Nvidia unveils AI chips for video, software generation

Nvidia (NVDA.O) announced on Tuesday that it plans to release a new artificial intelligence chip by the end of next year, designed to manage complex tasks like video creation and software development.

The chips, named "Rubin CPX," will be built on Nvidia's next-generation Rubin architecture, which will succeed the current "Blackwell" technology that marked the company's expansion into providing larger processing systems. As AI systems become more advanced, the demand for handling data-heavy tasks such as "vibe coding" or AI-assisted code and video generation is increasing.

AI models can require up to 1 million tokens to process just an hour of video content, a daunting task for traditional GPUs, according to the company. Tokens refer to the units of data processed by an AI model.

To address this challenge, Nvidia will incorporate several stages of the processing chain, such as video decoding, encoding, and inference (when AI models produce results), into its new chip.

By investing $100 million in these systems, Nvidia believes it could generate $5 billion in token revenue, as Wall Street places increasing focus on the returns from the massive investments in AI hardware. The race to build the most advanced AI systems has elevated Nvidia to the position of the world's most valuable company, securing a dominant share of the AI chip market with its premium, high-performance processors.

Copied