According to the official NVIDIA blog, the company is expanding its CUDA-X scientific research tools. These include cuPhoton for astronomical and simulation analysis, cuPotential for accelerating molecular dynamics workflows, and the open-source cuSpa for cell polygon computation. These libraries are designed for scientific computing and research data processing, and can also serve as tools for AI agents. The acceleration data cited by the vendor comes from specific scenarios and does not imply a uniform speedup across all scientific computing workloads.