Nvidia, a prominent player in the chip industry, has announced that its latest AI superchip platform, Vera Rubin, is now in full production. CEO Jensen Huang revealed this development at a press event during the CES technology trade show. The Vera Rubin platform is designed to significantly reduce the cost associated with training and running AI models, offering a more cost-effective solution compared to Nvidia’s previous chip system, Blackwell.
According to Nvidia, the Rubin chips can train large models using a fraction of the chips required by its predecessor, potentially making advanced AI systems more accessible. This advancement is expected to further solidify Nvidia’s position in the hardware market, making it challenging for customers to consider alternatives.
Microsoft and CoreWeave are among the first companies to leverage Rubin chips for their services, with Microsoft incorporating thousands of Rubin chips into its upcoming AI data centers. Nvidia is also collaborating with Red Hat to expand product compatibility with the new chip system.
The Vera Rubin platform, named after the renowned astronomer Vera Rubin, comprises six distinct chips, including the Rubin GPU and Vera CPU, fabricated using cutting-edge technology. Nvidia’s commitment to innovation is evident in the sixth-generation interconnect and switching technologies used to integrate these chips seamlessly.
Source: WIRED