NVIDIA has introduced technologies that accelerate the development and production of chips

AI for Chip Design

NVIDIA has introduced the cuPotential library, which enables developers to create AI agents for simulating interactions between atoms and molecules. Such simulations help study material behavior at the atomic level and better understand the physical processes underlying the development of new materials and semiconductor technologies.

AI agents can use these models to explore various scenarios and accelerate computations. This is crucial for chip design, as developers must analyze complex physical processes and test numerous parameters. The faster these calculations are performed, the more options can be explored within the same timeframe.

Early tests of cuPotential demonstrated an additional two- to threefold acceleration for certain workflows already optimized for GPUs. The library is intended for developers who wish to build their own AI agents for scientific and engineering tasks.

Lam Research Accelerates Manufacturing Simulation

Lam Research plans to integrate NanoVDB technology into the SEMulator3D platform, which is used for virtual simulation of semiconductor manufacturing processes.

During chip fabrication, manufacturers must precisely control material processing and the geometry of microscopic structures. Simulation allows for the pre-evaluation of how changes in technological parameters will affect outcomes and enables testing of different options without the need to conduct each experiment directly on the production line.

The integration of NanoVDB is expected to accelerate the simulation of individual semiconductor wafer etching processes by up to 30 times. This can help engineers faster verify technological solutions, optimize manufacturing processes, and reduce the time required to develop new chip fabrication technologies.

KLA Accelerates Computations for Semiconductor Analysis

KLA has introduced a new computational component, the NPP NCC, designed for tasks related to data analysis and processing in the semiconductor industry.

Chip production requires the analysis of large volumes of information to monitor product quality and detect deviations from specified parameters. Computational tools help process such data and support control and analysis processes at various stages of production.

The new component demonstrated performance 50 times higher than previous versions running on central processing units. Accelerated computations enable faster processing of complex tasks and can improve the efficiency of workflows related to semiconductor manufacturing.

Advantest Uses AI for Manufacturing Diagnostics

Advantest has integrated NVIDIA Nemotron and Kumo-Anomaly models into the ACS RTDI platform, designed for data analysis in semiconductor manufacturing.

On production lines, equipment continuously generates data about technological processes. Analyzing this data helps identify deviations, determine potential causes of malfunctions, and prevent issues that could affect product quality or lead to equipment downtime.

The integration of AI allows for the analysis of manufacturing data, detection of anomalies, and provision of recommendations for further actions. This helps specialists quickly resolve emerging issues and make data-driven decisions.

Using the solution can reduce response time to manufacturing problems from several hours to just a few minutes. This is particularly important for semiconductor fabs, as delays in detecting and resolving faults can impact production efficiency.

What Are AI Fabs

NVIDIA is also advancing the concept of AI-defined fabs—semiconductor factories where artificial intelligence is used to optimize manufacturing processes, analyze data, and detect faults.

Such facilities can combine AI agents, computational tools, and digital twins. The latter are virtual models of equipment or manufacturing processes. Engineers use them to test changes and evaluate potential outcomes before implementing new parameters on the actual production line.

Using AI at various stages of production helps identify problems faster, utilize equipment more efficiently, and reduce the time required to validate technological solutions. NVIDIA is developing this direction in collaboration with semiconductor industry companies, offering tools for simulation, computation, and analysis of manufacturing data.