Regulating Data Centers for a Sustainable Future: a World Overview
How governments and companies are racing to manage the environmental footprint of AI infrastructure
There is no doubt that artificial intelligence is one of the most rapidly expanding technologies worldwide. Increasingly, individuals and organizations are hoping to harness the power of AI to improve business processes, handle large amounts of data, and improve efficiency. Indeed, McKinsey & Co.’s 2025 State of AI Report shows that 88% of respondents make use of AI in at least one operational function, representing a ten percent increase from 2024’s report.
However, along with the increased use of AI comes a burgeoning demand for the infrastructure which supports artificial intelligence. Foremost amongst these is a growing need for data centers, the facilities which house the hardware which makes the training and daily delivery of AI applications and services possible. The demand is readily apparent; by the end of 2025 there will be over 11,000 data centers globally (1). But, the proliferation of these sites raises complicated questions around the environmental and social impact of AI. In this week’s newsletter, we dive into the impact of data centers, and how organizations and jurisdictions plan to address these concerns.
The Impact of Data Centers
Data centers are widely viewed as environmentally intensive facilities, primarily because of their enormous energy and water requirements. They are the backbone of AI and cloud computing, but that capability comes with a significant resource footprint.
On the energy side, data centers must power high-performance IT equipment and then cool it, often continuously, to prevent overheating. Their efficiency is commonly measured using Power Usage Effectiveness (PUE), which compares total facility energy use to the energy consumed solely by computing equipment. Even as operators strive for lower PUE values, rising AI workloads push up overall demand. In the United States, data centers are expected to need 22% more grid power in 2025 than the prior year, with demand projected to nearly triple by 2030 (2). Generative AI is a major driver as training AI models requires a tremendous amount of energy, and even more to power the usage of popular models like ChatGPT.
Renewable energy is frequently cited as the most important lever to reduce the climate impact of this growth. Many hyperscale operators have already committed to matching their electricity consumption with renewable power purchases (3). Yet the sheer speed of AI and cloud expansion raises a difficult question: can clean energy deployment scale fast enough, in the right locations, to keep pace with this new load on the grid?
Water is the second, but no less important, major pressure point. Large data centers can consume up to 5 million gallons of water per day (4), often for evaporative cooling but mainly for powering data centers through turbines using fossil fuel and water steam. This level of use is comparable to the daily water needs of a city of up to 50,000 people. Because much of the water used in cooling is evaporated, it cannot simply be reclaimed or recirculated, so even advanced efficiency measures may only partially mitigate demand. The result is growing concern in water-stressed regions, where communities face trade-offs between industrial water use and residential, agricultural, or ecological needs. Research has suggested that the water footprint of a single ChatGPT conversation can be comparable to that of a standard plastic water bottle (5). While this may seem small in isolation, it is incredibly significant when multiplied by millions of interactions per day.
Data centers also occupy substantial physical space. Many facilities exceed 30,000 square feet, with hyperscale campuses covering far more (6). Land conversion, construction impacts, noise from cooling equipment, and heavy electricity demand can all affect local communities. Polling by Beyond Fossil Fuels and conducted by Savanta in several European countries has found that residents are increasingly worried that new data centers could undermine clean energy transitions, strain water resources, or raise consumer energy costs (7).
Community Reaction and Growing Scrutiny
Unsurprisingly, the local impacts of data centers have become a recurring topic in news coverage and public debate. Communities near major data center clusters have raised questions about fairness: Who benefits from these facilities, and who bears the environmental cost? Survey data from multiple European countries indicates strong public support for rules that limit the impact of new data centers on energy systems, water supplies, and household bills. In other words, citizens are not necessarily anti-digital or anti-AI. However, they are increasingly insistent that growth comes with guardrails.
This tension is now shaping policy discussions worldwide, driving a wave of regulation, voluntary commitments, and new planning frameworks aimed at making data center growth more sustainable.
