Nvidia’s CEO Jensen Huang launched “Rubin,” the next generation AI chip platform during Computex 2024 in Taiwan, which is scheduled for release in 2026. This announcemnet exceeds their recently discussed “Blackwell” platform that has not yet become widely available.
Huang also presented their annual AI acceleration platforms upgrade plan. This will involve having a “Blackwell Ultra” chip by 2025 and the “Rubin” platform by 2026. The Rubin will incorporate advanced memory and interconnect technologies for enhanced performance and power efficiency.
Apart from this, Huang introduced another CPU based on ARM architecture called Vera named after an astronomer Vera Rubin which belongs to the new accelerator board titled Vera Rubin.
Nvidia is confident despite the risk of cannibalizing current product sales through early disclosure of future plans. They have a stronghold on the data center GPU market used to build AI models like ChatGPT and will keep expanding by consistently coming out with newer technology options. However, they are up against competitors like AMD and Intel as well as moves toward diversifying GPU manufacturing on behalf of industry giants such as Sam Altman from OpenAI.
Source: arstechnica
