Energy: optimizing consumption and increasing efficiency

Artificial intelligence (AI), cryptocurrencies, streaming services – technological development is proceeding at a rapid pace, and with it the energy demand of digital applications is increasing. Despite growing data volumes and network expansion, we are pursuing the goal of keeping energy consumption at least stable in the medium term (2027 compared to the base year 2023, excluding T‑Mobile US). In recent years, we have been able to continuously reduce energy intensity – i.e., our energy consumption in relation to the volume of data transmitted. In addition, the expansion of renewable energies also plays an important role for us: they can help limit energy-related emissions and reduce dependence on fossil fuels – especially in combination with battery storage systems.

We deal in more detail with the topics of energy consumption, mix and efficiency as well as climate protection under “Climate change” and “General disclosures” in our audited Sustainability statement 2025. There we describe our goals and the plans for their implementation. You can also find more information on climate protection here in the CR report.

Milestones achieved, ongoing projects and goals

Since 2021, we have been sourcing 100 % of our electricity from renewable energies (Scope 2, market-based method) throughout the Group – from long-term supply contracts, direct electricity purchasing and certified guarantees of origin. To ensure stable energy consumption in the medium term, we are focusing on modernizing our grid infrastructure and operating our networks and data centers as efficiently as possible.

Grid infrastructure: innovations for energy efficiency

The operation of the mobile and fixed network infrastructure in Europe and the USA accounts for the largest share of our energy needs. We want to avoid that the constantly increasing data consumption of users is accompanied by a corresponding increase in energy consumption. That is why we are continuously working to make the energy consumption of our networks as efficient as possible. We are pursuing the goal of keeping energy consumption at least constant by 2027 compared to 2023 (excluding T‑Mobile US). This goal is supported by programs and investments in energy-saving measures for all energy sources, by the optimization of infrastructure and by the use of innovative technology components.

In Group-wide innovation projects, we are developing new approaches to our grid operation: For example, we are modernizing the grid infrastructure, relying on operational energy-saving functions and other technologies to improve energy efficiency. In mobile communications, for example, so-called power saving features are used (excluding T‑Mobile US). They automatically switch off certain functions when the network is only under low load.

AI can also help to control the use of energy in the mobile network in a more targeted manner. In Greece, for example, AI and machine learning algorithms are used for this purpose: They analyze the utilization of the network and adjust the energy consumption of a network component – the Radio Access Network (RAN) – according to demand. For customers, this has no noticeable impact.

We also use AI to adapt network capacities more closely to expected loads, for example by dynamically controlling individual mobile phone cells: Antenna power can be increased at major events such as open-air concerts or football matches; during periods of low demand – for example, at night or on non-match days – certain frequencies can automatically switch to a sleep mode.

Modern telecommunications infrastructure generates a lot of heat during operation. Therefore, cooling systems at telecom sites and data centers also play an important role in overall energy consumption. Data-based and automated systems can control cooling demand according to demand and adapt the output to the actual heat load.

A look at Greece shows what this looks like in practice, where intelligent automation is used at more than 1,500 network locations. They are part of a central energy management system and support the control of the cooling.

In addition, the “Optimal Temperature Set Point” application is used: AI or machine learning-based automation continuously evaluates the temperatures of network elements and derives a suitable setpoint for the air conditioning system in technical rooms.

In the Zagreb data center, too, sensors record the current heat situation every minute. With the help of this data, AI‑based controls control fans and cooling units so that they adapt flexibly to the respective load and are switched on and off as needed. In this way, the cooling is adapted to the heat generation. The “White Space Cooling Optimization” (WSCO) project is being implemented in Croatia jointly by Hrvatski Telekom and Siemens.

We describe further measures to stabilize energy consumption and increase energy efficiency in the “Deep Dive” section and in our audited Sustainability statement 2025. Closely linked to our approach to energy-efficient grids are also the topics of Operational resource protection and Raising awareness among employees here in the CR report.

