Samsung and NVIDIA create an artificial intelligence factory to automate chip production
Samsung Electronics announced the creation of an artificial intelligence factory in collaboration with NVIDIA. The project envisages the deployment of over 50,000 NVIDIA GPUs to integrate AI into all stages of production, from development to quality control, with the aim of scaling the development of semiconductors, mobile devices, and robotics.
What is known
Samsung will integrate all stages of chip production into a single intelligent system. AI will analyze, predict, and optimize processes in real-time, reducing development cycles and increasing manufacturing accuracy.
The company already demonstrates a 20-fold improvement in optical lithography productivity thanks to NVIDIA cuLitho and CUDA-X, allowing AI to more accurately predict deviations in chip patterns.
Samsung and NVIDIA are developing joint HBM4 developments, based on the 6th generation 10nm DRAM and 4nm logic chip, providing speeds up to 11 Gbit/s and significantly better average performance (8 Gbit/s).
Samsung plans to create virtual "digital twins" of factories using NVIDIA Omniverse. These are computer models that reflect real manufacturing processes.
In such environments, AI can:
- test changes without risk to equipment;
- identify problems before they occur;
- increase manufacturing efficiency and accuracy.
Thanks to NVIDIA cuLitho and CUDA-X libraries, Samsung has already increased computation speed 20 times, accelerating chip production and reducing defects.
The companies are also developing AI-RAN, a new generation of mobile networks where artificial intelligence is combined with the network, allowing robots, drones, and other devices to quickly process data and act intelligently without delays.
Source: Samsung
