Nvidia designed the CUDA parallel-computing platform used widely with its graphics processors.
Nvidia introduced CUDA in 2006 as a software platform and programming model for using graphics-processing units (GPUs) for general-purpose computing. Before CUDA, GPUs were mainly associated with rendering images and video. CUDA allowed developers to write programs that assign highly parallel workloads to Nvidia GPUs.
CUDA became important in scientific computing, machine learning, engineering simulation, and data analysis. Its ecosystem includes development tools, libraries, and programming interfaces. The platform is proprietary to Nvidia hardware, which distinguishes it from standards intended to work across multiple GPU brands.
A common mix-up is confusing CUDA with the GPU itself. CUDA is the software and computing platform; Nvidia’s GeForce, A100, H100, and other products are hardware families that can support it. The company’s strong position in accelerated computing helped make Nvidia one of the world’s most valuable listed companies.