Telehouse Germany Cuts Cooling Electricity Use by 10.4% with AI-Supported Cooling Optimization

Telehouse Germany and etalytics have completed a joint AI-supported cooling optimization project at Telehouse's Frankfurt data center, achieving a 10.4% reduction in electricity consumption on one of five cooling units. The result was delivered under live operating conditions and using existing infrastructure in a facility that was already considered well optimized, underscoring the potential for AI-driven efficiency gains even in mature data center environments.

The project relies on etalytics' etaONE platform, which combines digital twin modelling, predictive analytics and automated setpoint optimization within predefined operating limits. The system continuously identifies more efficient operating points as thermal demand, ambient conditions and equipment behavior change. Within the current optimization scope, etaONE adjusts two variables in the recooling circuit: cooling tower outlet temperature and recooling pump volume flow. The optimization accounts for site-specific constraints, including nighttime noise requirements that limit cooling tower flexibility.

The 10.4% reduction was determined by comparing optimized operation between April 14 and August 5, 2026, with historical periods under similar operating conditions, with the assessment accounting for ambient temperature and thermal demand. Rollout of the same optimization approach across the remaining units is already underway. Based on the performance achieved to date, the complete five-unit system is projected to save approximately 497 MWh of electricity per year. Under the conventional German electricity mix, this corresponds to around 180 tonnes of avoided CO2 emissions annually. Telehouse Germany has won the Infrastructure Sustainability Award at the Platform Global Awards 2026 for the project.

Beyond the energy savings, the deployment has improved operational visibility. etaONE monitors indicators of cooling-tower filter fouling, enabling Telehouse to identify changes earlier and plan maintenance more precisely. The platform operates within approved technical boundaries and alongside Telehouse's existing building automation system. Fallback mechanisms automatically return the plant to conventional control if defined requirements are not met, while operators can intervene at any time. During the reported assessment period, no incidents or limit violations occurred, and normal data center operations were not disrupted, including during scheduled maintenance.

"With AI-based optimization by etalytics, we are making energy flows and operating conditions transparent and manageable. For me, innovation is not about adding more and more technology. It is about achieving better results with less effort," said Michael Nicolai, CEO of Telehouse Germany. Dr. Thomas Weber, Co-Founder and Managing Director of etalytics, added: "This project shows what AI-supported optimization can achieve under real operating conditions: measurable efficiency gains in a live data center. The result was achieved in an already well-optimized environment and within clearly defined operating limits. That is what makes this approach scalable for mission-critical operations: measurable performance improvements without compromising reliability or operational control."

The current optimization approach is being rolled out across the remaining cooling units within the system, with further optimization of the chilled-water side also planned and additional potential identified across the wider cooling infrastructure. Telehouse also plans to expand its use of etaONE for integrated energy and operational reporting. Because the platform is directly connected to the underlying operational systems, reporting can draw on a consistent data foundation without requiring information to be manually consolidated from separate sources. Automated reporting and dashboard functions are intended to simplify reporting processes and provide a more consistent view of energy and operational performance. Longer term, the digital twin foundation can support increasingly complex energy strategies such as waste heat utilization, district heating integration and thermal storage.