Nvidia developed the CUDA platform, widely used to program its graphics processors.
Nvidia introduced CUDA in 2006 as a parallel-computing platform and programming model for its graphics processing units, or GPUs. CUDA lets developers use GPU hardware for general-purpose calculations rather than limiting it to displaying graphics.
The platform became important in scientific computing, machine learning, and artificial intelligence because GPUs can perform many similar mathematical operations simultaneously. Nvidia’s GPUs and software ecosystem became central to training and running large neural-network models.
CUDA is proprietary to Nvidia hardware, which helps distinguish it from alternatives such as OpenCL and software frameworks that support multiple chip makers. Nvidia began as a graphics-chip company in 1993, but its data-center and AI businesses later became major contributors to its market value. The company’s famous GeForce brand is aimed primarily at consumer graphics and gaming.