At the conference «FI Day. Blockchain: On the Threshold of the Financial Era», representatives of two locomotives of artificial intelligence development in Russia — Sberbank and T-Bank — spoke about why it is still lagging behind OpenAI and Google products. One of the main reasons is the lack of necessary computing power and territories for the construction of large data processing centers, which are important for the development of generative AI.
According to Vedomosti, the chairman of the board of Sberbank, German Gref, said that the bank used to build data centers with a capacity of 40 MW, and this was considered «very much». Now Sber is not even considering sites with a capacity of less than 200 MW for construction, «and better — 300–400 MW». At the same time, as Gref recalled, 300–400 MW is the energy consumption of a nuclear power plant unit. It is not easy to find such capacities, and, according to him, the shortage of computing power is one of the main reasons for Russia's lag behind OpenAI and Google AI products by 6–18 months.
The head of T-Bank, Stanislav Bliznyuk, confirmed the existence of this problem: it is very difficult to find a place even for two data centers with a capacity of 50 MW. He also noted the issue of tariff adequacy and the limited amount of capacity.
Representatives of VTB and the consulting company «DC Square», speaking about this problem, touched upon the difficulties of arranging future sites for Russian data centers. It's not just about providing capacity, but also about improving the territory for data processing centers. They must have access to several main communication channels, be located away from residential areas and social infrastructure facilities, and at the same time provide employees with the opportunity to get to work without problems.
Read materials on the topic:
Russia must develop its own sovereign artificial intelligence — Putin
Sberbank's AI solutions may appear in Lada and other AvtoVAZ cars