Listen to the article
Ukrainian drones struck three major Yandex data centres inside Russia between 8 and 11 October, triggering fires, module failures and widespread service outages that reached users across the CIS, including in Uzbekistan. The attacks disabled at least three supercomputers central to the company’s artificial intelligence research, according to initial assessments reported by Kun.uz.
Three facilities, three nights
The first strike came overnight on 8–9 October, targeting a data centre in Sasovo, Ryazan Oblast. Opened in 2014, the facility is one of Yandex’s primary data-processing hubs, with an electrical capacity of up to 12 megawatts and tens of thousands of servers. A large fire broke out after the drone hit, forcing the centre to suspend operations and causing interruptions to services running on Yandex Cloud infrastructure.
On 9 October a second facility was struck — the data centre located in the Grabsevo industrial park in Kaluga Oblast. Commissioned in 2022, the site had been described by Yandex itself as one of the largest data-processing centres in Russia, with a design capacity of 63 megawatts. The attack put several of the complex’s modules entirely out of service.
Overnight on 11–12 October, drones hit a third Yandex data centre in the city of Vladimir. The sequential nature of the strikes amplified their cumulative effect on digital services not only within Russia but across other CIS countries. Users in Uzbekistan reported difficulties accessing a range of Yandex services during the period.
Supercomputers among the losses
Early analysis suggests the physical damage could deal a significant blow to Russia’s artificial intelligence development. Three of Yandex’s leading supercomputers — used to train neural networks, large language models and machine-vision systems — are reported to have been put out of action entirely, though the full extent of the damage has not been independently confirmed.
Two of the machines, named Chervonenkis and Lyapunov, were housed at the Sasovo facility. A third, Galushkin, was located at the Kaluga site. All three were dedicated to computationally intensive AI workloads. Chervonenkis is described as the most powerful supercomputer built in Russia in the past 15 years; when it was commissioned in 2021 it ranked 19th on the global TOP500 list of the world’s fastest supercomputers. Galushkin was similarly designed for high-performance tasks such as training large AI models.
The precise condition of all three machines and the exact scale of data loss remain unclear. No comprehensive damage assessment has been made public.
Recovery complicated by sanctions and chip shortages
Rebuilding the damaged infrastructure is expected to be difficult. Technology sanctions imposed on Russia have restricted access to advanced microchips, and the risk of further drone strikes adds another layer of uncertainty to any reconstruction effort. Together, these factors could significantly delay Yandex’s ability to restore its AI computing capacity to pre-attack levels.