Samsung and NVIDIA Unveil AI-Driven Semiconductor Manufacturing Platform

Strategic Insights -

Samsung and NVIDIA have forged a deep partnership spanning over 25 years to develop next-generation semiconductors and accelerate chip manufacturing. Their collaboration has led to significant breakthroughs in computational lithography, technology computer-aided design simulations, and AI-driven predictive maintenance. The companies are building a state-of-the-art AI factory, powered by more than 50,000 NVIDIA GPUs, to establish the foundation of AI-driven production. This new era of intelligent computing and chip manufacturing aims to redefine how the world designs, builds, and manufactures semiconductors.

Key Takeaways:

  • Samsung and NVIDIA are working together to innovate breakthroughs for next-generation semiconductors, including HBM and custom solutions.
  • The companies' partnership has led to 20x performance gains in computational lithography and technology computer-aided design simulations.
  • Samsung is harnessing NVIDIA GPUs, CUDA-X libraries, and solutions from Synopsys, Cadence, and Siemens to achieve massive speedups in simulation, verification, and manufacturing analysis.
  • Samsung is building digital twins with NVIDIA Omniverse for global fabs to shorten time from design to operations and enable AI-driven predictive maintenance, operational optimization, and real-time decision-making.
  • NVIDIA and Samsung are collaborating on AI-RAN network technology, bringing together AI and mobile network workloads.
  • Samsung is using NVIDIA robotics technologies on NVIDIA RTX PRO Servers to accelerate intelligent robot deployment across manufacturing automation and humanoid robot applications.
  • The companies' collaboration has led to the development of a new AI factory, combining Samsung's semiconductor technologies with NVIDIA platforms to establish the foundation of next-generation, AI-driven production.

Statistics:

  • 25 years: the duration of Samsung and NVIDIA's partnership
  • 50,000: the number of NVIDIA GPUs powering Samsung's AI factory
  • 20x: the performance gain achieved in computational lithography and technology computer-aided design simulations
  • 400 million: the number of Samsung devices powered by Samsung's proprietary AI models
  • 10: the number of global fab digital twins being built with NVIDIA Omniverse
  • 1995: the year of Samsung's collaboration with NVIDIA on the DRAM for NVIDIA's first graphics card

Sources:

  • GLOBE NEWSWIRE, "NVIDIA and Samsung Announce Plans for AI Factory at APEC Summit"
  • NVIDIA press release, "Samsung Accelerates Smart Manufacturing With Digital Twins, Robotics, and Generative AI"
  • NVIDIA Annual Report on Form 10-K for the fiscal year ended January 31, 2022