Drones knock out Yandex's third data center in four days

Yandex has lost a third major data center in four days. On Sunday, October 11, the company said a drone attack had knocked out its site in the Vladimir region, east of Moscow. Reuters reports that there were no casualties, but some user-facing services are currently down. For a technology company of this size, three sites lost in a row is a stress test that no fault-tolerance architecture is designed for.

What we know so far

Public details are scarce. The only source is Yandex's own statement, relayed by Reuters and then TNW. Judging by the available text, the company has not disclosed the extent of the damage, the recovery timeline, or which services are affected.

• Site: a data center in the Vladimir region, east of Moscow.
• Cause of the outage: a drone attack, according to Yandex.
• Date of the announcement: Sunday, October 11, 2026.
• Tally: the company's third major data center in four days.
• Casualties: none.
• User impact: some services are unavailable; which ones is unknown.

The source material has no details on the two earlier incidents, so they cannot yet be compared by scale or consequences.

Why three sites in four days is a special case

The industry learned long ago to survive the failure of a single data center. The tools are availability zones, data replication between sites, and automatic traffic failover. The design assumes that failures are independent: a fire, a power cut, or a maintenance error at one site should not coincide with a similar problem at another.

A series of attacks breaks that assumption. Three sites in four days means the redundancy margin is being used up faster than it can be rebuilt. The remaining nodes take on extra load, and any new failure on them hits users harder. Engineers call this cascading risk. The danger lies less in the outage itself than in the fact that the system is running without a safety net.

AI workloads raise a separate question. Yandex has its own language models and a voice assistant, and these require GPU clusters with heavy power and cooling. The company has not said which capacity was housed at the affected sites. Any conclusions about the impact on its models would be guesswork.

In practice: what teams and customers should check

Even if you don't use Yandex services, this story works as a checklist for any infrastructure.

First, look at how your redundancy is actually set up. Two "zones" in one region or at one provider do not protect against regional events. Second, test failover in practice: failure drills show how many minutes or hours recovery really takes, not what a document promises. Third, keep copies of critical data and configurations outside your primary site, ideally with another provider or in another jurisdiction. Fourth, decide in advance which product features you can switch off to keep the core of the service running, and prepare a degraded mode.

For Yandex customers, the next few days will answer the main question: whether the unavailable services return within hours or the recovery drags on. Follow the company's official status pages, and do not plan critical launches for the coming days until the picture for the remaining sites becomes clear.