Nvidia is revolutionizing AI with new chips: Bigger, faster, better!

Nvidia is revolutionizing AI with new chips: Bigger, faster, better!

Nvidia, the semiconductor giant, is faced with the challenge of meeting the rapidly growing demand for computing power for artificial intelligence. At the GTC developer conference, CEO Jensen Huang presented the new “Vera Rubin” chip generation, which is scheduled to come onto the market in autumn 2026. These chips are intended not only to reduce the costs of running AI software, but also to meet increasing demands for computing capacity, such as krone.at reported. Huang explained that the world is increasingly relying on generating answers on the fly using artificial intelligence rather than relying on stored data.

A key example during the presentation was the Chinese AI model Deepseek R1, which requires 150 times more computing power than traditional AIs to calculate a simple task such as seating arrangements at a wedding. Huang emphasized that the real requirements for computing capacity lie in the response generation, not in the training phase of AI models as is often assumed. To meet the enormous demand, which is now 100 times higher than estimated a year ago, Nvidia also announced a further development of the existing Blackwell platform this year, which, according to Huang, promises increased efficiency, such as spiegel.de reported.

Nvidia has established itself as a key technology for artificial intelligence by deploying its chips in numerous data centers worldwide. Big tech companies like Google and Meta, as well as emerging AI startups like OpenAI, rely on Nvidia to fuel their AI offensives. This has catapulted Nvidia's business to unprecedented heights in recent years and shows the central role the company plays in future technological development.