Nvidia developed the CUDA parallel-computing platform.
CUDA, short for Compute Unified Device Architecture, was created by Nvidia to let programmers use the company’s graphics-processing units for general-purpose parallel computation. Nvidia introduced CUDA in 2006, and its software ecosystem became important in scientific computing, machine learning, and artificial-intelligence research.
Graphics processors can perform many similar calculations simultaneously, making them useful beyond rendering images. CUDA supplied programming tools, libraries, and a software model that helped researchers and developers access Nvidia GPUs without designing specialized hardware for every task.
CUDA is not a graphics chip and is not an independent semiconductor company. A common confusion is between CUDA and OpenCL, a cross-platform framework supported by multiple hardware vendors. CUDA is specifically associated with Nvidia’s hardware and software ecosystem.