Samsung Electronics announced plans to establish a new AI Megafactory in partnership with NVIDIA, marking a major step toward AI-powered manufacturing across semiconductors, mobile devices and robotics. The initiative will use over 50,000 NVIDIA GPUs and the NVIDIA Omniverse platform to integrate design, production and operations within one intelligent network where AI continuously analyzes and optimizes processes in real time.
The partnership extends a 25-year collaboration between the two companies, now strengthened through joint development of HBM4 memory built on 6th-generation 10nm DRAM and 4nm logic base die, capable of speeds up to 11Gbps. Samsung will also continue advancing HBM, GDDR and SOCAMM memory to support future AI applications.
By applying NVIDIA accelerated computing, cuLitho and CUDA-X libraries, Samsung achieved a 20× performance gain in computational lithography, reducing design and production cycles. The company is also creating digital twins of fabrication plants using NVIDIA Omniverse for predictive maintenance and optimization, expanding this model to global hubs including Taylor, U.S.
Samsung is further integrating NVIDIA AI platforms like Jetson Thor and RTX PRO 6000 Blackwell Server Edition to enhance intelligent robotics, combining simulation and real-world data for more autonomous systems.
Samsung and NVIDIA are working with partners on AI-RAN, a next-generation network technology that brings AI to the mobile edge to enable intelligent and responsive operations for connected devices and robots.
Leave a comment