Ukraine hits Russian data centres in blow to Moscow's AI capabilities

The strikes targeted five data centres operated by three major Russian technology companies: Yandex, Sber and VKontakte. These firms provide much of Russia’s consumer internet infrastructure. Damage therefore threatened everyday services, not just specialist technology systems.

Yandex facilities in Sasovo, Kaluga and Vladimir were hit on 7, 9 and 11 October. A Sber centre in Balakovo was struck on 14 October. A VKontakte facility near St Petersburg was attacked on 18 October. Their systems support card payments, taxi apps, navigation, messaging and cloud storage.

The companies have not disclosed the damage’s full scale. Russians reported persistent problems paying for goods, ordering taxis, using navigation and accessing cloud storage. Experts said repairs could take months or years, while Yandex’s damaged computing capacity could also slow Russia’s artificial-intelligence work.

A data centre is a facility that houses servers and the equipment needed to keep them operating. The article describes vast halls containing servers, power systems and cooling equipment. These buildings provide the physical foundation for digital services that people use through phones and computers.

Servers process requests, store information and run applications. A company can use them to handle card payments, match taxi passengers with drivers, provide navigation data or store files in the cloud. Reliable power and cooling are essential because the equipment is fragile and operates continuously.

When a data centre is damaged, services may become slow, unreliable or unavailable. Fires can destroy microchips and contaminate electronics with soot. The article reports persistent failures after the strikes, showing how physical damage to computing facilities can reach people nationwide through familiar online services.

The attacks affected five facilities in five Russian regions. Three belonged to Yandex, while Sber and VKontakte each lost one targeted facility. The strikes happened between 7 and 18 October, creating a concentrated sequence of attacks on major technology infrastructure.

Yandex sites in Sasovo, Kaluga and Vladimir were hit. The Sber data centre was in Balakovo, Saratov region, and the VKontakte facility was in the Leningrad region near St Petersburg. Together, these locations supported systems used throughout Russia rather than only by local customers.

The reported disruption was broad and persistent. Russians experienced difficulties paying for goods, ordering taxis, using navigation services and accessing cloud storage. The article also links the damaged infrastructure to messaging and nationwide cloud-service failures, although the companies have not disclosed the full scale of physical damage.

Large technology companies operate infrastructure that serves users far beyond the building where it is located. Payments, taxi bookings, navigation and cloud storage depend on shared computing systems. If those systems fail, customers across the country can lose access even when their own phones and local networks still work.

The article identifies Yandex, Sber and VKontakte as central to Russia’s consumer internet. Their facilities process card payments, support taxi apps, provide navigation, host cloud storage and run messaging services. A fire can damage servers, power equipment and cooling systems, while soot can harm surviving electronics.

The result is a chain reaction across everyday services. The article reports persistent difficulties paying for goods, ordering taxis, using navigation and accessing cloud storage. Because companies have not disclosed the damage, the exact technical failure points remain unclear, but experts expect repairs to take months or years.

Russia requires companies to store all data on servers inside the country under legislation intended to create a “sovereign internet.” This rule keeps national digital infrastructure physically within Russia. It may support domestic control, but it also limits where companies can place backups and computing resources.

The article says this concentration makes Russian facilities easier to target. Analysts describe data centres as soft targets because they contain large amounts of fragile equipment in halls that depend on power and cooling. Nick Reiners said Russia must defend a vast territory with insufficient air-defence resources.

Ukrainian infrastructure is less concentrated because Ukraine moved most of its data to servers in European countries after earlier attacks. That reduced the effect of later strikes. Russia’s domestic-storage requirement creates the opposite vulnerability: damage to a small number of facilities can affect services nationally, while relocating resources may pressure authorities to ease internet controls.

Artificial-intelligence development depends on powerful computing systems. Yandex’s three supercomputers are used to develop AI models, and the company’s most powerful system was housed at the Sasovo facility. Damage there could remove important computing capacity and delay research.

The article says Yandex’s supercomputers use Nvidia A100 processors. Russia cannot legally import these chips because of Western sanctions. Fires may also destroy microchips and other sensitive components, while soot can damage equipment that appears to have survived. Rebuilding therefore requires both hardware and access to restricted parts.

Experts expect the strikes to delay Russian AI development further. Before the attacks, Russia was already several years behind the United States and China, according to the article. Repairs could take months or years, and the global chip shortage may add difficulty. The attacks could also push companies to rent computing capacity in Europe.

Online services rely on several connected layers. Servers run the applications and process requests. Storage keeps account details, files, maps and transaction information. Computing power determines how quickly systems can analyse data and respond. Internet connections carry requests from users to data centres and return the results.

For example, a taxi app sends a booking request through an internet connection to servers. Those servers check stored account and location data, calculate a match and send the result back. A payment system similarly processes transaction information, while cloud storage retrieves or saves files on remote servers.

A failure at any layer can interrupt the service. Damaged servers cannot process requests. Lost storage can remove access to information. Power or cooling failures can shut equipment down, and damaged connections can prevent users reaching it. The article’s strikes disrupted payments, taxis, navigation and cloud storage because these services share physical infrastructure.