NVIDIA’s Omniverse platform to build an autonomous plant. This allows the company to verify and test production flows, material movements, and other processes in a virtual environment in advance, without having to shut down operating plants.
Meanwhile, NVIDIA announced that it will enter into a collaboration with pharmaceutical company Roche to build an AI factory. Roche plans to introduce more than 3,500 units of NVIDIA’s “Blackwell” graphics processing units (GPUs) to use AI in new drug development and other areas.
Park Jong-min
AI-translated with ChatGPT. Provided as is; original Korean text prevails.
ⓒ dongA.com. All rights reserved. Reproduction, redistribution, or use for AI training prohibited.
