NVIDIA has announced its intention to release Rubin, its next-generation artificial intelligence (AI) graphics processing unit (GPU), in 2026. This comes just three months after the announcement of the Blackwell GPU, marking a swift mention of a next-generation GPU.
According to Reuters on the 2nd (local time), NVIDIA CEO Jensen Huang revealed plans to launch a series of AI semiconductor products called Rubin in 2026, following Blackwell Ultra in 2025, during his keynote speech at an event held at National Taiwan University in Taipei, ahead of Computex 2024. Blackwell was first revealed in March and is currently under production.
Huang explained that the Rubin chip series will include a new GPU, central processing unit (CPU), and networking chip. However, he refrained from giving detailed specifications for Rubin.
He also mentioned that NVIDIA’s proprietary CPU will be called Vera, and Rubin will adopt the 6th generation high-bandwidth memory (HBM) HBM4. He further stated, “The Blackwell Ultra GPU, which is scheduled for release next year, will be equipped with the 5th generation HBM, HBM3E.”
In addition, he unveiled the GB200 NVL72 server that was produced in collaboration with Taiwan’s Foxconn (Hon Hai Precision Industry).
GPUs, which are semiconductors that process large amounts of data in parallel, are primarily used in the AI field. NVIDIA, which controls about 80% of the AI chip market, holds a dominant position as a major supporter and beneficiary of the rapidly growing AI development. Huang stated, “We plan to release a new AI chip every year and accelerate the release schedule every two years.”
Most Commented