European startups cut AI data center energy with smarter cooling, batteries and heat reuse

Europe's AI data center buildout is running into an electricity problem that more servers alone cannot fix, and a cluster of European startups is responding on three fronts: smarter cooling, flexible battery-based energy management, and reuse of the heat that facilities would otherwise throw away. According to a dev.to article that draws on Euronews reporting about the region's data center energy innovators, the common thread is treating power, cooling capacity and thermal output as things to be managed intelligently rather than accepted as fixed operating costs.

Cooling as the first target

Air-based cooling is under the most immediate pressure, because high-density AI hardware concentrates heat in ways that conventional airflow struggles to handle. EkkoSense has published case studies from optimization work at Virgin Media O2 sites showing roughly 15% savings on cooling energy. Instead of holding conservative settings across an entire facility, its software mines operational data to find where cooling performance can be improved.

Telefónica Germany has separately highlighted AI-driven digital twin capabilities for cooling optimization. A digital twin is a software model of a physical facility, and its practical value is not that it removes the need for cooling, but that it lets operators test and identify better operating conditions while staying within what the IT equipment requires.

Liquid cooling is also moving into real deployments. Submer has run immersion-cooling pilots with Telefónica, Telefónica Germany and other partners, placing hardware in a dielectric fluid rather than depending on airflow. The dev.to article notes that this is why liquid and other advanced cooling methods are drawing attention for dense computing environments, where air cooling becomes progressively harder to optimize.

Batteries as grid assets

GridBeyond has deployed behind-the-meter battery energy storage systems at Keppel DC REIT data centers in Ireland. The reported use case treats stored power as more than emergency backup: the batteries can provide flexible support to the electricity grid and help facilities shape their demand, with the deployment linked to lower carbon intensity and potential cost benefits rather than pure resilience.

Selling the heat

Deep Green is building a business around waste heat, piping thermal output to nearby users such as swimming pools and district heating networks. Its funding and deployment updates between 2024 and 2026 suggest heat reuse is being developed as a deployable local-energy model rather than a theoretical sustainability concept.

No universal checklist

The article is careful about limits. Heat reuse only works where a practical nearby heat consumer and suitable local infrastructure exist. Advanced cooling may require hardware, facility or operational changes. Battery projects depend on a site's energy profile and the applicable grid arrangements. The strongest case is therefore site-specific rather than a fixed menu of technologies.

For businesses buying cloud and AI services, the write-up draws a line between what is established and what is not. The evidence supports potential cost benefits from flexible battery use, but it does not establish that such savings are passed directly to cloud customers.

Why it matters

AI workloads raise power density and total consumption at the same time, which turns energy into an infrastructure constraint rather than a line item. Taken together, the three levers point to a broader reframing: the data center as part of a local energy system, where computing load, cooling, batteries, the grid and nearby heat demand all interact. EU-backed research and efficiency analysis cited in the article identify waste-heat recovery, AI-driven cooling and demand response or thermal management as active European research and deployment themes.

For cloud and AI customers, the practical consequence is that efficiency is becoming a due-diligence question. Reliability increasingly involves energy flexibility and thermal management, not only traditional backup arrangements, and sustainability claims carry more weight when tied to specific mechanisms such as measured cooling savings, actual battery deployments or named heat-reuse projects. Asking providers concrete questions about their data center efficiency initiatives, and the evidence behind them, is more informative than relying on broad environmental language. Over time, how well operators run their facilities will shape both the economics and the resilience of the cloud and AI services built on top of them.