Elon Musk makes prediction on Nvidia’s Rubin AI chips

Tesla CEO Elon Musk has shared his thoughts on Nvidia’s newly unveiled Vera Rubin chips, revealed at CES 2026. While Musk acknowledged the impressive design and performance, he predicted that the technology will take another nine months to become operational at scale.

His comments came in response to a video shared by influencer Sawyer Merritt, which highlighted the architecture and capabilities of the Rubin chip.

What are Nvidia’s Rubin chips?

Nvidia CEO Jensen Huang introduces the Rubin chips as an upcoming successor to the Blackwell microarchitecture. The end-to-end AI computer platform comprises Nvidia’s first-ever six-chip AI system and features an extreme codesign approach, where GPUs, CPUs, networking cards, and data processing units are all designed in a highly collaborative way.

Huang stated that the Nvidia's Rubin architecture is capable of as much as five times the performance of their current system; the 50 petaflops of inference performance in the Rubin GPU runs on NVFP4 precision.

The Vera CPU is designed to support the application of data movement and the AI agents, whereas the NVLink 6, Spectrum X, ConnectX-9, and BlueField-4 are designed to overcome the current bottlenecks in the large-scale AI training/inference.

Despite the hype, Musk warned that scaling Rubin chips will take time. “It will take another nine months before the technology becomes operational at scale and the software can function smoothly,” he wrote on X.

Notably, Huang described Nvidia’s Alpamayo autonomous vehicle AI as a “ChatGPT moment” for autonomous vehicles. Musk expressed scepticism about distribution, calling it “super hard” to solve. Tesla AI Chief Ashok Elluswamy echoed Musk’s concerns, though Musk wished Nvidia success in advancing autonomous driving systems.