Nvidia created the CUDA platform that powers much modern artificial-intelligence computing.
CUDA is Nvidia’s parallel-computing platform and application-programming interface. It allows software developers to use Nvidia graphics processing units for general-purpose computation rather than limiting them to graphics rendering.
Nvidia introduced CUDA in 2006. The platform gave researchers and engineers tools for accelerating workloads such as scientific simulations, image processing, data analysis, and machine learning. Its libraries and developer ecosystem became especially important as neural-network models required large amounts of parallel calculation.
Graphics processors contain many smaller processing cores designed to handle numerous operations simultaneously. That architecture makes them well suited to the matrix calculations used in training and running many AI systems. Nvidia’s hardware, CUDA software, and specialized data-center products together helped the company become a central supplier to the generative-AI industry.
CUDA is not a separate company or chip. It is Nvidia’s software platform, and its close connection to Nvidia hardware is a major reason developers often build AI systems around Nvidia accelerators.