Nvidia will launch Rubin CPX in late 2026, a GPU designed to improve
AI efficiency by separating input processing from response generation. Targeting code and video generation, the chip can be deployed as standalone systems or server cards. Nvidia says a USD 100 million deployment could generate USD 5 billion in revenue.