Samsung plans to combine digital twins, AI, and robotics throughout its manufacturing infrastructure, beginning with a brand new ‘Megafactory’.
For COOs and CIOs in industrial sectors, the good manufacturing facility announcement – in-built collaboration with NVIDIA – is an instance of the shift from contained AI pilots into manufacturing. The purpose is a totally clever and predictive setting, beginning with the complicated calls for of semiconductor, cell machine, and robotics manufacturing.
Samsung will deploy greater than 50,000 NVIDIA GPUs to embed AI all through the manufacturing movement. This goes additional than typical automation. The manufacturing facility will use a single clever community that can let AI constantly analyse, predict, and optimise manufacturing environments in real-time.
Digital twins for bodily features
Operationally, a key element is the widespread use of digital twins, powered by NVIDIA Omniverse libraries.
Samsung is constructing digital twins able to visualising total fab operations just about. The enterprise software is to make use of these digital environments to establish anomalies, carry out predictive upkeep, and optimise manufacturing earlier than adjustments are utilized within the bodily world. This method targets a discount in downtime and permits testing course of enhancements with out risking bodily line disruption.
In a high-value use case, Samsung detailed effectivity features in its computational lithography course of. By utilizing NVIDIA cuLitho and CUDA-X libraries for its optical proximity correction (OPC) course of, the corporate achieved a 20x achieve in computational lithography efficiency. As OPC is a key step in correct wafer patterning, the enhancement permits AI to foretell and proper circuit sample variations with far larger velocity and precision, thereby lowering growth cycles.
This ‘AI Manufacturing facility’ idea displays a wider business push. Many enterprises at the moment are trying to consolidate AI growth, leveraging platforms like Google Vertex AI or IBM watsonx to handle fashions and knowledge. Samsung’s method, constructed on a 25-year collaboration with NVIDIA, is a hardware-centric mannequin for reaching this integration at scale.
Samsung places AI on the core of a full manufacturing ecosystem method
The implementation extends to next-generation {hardware} and robotics. The businesses are working collectively on HBM4, with Samsung’s design utilizing Sixth-generation 10-nanometer-class DRAM and a 4nm logic base die. Samsung states its HBM4 processing speeds can attain 11Gbps, far exceeding the 8Gbps JEDEC customary. This superior reminiscence is meant to type a basis for the AI-driven manufacturing infrastructure.
On the manufacturing facility flooring, the plan entails bodily automation. Samsung is utilizing the NVIDIA RTX PRO 6000 Blackwell Server Version platform to advance manufacturing automation and humanoid robotics. Additionally it is utilizing the NVIDIA Jetson Thor robotic platform to speed up real-time AI reasoning, job execution, and security controls in clever robots.
This isn’t a single-site venture, both. Samsung is planning to increase its AI manufacturing facility infrastructure to its international manufacturing hubs to deliver larger intelligence and agility to its worldwide semiconductor operations.
Past the manufacturing facility partitions, the collaboration is tackling the community infrastructure required to assist widespread bodily AI. The companions are working with Korean telecom operators and researchers on AI-RAN growth.
AI-RAN is a next-generation communication know-how that integrates AI computing energy into cell community capabilities. For a CTO, this issues as a result of it permits robots, drones, and industrial automation gear to course of knowledge and run inferences in real-time on the community edge. This AI-powered cell community is positioned as a neural community that’s important for widespread adoption of bodily AI.
For enterprise decisionmakers, Samsung’s blueprint exhibits that an efficient ‘AI manufacturing facility’ have to be an end-to-end system. It requires unifying {hardware} (like HBM4), proprietary AI fashions, operational know-how (robotics), and edge networking (AI-RAN) right into a single and predictive knowledge movement.
See additionally: ASUS IoT platform makes use of NVIDIA tech for edge AI robotics
Need to be taught extra about IoT from business leaders? Take a look at IoT Tech Expo going down in Amsterdam, California, and London. The great occasion is a part of TechEx and is co-located with different main know-how occasions together with AI & Massive Knowledge Expo. Click on right here for extra data.
IoT Information is powered by TechForge Media. Discover different upcoming enterprise know-how occasions and webinars right here.
