Nvidia unveiled the Rubin CPX, a new GPU optimized for processing context windows exceeding 1 million tokens, at the AI Infrastructure Summit. The processor is part of the upcoming Rubin series and is designed for disaggregated inference, enabling stronger performance on complex, long-context workloads such as video generation and software development.
The announcement highlights Nvidia’s continued push to expand its dominance in AI hardware, following record-breaking momentum in its data center business, which generated $41.1 billion last quarter alone. By extending context capacity at scale, the Rubin CPX is positioned to support the next wave of advanced AI applications. The GPU is expected to become commercially available by late 2026.
