← All news
Global · Education & Research
Global·Education & Research

NVIDIA expands CUDA-X libraries for AI-assisted scientific research

NVIDIA announced expanded CUDA-X libraries for scientific computing, including tools supporting AI agents and training-data preparation. The updates span molecular simulations, astronomy, materials research and single-cell biology.

AI Hubs United Editorial ·

NVIDIA has expanded its CUDA-X scientific computing libraries with cuPhoton, cuPotential and cuSpa, alongside tools for AI-assisted research and biological model development. According to NVIDIA, the additions help researchers process larger datasets and connect accelerated scientific workflows with AI agents. In fusion research, Pacific Fusion used cuPhoton.xLens to examine results during plasma simulations. Researchers can connect that workflow to a locally running NVIDIA Nemotron-powered agent for interactive analysis that could guide simulations and reactor design.

For drug discovery, NVIDIA says cuPotential accelerates existing molecular simulation software by up to three times while improving calculation stability and accuracy. The library also speeds synthetic data generation for generative AI models. NVIDIA is working with Schrödinger and several molecular dynamics software communities to integrate the technology, while adopters include Cadence and Terray Therapeutics.

The biology updates include newly open-source cuSpa and an agent-ready container for rapids-singlecell, a framework codeveloped with scverse. NVIDIA reports that Therna Biosciences processed data 200 times faster using rapids-singlecell for therapeutic design. These tools target practical bottlenecks in preparing scientific data and interpreting results, potentially shortening the path between computation and researchers’ next experimental decisions.

AI Hubs United original short summary of the source linked above. It is not the original article; read the full story at the source.