Russia's Yandex says data centre in Vladimir area shut down after drone attack

A drone attack damaged Yandex’s infrastructure in the Vladimir region and forced the company to shut down that data centre completely. The immediate effect was service disruption, while the company reported no injuries. This matters because data centres support the digital systems that users depend on.

Yandex said some of its services were unavailable after the attack. The shutdown was not an isolated incident. The company had already reported damage at another facility in the Kaluga region, where a drone strike partly disabled operations.

The Vladimir shutdown is part of a wider series of attacks affecting Yandex facilities. A separate attack had also shut down the company’s data hub in Sasovo. The article does not state how quickly Vladimir services will return or what backup systems Yandex is using.

A data centre is a physical facility that houses computer servers, data-storage systems, networking equipment, power supplies, and cooling systems. These machines process requests and store information for online services. Although users see an app or website, much of its work happens inside data centres.

For example, when someone searches, sends a message, or uses an online map, servers process the request and return information. A data centre can also provide the computing power needed to develop AI models. Its equipment must run reliably because many digital functions depend on it.

The article shows why this matters in practice. Yandex’s Vladimir data centre was completely shut down after a drone attack, and some services became unavailable. Other facilities were also affected. The article does not describe Yandex’s internal architecture, but companies generally use redundancy to reduce the effect of one facility failing.

The disruption reached three Yandex facilities in different parts of Russia. The company said its centre in the Vladimir region was completely shut down after a drone attack. It had earlier reported that its Kaluga-region centre was struck and partly put out of action.

A day before the Kaluga report, an attack shut down Yandex’s data hub in Sasovo, southeast of Moscow. That hub housed two of the three supercomputers used to develop the company’s AI model. These locations therefore affected both ordinary services and AI-related computing.

The article calls this the third major data centre taken out of action in the last few days. It does not quantify the total amount of lost capacity or say whether the facilities shared the same systems. It does state that some Yandex services were unavailable after the Vladimir shutdown.

When a data centre is damaged or shut down, the services running there may stop responding. Users might be unable to open an app, retrieve stored information, complete searches, or use other online features. The exact effect depends on which systems were located there and whether alternatives are available.

The article gives a direct example. After the Vladimir-region data centre was shut down, Yandex said some of its services were unavailable. The Kaluga centre was partly put out of action, showing that disruption can also be limited rather than total. The article does not identify the specific services affected.

Companies can sometimes reduce user impact by shifting requests to other facilities or restoring systems from copies. Those protections require available capacity and functioning connections. The article does not say whether Yandex successfully rerouted traffic or recovered data, so the full operational effect remains unclear.

Yandex’s data centres matter because the company is heavily involved in Russia’s AI development and operates large-scale digital services. These facilities provide the computing, storage, and network equipment needed to keep services running and to develop AI models. Losing a centre can therefore affect both customers and research work.

The clearest example is Sasovo, southeast of Moscow. A Ukrainian attack shut down that data hub, where two of the three supercomputers used to develop Yandex’s AI model were housed. This concentration meant that one location contained most of the supercomputers named in the article.

The Vladimir and Kaluga incidents show a broader operational risk. Vladimir was completely shut down, while Kaluga was partly disabled. Yandex said some services were unavailable, but the article does not state whether AI development stopped or how much computing capacity remained elsewhere.

Companies often design online systems so one damaged facility does not stop everything. They may operate several data centres in different locations, copy important data between them, and keep spare computing capacity available. They can also use traffic controls to send users to functioning facilities.

For example, a company might store the same service data in two centres. If one centre fails, network systems can direct requests to the other. Regular backups can help restore information, while spare servers can replace damaged equipment. These methods work only when copies, capacity, and connections are available.

The article confirms that Yandex had other affected facilities in Kaluga and Sasovo, but it does not describe any backup arrangement. It also does not say whether traffic was rerouted or data duplicated. Therefore, these are general industry options, not measures confirmed for Yandex in this incident.

Training an AI model means exposing it to many examples and repeatedly adjusting its internal settings when its answers are wrong. Supercomputers divide these calculations across many powerful processors. They handle the data, calculate errors, and update the model through many rounds of training.

This requires substantial infrastructure because modern models may involve enormous datasets and billions of numerical settings. Processors must exchange results quickly, while storage systems supply training data. The machines also need reliable power, cooling, and networking. If part of the system is unavailable, training can slow or stop.

The article gives a concrete Yandex example. Its Sasovo data hub housed two of the three supercomputers used to develop the company’s AI model. An attack shut down that hub. The article does not state how training changed afterward, but the loss removed a facility containing most of those named supercomputers.