Cooling for data centers.

The cooling of a data center allows no stoppages: any failure compromises the service. APINA brings to this environment the same high-reliability industrial cooling engineering that operates on vessels and critical energy infrastructure, with natural refrigerants, redundancy and precise control.

The rise of AI has driven up rack density —from 15 kW to 80-120 kW—, making liquid cooling (direct-to-chip and immersion) the standard. Behind those systems, an efficient, continuous central cold production is needed: that is where APINA's technology fits.

Data centers already consume around 1.3 % of the world's electricity —between 240 and 340 TWh per year— and the figure is soaring with artificial intelligence. Electricity is, by far, their largest operating expense.

Cooling defines two metrics: PUE (total energy versus compute energy) and WUE (litres of water per kWh). Air uses more energy; water and immersion, less energy but more water. And efficiency has stalled: the average PUE has barely dropped since 2018.

The heat a data center expels —around 28 °C— is usually wasted. It is exactly the kind of low-temperature energy APINA turns into value: absorption refrigeration that produces cooling from that heat, or recovery to feed district-heating networks. Waste becomes a resource.

The pressure is global. The European Energy Efficiency Directive (2024) requires large data centers to report and reuse their heat; the F-Gas Regulation bans high-GWP refrigerants from 2027; and markets like Germany demand a PUE ≤ 1.3. From North America to the Middle East, the direction is the same: natural refrigerants and heat recovery are no longer an option but a requirement.