AI Data Centers Pose Regulatory Challenge, Jeopardizing Climate Goals – Study
A new research paper calls for binding renewable energy and efficiency targets for data centers amid the AI boom.
With demand for AI and large-scale data processing skyrocketing, the environmental impact of data centers – which serve as the backbone of these operations – is also increasing.
A new research paper examining the environmental and regulatory landscape highlighted a pressing need for policies addressing the rising energy and water usage associated with AI data centers
The report highlights gaps in both EU and US frameworks and proposes 12 actionable steps, including binding renewable energy and efficiency targets, to align the sector with global climate goals.
EU Data Center Regulations: A Promising Start Amid Complex Challenges
The paper, "AI, Climate, and Regulation: From Data Centers to the AI Act," says the EU’s recent data center regulatory developments mark a “good first step” but requires further development by including binding renewable energy and efficiency targets for data centers
The EU’s Energy Efficiency Directive (EED) and its supporting Delegated Regulation mandate data centers with a power demand above 500 kW must report energy and water consumption annually.
This move aims to increase transparency around resource usage and create accountability for environmental impact.
As AI’s power demands soar, data centers face growing pressure to balance efficiency and sustainability (Image: Alamy)
Germany’s recent Energy Efficiency Act builds on the EED’s requirements, applying regulations to data centers as small as 300kW and setting targets for renewable energy use. It requires data centers achieve 100% renewable energy reliance by 2027.
The act also mandates data centers disclose energy consumption attributable to each customer, providing clients with insights into their carbon footprint.
Philipp Hacker, co-author of the paper and chair for law and ethics of the Digital Society at the European New School of Digital Studies (ENS) at European University Viadrina, Frankfurt an der Oder, told Data Center Knowledge achieving 100% reliance on renewable energy by 2027 would be challenging.
The challenge would be especially daunting for smaller data centers with limited access to renewable energy sources.
“This is why, as we argue, there may be a need for larger AI model providers and data centers to not only consume renewable energy but also actively invest in and develop new sources of renewable energy to meet these ambitious goals,” he said.
Read more of the latest news on data center regulations
Hacker explained to comply with the EU’s reporting requirements, data centers will need to install appropriate measurement devices.
“While these devices are generally available, they are not yet consistently installed across facilities, so implementation efforts should focus on broad deployment of this infrastructure,” he said.
Kai Ebert, report co-author and PhD researcher at Viadrina, told Data Center Knowledge that under the EU’s taxonomy, nuclear power and natural gas are both considered renewable energy.
He calls this designation a “very controversial decision” but added large European data center operators might end up going a similar way as big US tech firms are now – heavily investing in nuclear power.
“If data center-specific renewable energy targets are implemented they will render the sector more carbon-neutral but not necessarily more sustainable,” he said.
Softly, Softly: Data Center Regulation in the US
Across the Atlantic, the US has introduced a softer regulatory approach. The proposed AI Environmental Impacts Act, currently pending in the Senate, emphasizes voluntary reporting.
It instructs agencies like the Environmental Protection Agency (EPA) and the National Institute of Standards and Technology (NIST) to conduct studies and set guidelines.
This approach could offer a foundation for future, more concrete environmental regulations.
From Hacker’s perspective, the US presents a more complex regulatory landscape due to its patchwork of state-level regulations.
“Political challenges in mandating renewable energy targets add further complications, making implementation more difficult across the US,” he said.
However, the report noted both the EU and US frameworks have gaps – notably the lack of enforceable efficiency standards at the EU level – leaving much of the responsibility for environmental practices to the discretion of individual member states.
While Germany’s targets could serve as a regulatory model for Europe, these standards are not yet universally applicable across the EU.
The US, meanwhile, continues to rely on recommendations rather than strict compliance measures, potentially slowing environmental progress.
John Hodges, a lawyer with HWG and co-author of a September paper on the need for data center energy efficiency, said he anticipates further action at the state level, regardless of federal moves.
“The federal government may stall, but states are where much of the regulatory action is happening,” he said. “What’s going to happen next year is an open question at this point.”
He added current US regulations on data centers vary, with some mandatory energy requirements and voluntary standards like the Energy Star program, which is “highly prized for efficiency”.
Regulatory Actions Proposed
Considering these challenges, the team of German academics behind the latest research paper proposed 12 key regulatory actions to enhance environmental accountability within data centers.
These include strengthening energy reporting obligations, closing regulatory loopholes for open-source AI, and setting renewable energy targets at the EU level.
The authors of the paper also suggest imposing peak-hour energy restrictions for AI operations, mandating renewable energy creation by data centers, and introducing tradable energy budgets to incentivize efficient energy use.
The researchers’ 12 proposals span four main regulatory areas:
Energy and Environmental Reporting Obligations: Include energy usage from AI inferences, indirect emissions, and water consumption in reporting requirements, and set reporting standards at the cumulative server level for accurate data tracking.
Legal and Regulatory Clarifications: Define the responsibilities of AI providers, including those modifying existing models, eliminate the reporting exemption for open-source models, and establish binding renewable energy and efficiency targets for data centers.
Transparency and Accountability Mechanisms: Increase public access to environmental disclosures, require energy consumption data for High-Risk AI (HRAI), and integrate environmental impact assessments into AI risk management.
Future-Oriented Measures: Introduce peak-hour restrictions for AI operations, require renewable energy development by AI and data center firms, and establish a tradable energy budget system to manage AI-related energy demand.
Hacker said data centers will likely need to invest in advanced load-balancing solutions to distribute demand more evenly over time.
They may also need to invest in the most efficient infrastructure and technologies available to reduce peak energy use and better align with environmental restrictions.
He admitted that overall, meeting these regulatory requirements will demand substantial efforts from data centers.
“It’s a big ask,” Hacker said. “But it reflects the broader need for all sectors to undergo significant changes in response to climate change mitigation and adaptation.”
About the Author
You May Also Like