Nvidia is set to launch a new chip, called Rubin CPX GPU, designed specifically for extensive AI applications. The Silicon Valley chipmaker made the announcement at the AI Infra Summit in Santa Clara, California.
This new graphical processing unit (GPU), expected to become available at the end of 2026, will enable AI systems to manage million-token software coding and generative video with improved speed and efficiency.
The Rubin CPX operates in conjunction with NVIDIA Vera CPUs and Rubin GPUs within the new NVIDIA Vera Rubin NVL144 CPX platform.
Nvidia Founder and CEO Jensen Huang said, “The Vera Rubin platform will mark another leap in the frontier of AI computing — introducing both the next-generation Rubin GPU and a new category of processors called CPX.”
“Just as RTX revolutionised graphics and physical AI, Rubin CPX is the first Compute Unified Device Architecture (CUDA) GPU purpose-built for massive-context AI, where models reason across millions of tokens of knowledge at once.”
The company said that AI models may require up to 1 million tokens to process an hour of video content, which is challenging for traditional GPUs.
To address this, the Rubin CPX incorporates video decoders and encoders, along with long-context inference processing, into a single chip. This integration provides enhanced capabilities for long-format applications, including video search and high-quality generative video.
Built on the NVIDIA Rubin architecture, the Rubin CPX GPU features a cost-efficient monolithic die design with powerful NVFP4 computing resources. The GPU delivers up to 30 petaflops of compute with NVFP4 precision, providing high performance and accuracy.
It is equipped with 128GB of GDDR7 memory to support demanding context-based workloads and offers three times faster attention capabilities compared to the NVIDIA GB300 NVL72 systems.
The Rubin CPX will be available in various configurations, including the Vera Rubin NVL144 CPX. According to Nvidia, the Vera Rubin NVL144 CPX allows companies to achieve significant monetisation, with potential token revenue of $5bn for every $100m invested.
AI companies, such as Cursor, Runway and Magic, are already exploring the potential of Rubin CPX to enhance their applications.Nvidia Rubin CPX will be supported by the complete NVIDIA AI stack, ranging from accelerated infrastructure to enterprise-ready software.
