Samsung Electronics as we speak introduced plans to create a brand new AI Megafactory in collaboration with NVIDIA, marking a significant milestone within the firm’s efforts to guide the worldwide paradigm shift towards AI-driven manufacturing. By deploying greater than 50,000 NVIDIA GPUs, AI might be embedded all through Samsung’s total manufacturing stream, accelerating growth and manufacturing of next-generation semiconductors, cellular gadgets and robotics.
Samsung’s AI Factory will combine each side of semiconductor manufacturing, from design and course of to gear, operations and high quality management, right into a single clever community, the place AI repeatedly analyzes, predicts and optimizes manufacturing environments in actual time.
The Samsung AI Factory goes past conventional automation, serving as an clever manufacturing platform that connects and interprets the immense knowledge generated throughout chip design, manufacturing and gear operations.
From 25+ Year Collaboration to a Stronger AI Chip Alliance
Samsung and NVIDIA rejoice a collaboration that spans greater than 25 years, starting with Samsung’s DRAM powering NVIDIA’s early graphics playing cards and lengthening by way of foundry partnership.
In addition to their ongoing collaborations, Samsung and NVIDIA are additionally working collectively on HBM4. With extremely excessive bandwidth and vitality effectivity, Samsung’s superior HBM options are anticipated to assist speed up the event of future AI functions and type a vital basis for manufacturing infrastructure pushed by these applied sciences.
Built with the corporate’s Sixth-generation 10-nanometer (nm)-class DRAM and a 4nm logic base die, Samsung HBM4’s processing speeds can attain 11 gigabits-per-second (Gbps), far exceeding the JEDEC normal of 8Gbps.
Samsung may also proceed to ship next-generation reminiscence options, together with HBM, GDDR, and SOCAMM, in addition to foundry providers, driving innovation and scalability throughout the worldwide AI worth chain.
Accelerating the Transition to Intelligence-Driven Manufacturing
Over the subsequent a number of years, Samsung plans to implement NVIDIA accelerated computing to scale its AI Factory and speed up digital twin manufacturing powered by NVIDIA Omniverse libraries throughout one of many world’s most complete chip manufacturing infrastructures ― spanning reminiscence, logic, foundry and superior packaging.
Leveraging NVIDIA cuLitho and CUDA-X libraries for its optical proximity correction (OPC) course of, Samsung has achieved a 20x achieve in computational lithography efficiency. As a vital step in correct wafer patterning, the improved OPC permits AI to foretell and proper circuit sample variations with far higher velocity and precision, decreasing growth cycles.
For digital design automation (EDA), the businesses are collaborating with EDA companions to develop next-generation GPU-accelerated EDA instruments and design applied sciences.
Using NVIDIA Omniverse libraries, Samsung is constructing digital twins that may visualize total fab operations nearly. These digital environments determine anomalies, carry out predictive upkeep and optimize manufacturing earlier than adjustments are utilized within the bodily world.
Samsung plans to increase its AI Factory infrastructure to its international manufacturing hubs, together with Taylor, U.S., bringing higher intelligence and agility to its worldwide semiconductor operations.
Building Out the AI Ecosystem Across Generative AI and Robotics
Samsung has been growing proprietary AI fashions that energy greater than 400 million of its gadgets. These fashions are additionally built-in into the corporate’s inner manufacturing methods, driving intelligence and innovation throughout its manufacturing processes.
Built on NVIDIA accelerated computing and the Megatron framework, Samsung’s AI fashions reveal superior reasoning capabilities that ship distinctive efficiency in real-time translation, multilingual conversations and…
