Samsung Electronics has announced plans to establish an “AI Megafactory” in partnership with U.S.-based technology company NVIDIA, marking a major milestone in Samsung’s shift toward AI-driven production.
According to the company statement, more than 50,000 NVIDIA GPU solutions will be deployed as part of the collaboration, enabling Samsung to integrate artificial intelligence throughout its entire manufacturing workflow. This integration is expected to accelerate the development and production of next-generation semiconductors, mobile devices, and robotics technologies.
The Samsung AI Factory will unify every phase of semiconductor production—ranging from design and manufacturing to equipment, operations, and quality control—within a single intelligent network. Throughout this network, AI will continuously analyze manufacturing environments in real time, generate predictions, and optimize processes.
Moving beyond traditional automation, the AI Factory will function as an intelligent production platform that connects and interprets data from chip design, manufacturing, and equipment operations.
A 25-Year Collaboration Evolves into an AI Chip Alliance
Samsung and NVIDIA’s partnership began over 25 years ago, when Samsung’s DRAM solutions were used in the earliest NVIDIA graphics cards. Over time, this relationship has expanded to include high-performance memory and now a full-scale AI-enabled manufacturing initiative.
The two companies are currently co-developing HBM4, a next-generation high-bandwidth memory solution. Built using Samsung’s 6th-generation 10nm-class DRAM and 4nm logic chip, HBM4 reaches speeds of 11 Gbps—well beyond the JEDEC standard of 8 Gbps. Samsung continues to offer next-generation memory solutions such as HBM, GDDR, and SOCAMM, alongside its foundry services, contributing to innovation and scalability across the global AI value chain.
Digital Twins and Accelerated Computing
Samsung plans to scale the AI Factory using NVIDIA’s accelerated computing technologies. Its production infrastructure—including memory, logic chips, foundry services, and advanced packaging—will leverage digital twin models built using NVIDIA Omniverse.
Samsung has boosted computational lithography performance twentyfold by utilizing NVIDIA’s cuLitho and CUDA-X libraries for optical proximity correction (OPC). This advanced OPC process—a critical step in precision wafer patterning—uses AI to predict and correct circuit variations with greater speed and accuracy, shortening development cycles.
In electronic design automation (EDA), Samsung and NVIDIA are working with industry partners to develop next-generation GPU-accelerated design tools and technologies.
Samsung is also developing proprietary AI models supporting over 400 million devices. These models, integrated into its internal production systems, enhance decision-making and innovation. Built on NVIDIA’s Megatron platform, they feature advanced reasoning capabilities.
Robotics and Physical AI Integration
In the robotics domain, Samsung is using the NVIDIA RTX PRO 6000 Blackwell Server Edition platform to advance industrial automation and humanoid robotics. Several NVIDIA AI platforms enable Samsung to link virtual simulations with real-world robot data—allowing robots to perceive their environment, make decisions, and operate intelligently.
Through the NVIDIA Jetson Thor robotics platform, Samsung is improving real-time AI-based reasoning, task execution, and safety checks, while expanding these technologies beyond the AI Factory into broader business applications.
Connecting Consumers and Industries Through AI
Samsung is also collaborating with South Korean telecom operators, academic institutions, and research centers to advance AI-RAN (AI + Radio Access Network) technologies.
AI-RAN integrates AI computing capabilities directly into mobile network infrastructure. It stands out as a next-generation communication technology that enables intelligent perception and data processing for robots, drones, and industrial automation systems.
AI-RAN’s edge-based AI inference capabilities will act as a neural network enabling widespread adoption of physical AI.
Based on the collaboration launched last year, Samsung combined its software-defined network with NVIDIA’s GPU solutions to demonstrate the viability of AI-RAN. Both companies plan to deepen their cooperation in this area as they continue building the future of AI-powered telecommunications and manufacturing.
Source: Anadolu Ajansı/ Prepared by: İlayda Gök