What happened
The German digital association Bitkom released a study projecting a drastic increase in electricity demand for German data centers and IT installations, driven primarily by the expansion of artificial intelligence and cloud services. The report estimates current 2026 consumption at 23 billion kilowatt-hours, with projections reaching 56 to 70 billion kilowatt-hours by 2035. AI and high-performance computing are expected to consume 55% of total capacity by 2035, up from 20% in 2026.
Bitkom, the German digital association, published findings from a study conducted with the Borderstep Institute regarding the energy consumption of data centers and smaller IT installations in Germany. The report states that in the current year 2026, these facilities consume approximately 23 billion kilowatt-hours of electricity. This volume is equivalent to the annual consumption of roughly 6.5 million households, exceeding the total private household consumption of the states of North Rhine-Westphalia or Bavaria.
The study projects a significant upward trajectory in energy demand. By 2030, consumption is expected to rise to between 35 and 38 billion kilowatt-hours. By 2035, the demand could reach between 56 and 70 billion kilowatt-hours. The primary driver identified for this surge is the boom in generative and industrial AI applications, alongside general cloud service expansion.
Specifically regarding AI, the share of total German data center capacity used for AI and high-performance computing is estimated at 20% in 2026. This proportion is projected to increase to 55% by 2035. In terms of absolute connection power, this represents an increase from 670 megawatts in 2026 to approximately 5,300 megawatts by 2035, nearly an eightfold increase.
The industry reports that power supply bottlenecks are now the dominant obstacle for new projects. According to the Borderstep Institute study cited by Bitkom, around 80% of surveyed operators and industry experts identified limited grid connection capacities as the greatest challenge. Bitkom President Ralf Wintergerst stated at the federal government's Digital Summit that the AI competition is also a competition for computing power, and falling behind in infrastructure leads to falling behind in AI.
Why it matters
This projection highlights a critical infrastructure bottleneck for the European AI ecosystem. As AI workloads become the dominant driver of data center energy consumption, the physical limits of the German power grid and permitting processes threaten to constrain the region's competitive position in the global AI race. The study indicates that without policy interventions such as reduced electricity taxes or faster planning procedures, companies may relocate infrastructure abroad, increasing dependency on foreign data centers. This shift has significant implications for data sovereignty, energy policy, and the long-term sustainability of AI deployment in Europe.
The projected tripling of energy demand underscores the physical constraints facing the deployment of large-scale AI models in Germany. As AI workloads become the majority of data center consumption, the efficiency and availability of electricity become as critical as the algorithms themselves. This creates a direct link between energy policy and AI competitiveness.
Bitkom argues that companies will not delay their digital transformation due to a lack of domestic data centers. Instead, they will build infrastructure abroad, leading to increased dependency on foreign data centers. This poses risks to data sovereignty and the local economic benefits associated with digital infrastructure investment.
The study highlights a policy gap, with the industry calling for targeted relief measures such as a reduction in electricity tax or an industrial electricity price for data centers. Additionally, faster planning and permitting procedures are requested to address the grid capacity bottlenecks that currently hinder project development.
Investments in German data centers are expected to reach a record 23.5 billion euros in 2026, a 57% increase compared to the previous year. This capital influx indicates strong market confidence in the long-term growth of AI and cloud services, but it also intensifies the pressure on the existing energy grid to accommodate these new loads.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
What should a useful AI forecast state?
What to watch next
Monitor legislative responses in Germany regarding electricity taxes and industrial power pricing for data centers. Track the actual timeline for grid connection approvals and whether the projected 57% increase in 2026 investments materializes. Observe if major AI providers announce shifts in infrastructure location due to German regulatory or energy constraints.
Legislative developments in Germany regarding the electricity tax (Stromsteuer) and potential introduction of specific industrial power pricing mechanisms for data centers. Any changes to these policies could significantly impact the cost structure of AI infrastructure in the region.
The pace of grid expansion and the approval timeline for new data center projects. If permitting processes remain slow, the projected growth in AI capacity may be delayed or shifted to other European countries with more flexible regulatory environments.
Announcements from major AI and cloud providers regarding their infrastructure strategies in Europe. Watch for signals of relocation or reduced investment in Germany if energy costs or grid access remain unresolved.
Actual energy consumption data from 2026 to verify if the 23 billion kWh baseline and the 20% AI share hold true, providing a for future projections.