New Vera Rubin Platform from Nvidia Significantly Boosts AI Application Power
At the CES in Las Vegas, Nvidia CEO Jensen Huang announced the launch of a new server platform, Vera Rubin, which offers five times the computational power for chatbots and AI applications.
The Vera Rubin platform consists of six separate chips, includes 72 graphics processors, and 36 new central processors. This allows for significant performance growth in generating basic AI system units, known as “tokens,” even in large systems with over 1000 chips. Nvidia uses a unique data type that the company plans to make a new industry standard.
In addition to hardware innovations, Nvidia introduced the Alpamayo technology for autonomous driving, which goes beyond simply reacting to situations. It uses a “chain of thoughts” approach that allows the car to logically analyze complex road circumstances. This, according to Huang, marks a “ChatGPT moment for physical artificial intelligence,” where robotaxis become the first to receive such capabilities. The Mercedes-Benz CLA will be the first car equipped with Alpamayo.
The announcement came amidst increasing competition in the AI hardware market, where Nvidia faces pressure not only from traditional rivals such as AMD but also from its own clients, including Alphabet/Google.
| Key Features of the Vera Rubin Platform | |
|---|---|
| Number of Chips | 6 |
| Graphics Processors (GPU) | 72 |
| Central Processors (CPU) | 36 |
| Token Generation Growth in Systems with Over 1000 Chips | 10 times |
