Nvidia to release Vera Rubin AI computing platform ahead of schedule
Kyiv • UNN
Nvidia is launching a new Vera Rubin computing platform in early 2026. It will include six chips that will form one AI supercomputer and will provide five times more computing resources for AI training than Blackwell.
Nvidia is starting 2026 with an early launch of its new Vera Rubin computing platform, after a record year for Rubin's GPU predecessor, Blackwell, fueled by an AI boom (or "bubble"), The Verge reports, writes UNN.
Details
During a press briefing ahead of his CES 2026 keynote, Dion Harris, Nvidia's Senior Director of High-Performance Computing and AI Infrastructure Solutions, described Vera Rubin as "six chips that make up one AI supercomputer."
These six chips include the Vera CPU, Rubin GPU, 6th generation NVLink switch, Connect-X9 network card, BlueField4 DPU, and Spectrum-X 102.4T CPO. The platform will support 3rd generation confidential computing and, according to Nvidia, will be the first reliable rack-scale computing platform.
Nvidia claims that the Rubin GPU is capable of providing five times more computing resources for AI training than Blackwell. The Vera Rubin architecture can generally train a large AI model using a "mixture of experts" (MOE) structure in the same amount of time as Blackwell, while using four times fewer GPUs and at one-seventh the cost per token.
Rubin's launch was initially expected later this year. Its earlier release comes just a couple of months after Nvidia reported record-high data center revenues, which grew 66 percent year-over-year. This growth is driven by demand for Blackwell and Blackwell Ultra GPUs, which set a high bar for Rubin's success and served as an indicator of the "AI bubble." Products and services powered by Rubin will be available from Nvidia partners starting in the second half of 2026.