Nvidia has introduced Blackwell Ultra, a new version of its flagship AI chip architecture that the company says can deliver up to 40% more performance for AI training than the original Blackwell lineup. The launch underscores how quickly the semiconductor giant is pushing to extend its lead in the market for advanced AI hardware.

The new chips are aimed at data centers and large-scale computing systems that power generative AI models, where demand for faster processing and higher efficiency continues to surge. Nvidia framed the upgrade as a significant step forward for customers building and deploying increasingly complex AI workloads.

The announcement comes as competition intensifies across the AI hardware sector, with major technology companies and chipmakers racing to secure supply and improve computing power. Nvidia remains a central player in that contest, and the Blackwell Ultra rollout is likely to be closely watched by cloud providers, enterprise buyers, and investors.

Reuters reported the launch as part of Nvidia's broader effort to keep its product line ahead of rivals in the fast-moving AI chip market. For buyers, the key question will be whether the performance gains translate into practical advantages in cost, efficiency, and deployment at scale.