Nvidia Unveils Rubin Platform at CES 2026
Nvidia introduced its Rubin platform for AI data centers at CES 2026, promising up to a 90% reduction in token costs for certain AI models. This platform, featuring Vera CPUs and Rubin GPUs, aims to make diverse AI applications economically viable for enterprises. A rack-scale solution with 72 GPUs and a smaller system with 8 GPUs are expected to be available in the second half of 2026.
As AI costs have significantly declined—falling by a factor of 1,000 for certain models over the past three years—Nvidia’s Rubin could prove crucial for enterprises struggling with high token expenses. Industry experts emphasize that 2026 is a pivotal year for AI, with a critical focus on achieving meaningful returns on investment. Major AI cloud providers, including CoreWeave and Microsoft, plan to deploy the Rubin platform as they explore the future of AI development.
