Digitalization is advancing—and with it, the expansion of data centers. But what many people don’t realize is that data centers are no longer a marginal phenomenon; rather, they are a structural driver of global electricity demand. A recent study by Allianz Trade shows that their impact on the climate has been massively underestimated to date. Particularly alarming: Artificial intelligence (AI) is one of the biggest drivers of emissions.

Data centers already consume about two percent of the world’s electricity supply, and that figure is on the rise. The United Nations (UN) recently warned in a new study that AI could lead to more climate-damaging emissions as well as higher water and raw material consumption, as reported by the Handelsblatt .

AI applications already account for 15–20% of data centers’ electricity consumption, and this share could rise to 40% by 2030. At the same time, the installed capacity of data centers is expected to double by the end of the decade. Global investment in this infrastructure totaled $580 billion in 2025.

But the figures are even more dramatic than previously thought: Data centers worldwide emitted 286 million metric tons of CO₂ in 2025—57% more than earlier estimates. Of that total, 43–60 million metric tons of CO₂ are attributable to AI alone. Without countermeasures, emissions could rise to as much as 643 million metric tons by 2030.

Location Matters: Why the Electricity Mix Determines the Carbon Footprint of Data Centers

Not all data centers are the same—at least not when it comes to their carbon footprint, as PV Magazin points out. The location and the local electricity mix make all the difference. In India, Indonesia, and Malaysia, emissions exceed 600 g of CO₂ per kilowatt-hour (kWh).

In Norway and Sweden, the figures are below 30 g CO₂/kWh—thanks to renewable energy.
At 329 g CO₂/kWh, Germany ranks in the middle globally, but is among the lowest performers in Europe. The reason: an electricity mix consisting of 20.6% coal and 10% biomass. In 2025, German data centers thus generated 8.4 megatons of CO₂—equivalent to 1.2% of Germany’s total emissions.

But it’s not just the carbon footprint that’s cause for concern: Data centers consumed 814 billion liters of water worldwide in 2025; by 2030, that figure could reach 1.8 trillion liters. The climate damage caused by data centers currently amounts to $68 billion per year and could rise to $154 billion by 2030.

AI as Part of the Solution: How “Green AI” Can Succeed

Despite the alarming figures, AI also offers potential solutions. Patrick Hoffmann, senior climate economist at Allianz Trade, says: “AI has the potential to reduce net emissions—provided that its efficiency gains scale faster than the expansion of infrastructure.” The key to green AI lies in clean electricity.
Europe has structural advantages in this area: Thanks to a comparatively clean electricity mix, the region could become a pioneer in climate-friendly AI infrastructure. However, without consistent expansion of renewable energy, this lead is at risk of being lost. Hoffmann emphasizes: “The course for climate-friendly AI infrastructure must be set now.”

Conclusion: Urgent Need for Action by Policymakers and the Business Community

The studies cited show that data centers are an underestimated factor in climate change. Their expansion urgently requires sustainable energy solutions.