Nvidia is kicking off 2026 with the early launch of its new Vera Rubin computing platform, following a record-breaking 12 months for the Rubin GPU’s predecessor Blackwell, fueled by the AI increase (or bubble).
Throughout a press briefing forward of in the present day’s keynote, Dion Harris, Nvidia’s senior director of HPC and AI infrastructure options, described Vera Rubin as “six chips that make one AI supercomputer.”
These six chips embody the Vera CPU, Rubin GPU, NVLink Sixth-gen change, Join-X9 NIC, BlueField4 DPU, and Spectrum-X 102.4T CPO. The platform will help Third-generation confidential computing and, in accordance with Nvidia, would be the first rack-scale trusted computing platform.
Nvidia claims the Rubin GPU is able to delivering 5 occasions as a lot AI coaching compute as Blackwell. The Vera Rubin structure as complete can prepare a big “combination of consultants” (MOE) AI mannequin in the identical period of time as Blackwell whereas utilizing 1 / 4 of the GPUs and at one-seventh the token value.
The Rubin launch was initially anticipated for late this 12 months. Its early arrival in the present day comes simply a few months after Nvidia reported record high data center revenue, up 66 p.c over the prior 12 months. That development was pushed by demand for Blackwell and Blackwell Extremely GPUs, which have set a excessive bar for Rubin’s success and served as a bellwether for the “AI bubble”. Services operating on Rubin can be obtainable from Nvidia’s companions beginning within the second half of 2026.
