YADRO introduces hybrid cooling for high-density data centres
YADRO combines direct liquid cooling, a dry cooler, coolant distribution and monitoring. We review the architecture and the questions for a pilot.
Contents
On 24 July 2026, YADRO opened early access to an engineering architecture for cooling high-density data centres. It combines direct heat removal from processors and accelerators with an external circuit, coolant distribution, water treatment and central monitoring.
The relevant point for an enterprise is not an individual component but the attempt to coordinate computing and facilities infrastructure as one design. As rack density grows, cooling, power and the availability of engineering systems increasingly determine whether a workload can be installed in an existing machine room.
What the architecture includes
According to the vendor, Direct Liquid Cooling removes heat from CPUs and GPUs. The facilities layer includes the BOREY Poseidon circuit with a dry cooler, hydraulics and coolant distribution, together with RACKSCALE water treatment.
The BRO-300 module distributes coolant to the racks. YADRO says that key components are duplicated and that the architecture can be used in Tier III data centres. A Russian hardware and software controller manages the system, while the YADRO SUPRIM platform provides operational visibility. The software is listed in Russia’s software registry.
The vendor plans to open a demonstration and testing laboratory in the third quarter of 2026. This is a future milestone, not a completed independent test of a production site.
What the announcement does not establish
The publication does not provide PUE measurements from a named facility, a maximum supported rack density, equipment lifetime data or long-term operational results. A Tier III reference also does not automatically certify every data centre that deploys the system: facility level depends on the complete engineering design, procedures and implementation.
Before choosing the platform, a customer should establish:
- server heat profiles and permitted coolant temperatures;
- redundancy for pumps, controllers, power and external heat rejection;
- maintenance procedures that keep computing workloads online;
- water-quality, leak-detection and emergency-isolation requirements;
- compatibility across manifolds, cold plates and different rack generations;
- metrics and events exposed to external monitoring;
- spare capacity for future equipment generations.
How to run a pilot
A pilot should use a workload that resembles production rather than a single demonstration rack. Useful measurements include inlet and outlet temperature, coolant flow, pump and cooler power, stability under changing load and recovery time after a component failure.
Planned maintenance deserves its own test. The team should verify whether it can isolate a branch, replace a pump or controller, update software and return the equipment to service without breaching availability targets. Facilities events should feed the same observability layer as the IT systems; our IT infrastructure monitoring guide explains this operating model.
Practical conclusion
YADRO’s hybrid architecture addresses a real consequence of rising thermal density by treating servers, cooling and monitoring as one system. The potential benefit is a more predictable route to liquid cooling when an existing facility is modernised or a new one is designed.
The solution still needs an engineering pilot. Measured efficiency, server compatibility, maintainability, redundancy and documented leak or failure procedures are the decisive criteria. These scenarios should then become part of the wider disaster recovery plan.
Primary source: YADRO: integrated cooling for high-density data centres
