Nvidia CEO Jensen Huang revealed the Rubin AI Chips during GTC 2025, underscoring the company’s commitment to advancing AI technology. And at the same time driving major strides in AI computing and data center innovation.
Rubin AI Chips
Nvidia intends to transition from the Blackwell architecture to develop Rubin AI chips which will honor famed astronomer Vera Rubin. Nvidia will release the Rubin AI chips during the second half of 2026 to address the growing demands of sophisticated AI systems. Huang explains that these chips will achieve major performance improvements when utilized for AI training and inference activities.
Technological Advancements and Performance
The Rubin architecture achieves superior performance through high-bandwidth memory (HBM4) which resolves former AI accelerator bottlenecks. These performance enhancements could achieve massive improvements which will significantly boost processing efficiency during complex AI operations.
A Nod to Scientific Legacy
With the launch of its Rubin chips Nvidia carries forward its historical pattern of naming architecture after renowned scientists. Vera Rubin’s significant contributions to research about dark matter make her name fitting for this advanced technology.
Future Roadmap: Rubin Ultra and Feynman
The roadmap indicates that the Feynman architecture would debut in 2028 before the unveiling of the Rubin Ultra, which is scheduled for 2027. Other coming innovations by Nvidia are expected to strengthen their position as world leaders in the area of improvements in AI computing for scenes.
Implications for the AI Industry.
The AI sector will undergo substantial changes with the release of these AI chips. These enhanced chips will boost computation power allowing complex AI applications. To operate successfully across multiple sectors including autonomous vehicles and sophisticated data analytics.
If you want to read about NVIDIA Jetson: The Ultimate Game-Changer for AI and Robotics click here.
