Ukraine Is Now Striking Russia's AI Data Centers

Russia's AI data centers have become military targets, and it took just four days to prove it.

By Blumefield | October 13, 2026

A New Front Opens in the Drone War

For three years the war between Russia and Ukraine was fought over trenches, power plants and oil refineries. This month it reached the server hall. In the span of four days, Ukrainian long range drones struck three Russian AI data centers operated by Yandex, the technology group often described as Russia's Google, turning the physical backbone of the country's artificial intelligence program into a battlefield objective. The message from Kyiv is unmistakable. The compute that trains Russia's models and runs its cloud is now a target.

The sequence was deliberate. On October 8, drones hit the Sasovo facility in the Ryazan region, sparking a fire and forcing a complete shutdown. According to multiple accounts, Sasovo houses two of the three supercomputers Yandex uses to develop its flagship AI model. A day later, on October 9, strike drones reached a far larger site in the Kaluga region, disabling what Yandex described as multiple hardware modules. Then, overnight on October 11, a third wave struck a data center in Vladimir, east of Moscow, starting a fire and triggering outages across more than eighty Yandex services.

Yandex did not soften the blow in its own alerts. "Operations at the data centre have been completely suspended," the company told users on Telegram, warning that "the remaining resource configuration is considered unstable" and advising customers to switch to disaster recovery plans. For a company that commands roughly seventy percent of the Russian search market, running in emergency mode is not a footnote.

Inside the Compute Backbone

The scale of what was hit explains why these AI data centers matter so much. The Vladimir site was engineered as one of Yandex's foundational processing hubs, designed to hold up to 2,880 server racks and fed by two 25 MVA transformers and dedicated 110 kV lines running to a major regional substation. The Kaluga facility is larger still, spanning roughly 130,000 square meters with a designed power capacity of about 63 megawatts and more than 3,800 racks, each drawing as much as 15 kilowatts. These are not back office server closets. They are industrial installations on the same order as the hyperscale campuses being built across the United States and the Gulf.

What flows through them is the connective tissue of Russian digital life. The Vladimir outage alone took down Yandex Cloud, the YandexGPT API that developers across the country build on, and a long list of consumer applications. Sasovo reportedly served clients including Russian Railways and the mobile operator MegaFon alongside national media outlets. When a single strike can sever a railway operator, a telecom carrier and a generative AI platform at once, the concentration of so much capability inside a handful of AI data centers stops looking like efficiency and starts looking like exposure.

Yandex's response away from the battlefield was just as telling. The company quietly removed the locations of its data centers in four Russian regions from its own mapping platform, an implicit admission that the coordinates themselves had become a liability. Its shares fell about four percent on the Moscow Exchange during the week of the strikes.

Mirror-Like Fashion

Kyiv has not published a detailed claim of responsibility, and much of the granular strike reporting has flowed through Telegram channels rather than official military statements, a caveat worth keeping in view. The strategic logic, however, is not hidden. President Volodymyr Zelenskyy framed the campaign bluntly after the first strike, saying Ukraine "always respond in mirror-like fashion." His point was that Russia has spent the past month hammering Ukrainian data centers and telecom nodes, including targets in central Kyiv, in an effort to degrade the country's connectivity and digital economy.

That symmetry is the heart of the story. Both sides have concluded that the digital layer of a modern state is as legitimate a target as its grid or its fuel depots. Ukrainian military analysts cited in several reports argue that repeated strikes on enterprise AI data centers complicate Russian state logistics, disrupt corporate administration and degrade digital command and control. The goal is not only to dent a tech company's balance sheet. It is to pull on a thread woven through the entire economy.

Moscow is reacting to the new threat. The Kremlin has reportedly ordered data center operators to prepare for further drone attacks, and the Russian Defense Ministry claimed it intercepted 479 Ukrainian fixed wing drones in a single overnight period across sixteen regions, a figure presented as a Russian claim rather than an independently verified count.

The Fragility of Concentrated Compute

Strip away the specifics of this conflict and a larger lesson emerges, one that reaches well beyond Yandex. The global race to build artificial intelligence has produced an unprecedented concentration of value inside a small number of enormous, immovable buildings. AI data centers are the most capital intensive and least mobile assets in modern technology, and the strikes in Russia have just shown how that immobility becomes a vulnerability the moment a facility sits within reach of a cheap airframe carrying a warhead.

The economics are lopsided in a way that should unsettle infrastructure planners everywhere. A long range strike drone can be assembled for a figure in the tens of thousands of dollars. The targets cost hundreds of millions to build and years to bring online, and they cannot be moved when threatened. Transformers, cooling plants and high voltage feeds are single points of failure that take months to replace. An attacker does not need to destroy a GPU cluster to neutralize it. Cutting the power and burning the switchgear achieves the same end for a fraction of the effort.

This is a sovereignty question as much as a military one. Governments from Washington to New Delhi are pouring public money into domestic AI data centers precisely so that national capability sits on home soil. The Yandex strikes are a reminder that home soil is not the same as safe soil, and that resilience, meaning geographic distribution, hardened power and genuine redundancy, has too often been an afterthought in a build out optimized for speed and scale.

What the Strikes Signal

For Russia, the immediate damage is operational and reputational. Yandex will restore service, but the proof that its compute can be reached and burned will shadow every enterprise customer now weighing where its most sensitive AI data centers should sit. For Ukraine, the campaign extends a deep strike doctrine that has already reshaped how Russia defends its refineries and airbases, now pointed at the information economy.

For everyone else, the episode is a preview. As artificial intelligence becomes the organizing principle of national power, the buildings that house AI data centers will be treated as strategic assets, and strategic assets attract attack. Expect insurers to reprice coverage for large compute campuses, expect governments to fold data center resilience into critical infrastructure doctrine, and expect operators to start paying for the geographic dispersion they skipped on the way up.

The AI boom has been narrated almost entirely as a story of chips, models and capital. The fires over Vladimir are a reminder that all of it ultimately rests on concrete, copper and high voltage power, and that the physical layer can be the easiest thing in the stack to break. You can follow the fallout on Blumefield, where we keep tracking how the AI arms race is colliding with the physical world. For primary accounts of the strikes, see the statements relayed through Yandex and the reporting by Al Jazeera and Militarnyi. More sits in our ongoing coverage of AI infrastructure.