KPI “Energy Intensity”

We have been able to steadily reduce energy intensity (i.e., our energy consumption in relation to the volume of data transmitted) in recent years. In the reporting year, energy intensity fell from 57 kWh/terabyte (2024) to 48 kWh/terabyte (2025). This corresponds to a reduction of around 16 %. Investments in modern technology have made this development of recent years possible – as has the shutdown of outdated network technologies. Detailed information on our KPI “Energy Intensity” can be found in our Sustainability statement 2025. As an indicator of the increase in efficiency in our data centers, we also use the so-called PUE value (Power Usage Effectiveness). Detailed information on this metric can be found in the “Deep Dive”.

Renewable energies: electricity supply contracts and large-scale battery storage systems

We purchase electricity from renewable sources through various instruments – we conclude electricity supply contracts with electricity producers, so-called Power Purchase Agreements (PPAs), purchase more electricity from renewable sources directly or acquire corresponding guarantees of origin

These supply contracts offer us price stability and can increase planning and investment security.

At the end of 2025, we purchased 31.7 % (2024: 36.2 %) of our electricity Group-wide via PPAs and self-generation. Excluding T‑Mobile US, the share was 26.1 % (2024: 22.6 %). We continuously monitor the electricity markets in the individual countries to identify new PPA options and conclude corresponding contracts if they make economic sense and fit our hedging strategy. One example of this is the photovoltaic park in Tützpatz, Mecklenburg, which went into operation in 2025 – the largest of its kind in Germany. We purchase the PV electricity generated there in full via a ten-year PPA. An overview of the annual development of the PPA share since 2022 can be found in the “Deep Dive”.

In 2025, PASM operated large-scale battery storage systems at its Bamberg, Hanover and Münster sites. They are used to temporarily store electricity from renewable sources and make it flexibly available. At the end of 2025, a total capacity of 16 MW was reached. The storage capacity is a total of 96 MWh (per day).

USA: diversified energy portfolio

To manage energy sustainably, T‑Mobile US deploys energy efficient technologies and focuses on sourcing renewable energy. The company has strategically built a diverse renewable energy portfolio by engaging in a range of projects, including medium- to long-term virtual power purchase agreements (VPPAs) with wind and solar farms, on-site and community solar energy contracts, and shorter-term retail renewable agreements.

This strategy helps to reduce price volatility and maintain a diversified energy portfolio.

Germany’s first AI factory

In the reporting year, the planning and construction of the Industrial AI Cloud, Germany’s first AI factory, was pushed forward in Munich. The aim is to provide AI computing capacity to companies, research institutions and the public sector. The project was created in partnership with technology companies such as NVIDIA, and the opening took place on February 4, 2026. In the future, the AI factory is to form a central component of a sovereign European AI infrastructure. In addition to aspects of digital sovereignty and industrial competitiveness, climate protection considerations also played a role in the planning: The AI factory is supplied with electricity from renewable energies. Furthermore, the AI factory uses a cooling concept in which water from a nearby stream is included in the cooling. Moreover, applications such as digital twins and simulation-based Physical AI approaches are to be supported. These can help companies to make development and production processes more resource-efficient and energy-efficient.

Europe needs AI solutions for industry. After all, AI has the greatest impact in mechanical engineering, manufacturing, logistics and robotics. What was missing so far was an infrastructure – open, safe and sovereign. We have now created this with our Industrial AI Cloud and with a technology stack Made in Germany. In addition, our AI factory is deliberately designed to be sustainable: Instead of a new building, we renovated an existing building. And thanks to high capacity utilization and modern architecture, we achieve maximum energy efficiency per computing unit. Our AI factory thus achieves a PUE value of less than 1.2.

Looking ahead

In the coming years, we want to further stabilize our energy consumption by continuously increasing our energy efficiency – despite rapidly growing data volumes. An important focus is also on the further expansion of renewable energies and large-scale storage solutions. In addition, we are increasingly relying on the use of digital and AI‑based solutions to further optimize energy consumption, especially in data centers.