Hrvatski Telekom is using AI-led data center cooling optimization at its Zagreb data center following the completion of a Siemens-led efficiency project. The deployment introduces machine-learning controls and sensor-based monitoring to reduce energy consumption while improving temperature stability across critical IT systems.
AI Controls Replace Static Cooling Models
The project, known as White Space Cooling Optimization (WSCO), was delivered by Siemens in cooperation with Croatia’s largest telecommunications provider. As a result, cooling operations at the facility now adjust dynamically to real-time conditions inside server rooms.
According to Siemens, the system relies on an intelligent sensor layout combined with AI-driven control logic. Consequently, cooling output is matched more closely to actual demand, limiting excess airflow and unnecessary equipment runtime.
Martin Lang, head of the Smart Infrastructure Buildings business unit at Siemens Austria, said the solution enables continuous learning and automated decision-making. As cooling behavior adapts over time, energy savings in the six-figure euro range are expected, while the risk of heat-related failures is reduced.
Operational Gains at the Zagreb Facility
Meanwhile, Hrvatski Telekom reports measurable operational improvements. Ivan Visković, director of the company’s Core Network and Services Sector, said temperature control has become more consistent across system rooms.
In addition, hot spots have been eliminated, cooling units now operate fewer hours, and maintenance requirements have declined. As a result, both operating costs and thermal risk have been reduced.
Sensor Data and Machine Learning Drive Efficiency
At the technical level, WSCO combines hardware, software, and process changes. Wireless thermistors track temperature variations at the rack level by measuring electrical resistance. That data is then processed by an AI engine, which regulates fan speeds and airflow automatically.
Over time, a continuous optimization loop is applied. First, operating conditions are analyzed. Next, the system adjusts cooling parameters. Then, machine learning refines future responses based on observed outcomes. Through this process, up to 99 percent of hot spots can be removed, according to Siemens.
Beyond efficiency gains, the system also strengthens preventive maintenance. Faulty components are identified earlier, allowing corrective action before failures occur.
Hrvatski Telekom provides fixed and mobile telephony, internet, wholesale, and data services across Croatia. With AI-led data center cooling optimization now in place, the company is positioning its core infrastructure to operate more efficiently as digital demand continues to rise.
