Maker of the GPUs and AI software stack that train and serve most large language, speech and translation models.
NVIDIA designs the graphics processors and accelerated-computing systems on which most modern AI models, including the large language and speech models behind machine translation, are trained and served. Founded in 1993 in California as a graphics-chip company, it became the central supplier of AI compute with its data-center GPUs and the CUDA software platform. For the language industry, NVIDIA matters in two ways: its hardware costs shape the price of running LLM-based translation at scale, and its NeMo and Riva toolkits provide open speech recognition, text-to-speech and neural translation models that developers use to build multilingual voice products. LocStock tracks it as a bellwether for AI infrastructure spending.