Nvidia CEO, Jensen Huang, highlighted the crucial role of artificial intelligence (AI) in next-generation chip manufacturing. As CPUs face challenges in increasing efficiency and power due to the difficulty of fitting more transistors onto silicon wafers, demand for advanced computing hardware continues to rise.
Nvidia has emerged as a leader in combining the parallel processing capabilities of GPUs with CPUs, sparking the AI revolution and creating opportunities in manufacturing, robotics, and autonomous vehicles.
Chip manufacturing involves a complex process of over 1,000 steps, requiring precision and computational sciences. Huang emphasized that chip manufacturing is an ideal application for Nvidia’s accelerated and AI computing.
The company’s GPUs are being utilized by various firms in different aspects of chip manufacturing. For instance, D2S Inc., IMS Nanofabrication GmbH, and NuFlare Technology Inc. are employing Nvidia’s hardware to accelerate pattern rendering and mask process correction in the building of “mask writers.”
Taiwan Semiconductor Manufacturing Co., KLA Corp., and Laser Technology Inc. are leveraging Nvidia’s GPUs for mask inspection, utilizing classical physics models and deep learning to detect defects.
Furthermore, Nvidia is collaborating with chipmakers to accelerate computational lithography, which is a significant workload in chip design and manufacturing. Huang announced the introduction of Nvidia cuLitho, a new software library that optimizes GPU-accelerated computational lithography.
This advancement is expected to make chip design beyond two nanometers possible and significantly enhance efficiency by reducing power and cost.
As per some sources, Huang discussed the concept of “embodied AI,” which entails creating AI systems that can understand, reason about, and interact with the physical world. Nvidia has taken initial steps towards this vision by introducing multimodal embodied AI, such as Nvidia VIMA, which can perform tasks based on visual prompts.
Additionally, the company is developing Earth-2, a digital twin of the Earth, integrated within Nvidia Omniverse. Earth-2 incorporates a physics AI model called FourCastNet to simulate global weather patterns at an accelerated pace, supporting advancements in semiconductor manufacturing and addressing climate change.
Nvidia’s commitment to advancing chip manufacturing through AI, coupled with its focus on embodied AI and digital twin development, positions them at the forefront of innovation in the technology industry.