Nvidia introduced its Rubin platform at CES 2026, promising up to 90% lower token costs for specific AI models. This reduction could expand the viability of various AI applications for businesses.

The Rubin platform from Nvidia features Vera CPUs, Rubin GPUs, and proprietary networking technology. The platform will offer a rack-scale solution with 72 GPUs and a small-scale system with 8 GPUs in the latter half of 2026.

Lower AI inference costs on Nvidia’s Rubin platform could revolutionize the industry, facilitating more economical AI usage for enterprises. Industry experts emphasize the need for AI to deliver tangible ROI in 2026 to sustain growth and investments in the field.

Read more at Nasdaq: NVIDIA’s Next Chip Isn’t Just Faster — It Could Make AI 10 Times Cheaper to Run