Taipei, November 1
Samsung Electronics said Friday that it will create a next-generation artificial intelligence megafactory in partnership with US chip giant Nvidia, aiming to embed AI across its entire global semiconductor production line, according to Focus Taiwan.
The new platform will integrate AI into every phase of the manufacturing flow, from semiconductor design and processing to equipment control and quality assurance, and will run on more than 50,000 Nvidia GPUs.
"The Samsung AI Factory goes beyond traditional automation," a Samsung Electronics official said. "It connects and interprets immense data generated across chip design, production and equipment operations."
Described by the company as an "intelligent manufacturing platform," the factory will use AI to analyse, predict and optimise manufacturing conditions in real time.
The initiative builds on a partnership between Samsung and Nvidia that spans over 25 years. The two companies first collaborated when Samsung supplied DRAM for Nvidia's early graphics cards, and the relationship has since expanded to include foundry and memory technologies.
The latest joint efforts are focused on HBM4, a high-bandwidth memory solution built using Samsung's sixth-generation 10nm-class DRAM and 4nm logic base die.
"Samsung will continue to deliver next-generation memory solutions, including HBM, GDDR and SOCAMM, as well as foundry services, driving innovation and scalability across the global AI value chain," the official said.
The megafactory will utilise Nvidia's Omniverse and Cuda-X platforms to construct "digital twins" of entire chip fabs -- virtual models that allow engineers to simulate equipment operations and potential upgrades before they are applied to physical systems. These simulations can help with predictive maintenance and process optimisation.
On the production floor, Samsung is also leveraging Nvidia's cuLitho software to achieve up to 20 times faster performance in computational lithography. This boosts processes like optical proximity correction, which enables AI to more precisely predict and fix circuit pattern variations, improving development speed and chip yield.
In the design stage, Samsung is working with electronic design automation partners to develop GPU-accelerated tools to push the boundaries of semiconductor design capabilities.
In robotics and factory automation, Samsung is deploying Nvidia's RTX Pro 6000 Blackwell Server Edition to enhance the performance of humanoid robots and is using Jetson Thor for real-time AI reasoning and task execution in smart robots.
The partnership also extends to next-generation mobile networks. Through joint AI-RAN development, the two companies are building a system that allows edge devices -- like drones and industrial robots -- to carry out real-time AI inference and decision-making using Nvidia's GPU compute power combined with Samsung's software-based mobile network infrastructure.
"This AI-powered mobile network will play a crucial role as a neural network essential in the widespread adoption of physical AI," the official said.
Samsung said it plans to apply the AI factory infrastructure across its global sites, including the new chip fab under construction in Taylor, Texas. The move reinforces the company's ambitions to lead in every major semiconductor area: memory, logic, foundry and advanced packaging.
Samsung's AI models are already running in over 400 million consumer devices. Now, the company aims to embed these capabilities into its own factories through the Megatron framework, supporting multilingual interactions, intelligent summarisation and advanced reasoning.
"This is a critical milestone in our journey to lead the global shift toward AI-driven manufacturing," the official said.
- ANI
Reader Comments
We welcome thoughtful discussions from our readers. Please keep comments respectful and on-topic.