Summary
- UTELS reported damage to a Kyiv data centre containing core network equipment.
- Loss of power created network disruption while emergency teams worked on restoration.
- The incident follows other recent attacks affecting Ukrainian data centre and communications infrastructure.
A Russian attack on Kyiv has damaged a data centre housing core equipment for Ukrainian internet provider UTELS, disconnecting infrastructure from its power supply and causing network disruption during another sustained assault on the capital.
UTELS reported that emergency repair teams were working at the affected facility on 23 September. The company said loss of power to the equipment could interrupt network operation and access to services while recovery work continued.
The incident formed part of a wider series of attacks across Kyiv. Ukrainian authorities reported damage to transport, commercial, office, and other civilian infrastructure, while internet problems affected users across the city and surrounding region.
Ukraine’s judicial information systems also reported an outage linked to disruption on a backbone provider’s network. Court services were unavailable from 13:58 to 16:17 on 23 September before partial restoration, demonstrating how damage to communications infrastructure can propagate into public services even when the affected application is hosted elsewhere.
Physical attack collapses conventional redundancy assumptions
Data centre resilience is normally designed around component failure, utility outages, network faults, maintenance events, and other foreseeable technical incidents. Repeated military strikes create a different operating environment.
Generators and UPS systems can bridge the loss of normal utility power, but they depend on the building, switchgear, fuel systems, distribution paths, cooling equipment, and controls remaining physically intact. A facility that is directly damaged can lose several layers of redundancy simultaneously.
The same applies to connectivity. Multiple carriers and diverse fibre routes reduce exposure to an individual cable failure, but the resilience benefit falls if routes converge geographically or if a wider attack damages several communications nodes within the same urban area.
Restoration therefore becomes less about transferring load automatically and more about whether technicians can safely reach the site, assess structural and electrical damage, isolate failed systems, restore power, and reconnect equipment while the threat environment remains active.
That operational constraint is visible in Ukraine’s telecommunications sector. Providers have spent years adding batteries, generators, redundant routes, mobile equipment, and repair capability, but repeated attacks continually consume spares, fuel, engineering time, and replacement hardware.
Kyiv’s digital infrastructure has been hit repeatedly
The UTELS incident is not isolated. DataCentral reported earlier in September that an attack affected the BeMobile data centre used by UCloud, temporarily interrupting websites hosted through the provider. A later Russian claim that another Kyiv facility operated by De Novo had been struck could not at that point be independently confirmed.
The 23 September incident differs because UTELS publicly reported that equipment at the affected data centre had lost power and that emergency teams were carrying out restoration work.
For operators, the repeated incidents expose a difficult balance between concentrating equipment in hardened facilities and distributing systems across enough sites that one strike cannot remove critical services.
Geographic redundancy only works where secondary locations have independent power, networking, staffing, and operational control. Moving workloads between cities can also increase latency or create additional network dependencies, while cloud failover depends on applications having been architected and tested to use it.
The immediate priority at the UTELS facility is restoration. The broader engineering lesson is that critical digital infrastructure can lose power not because a utility fails somewhere upstream, but because the building and network serving the equipment become part of the incident itself.
That distinction changes the resilience problem. Backup systems are designed to keep equipment running through failures around the facility. In Kyiv, operators increasingly have to plan for scenarios in which the facility is what has failed.
