Russia Is to Build New Supercomputer for AI Development Systems
Once completed, the system is expected to become a foundation for training large language models and scientific computing.
Sber is investing up to 500 billion rubles ($~5.4 billion) in the construction of a new supercomputer designed primarily for training and supporting the development of artificial intelligence systems.
Supercomputers are used to perform extremely complex calculations and are integral to scientific research in fields such as physics, chemistry, genetics, and climate process modeling. In recent years, powerful computing systems have also become critical infrastructure for training modern AI models that require massive memory and processing power.
Experience in the Field
This initiative is not new for Sber. The company previously participated in the creation of the Kristofari and Kristofari Neo supercomputers. According to the Top500 ranking as of last November, those systems were ranked 231st and 147th among the world’s most powerful supercomputers.
Recently, work began in Balakovo on another computing complex based at what the bank calls a “super data center.”
The new supercomputer is intended to serve as a foundational platform for the company’s future developments in artificial intelligence. Having its own high-performance computing base should speed up AI model training and reduce dependence on external infrastructure components.
