Nvidia CEO Jensen Huang Confirms Full Production of Vera Rubin Chips
Nvidia’s CEO announced that their Vera Rubin AI chips are ready for customers, aiming to revolutionize AI model costs significantly.
The Rubin chips will make running AI models cheaper, costing about one-tenth of what Nvidia’s Blackwell chips do. They can also train large models while using fewer chips than Blackwell. These improvements mean that it will be easier for companies to stick with Nvidia’s technology.
Microsoft and CoreWeave are two partners who will use Rubin chips soon. Microsoft has big AI data centers in Georgia and Wisconsin that will have thousands of Rubin chips. Some of Nvidia’s partners are testing Rubin with their next-gen AI models.
Nvidia is also working with Red Hat, which makes software for banks and other industries. They aim to create more products that run on the Rubin chips.
The Vera Rubin chip system includes several types of chips, like the Rubin GPU and Vera CPU. They are made using advanced technology from TSMC, which helps speed up processing. Huang believes these new chips are the best in the world.
Nvidia has worked on the Rubin system for many years. Huang announced the chips would come during a speech in 2024, and last year, he said they’d arrive in the second half of 2026.
However, it is not clear what “full production” means. Normally, high-tech chips are made in small amounts first while they are tested and then made in larger amounts later.