The Emerging Regulatory Landscape
United States
In the U.S., the regulatory picture is complex and often fragmented. At the federal level, a recent executive order aims to accelerate permitting for data center infrastructure, highlighting its importance for national competitiveness and AI leadership (8). At the same time, many states actively court data center investment through tax incentives. Virginia and Texas, for example, have become major data center hubs in part because of these financial benefits, and other states are now following their lead (9).
However, the rapid build-out has also sparked pushback. More than 230 organizations have signed a letter, drafted by Food & Water Watch, calling for a national pause on new AI-related data centers, citing concerns about water use, energy demand, and local environmental impacts (10). This debate illustrates the broader dilemma of how to balance the economic promise of AI-driven infrastructure with the need to protect communities and ecosystems.
European Union
The European Union has opted for a more structured, data-driven approach anchored in the recast Energy Efficiency Directive (EED). Under the updated directive, owners and operators of data centers with an installed IT power demand of at least 500 kW must report detailed sustainability key performance indicators (KPIs). These include energy consumption, Power Useage Effectiveness (PUE), temperature set points, waste heat utilization, water usage, and the share of renewable energy used. The information is submitted either to national authorities or to a central EU database established under Delegated Regulation (EU) 2024/1364 (11).
Recommended by LinkedIn
Germany
Germany has embedded strict requirements into national law through its Energy Efficiency Act (EnEfG), which helps implement the EED domestically. The law tightens energy efficiency expectations for data centers, including mandatory Power Useage Effectiveness (PUE) thresholds. Existing facilities must achieve a PUE of less than 1.5 from July 2026 and less than 1.3 from July 2030, while new data centers are required to operate below a PUE of 1.2 starting in July 2026 (12). Germany also places strong emphasis on waste heat reuse and requires operators to implement energy or environmental management systems, signaling a comprehensive shift toward high-efficiency, low-carbon digital infrastructure.
Other EU countries have followed Germany’s lead and adopted their own measures to comply with the EED. They have transposed laws that regulate data centers for energy/water use, require reporting, promote waste heat reuse, and in some cases have even imposed local development caps.
Australia
In Australia, the National Greenhouse and Energy Reporting (NGER) Act requires large data centers to report their emissions and energy use, building a consistent baseline for climate and policy decisions. For government procurement, the Digital Transformation Agency’s Data Centre Panel sets the bar even higher: data centers serving federal agencies are expected to meet stringent energy performance standards, often including high NABERS ratings, low PUE, and significant use of green power (13). This approach effectively uses public-sector demand to push the market toward more sustainable facilities.
Singapore
Singapore has taken a certification-led approach with the BCA-IMDA Green Mark for Data Centres (GMDC) scheme (14). Under this program, data centers are evaluated on energy and water efficiency, operational practices, and overall environmental performance. Facilities that demonstrate strong performance can achieve higher Green Mark ratings (such as Gold or Platinum), which signal to customers and regulators that they are operating at the leading edge of sustainability. This aligns with Singapore’s broader strategy of tightly managing limited land and resource constraints while continuing to grow as a regional digital hub.
China
China has explicitly framed data centers as critical infrastructure for its digital economy, and also as a priority area for decarbonization. In 2024, the government announced an action plan for the green and low-carbon development of data centers (15). By 2025, the average Power Useage Effectiveness (PUE) of Chinese data centers is targeted to be below 1.5, and the utilization rate of renewable energy is expected to increase by 10% annually. By 2030, the plan envisions data centers reaching internationally advanced levels of energy and carbon efficiency per unit of computing power.
Canada
Canada has taken a more guidance-driven approach so far, publishing federal best-practice guidelines for data center design and operation, with a focus on efficiency, right-sizing, and effective cooling strategies (16). Some provinces and municipalities are beginning to integrate these principles into local planning and building rules, particularly around energy performance and, in some cases, water use. While not yet as prescriptive as the EU or Germany, Canada’s trajectory suggests increasingly formal expectations for data center sustainability in the years ahead.
Conclusion
The environmental impact of data centers is not to be underestimated, particularly as the number and size of these facilities continues to grow. While AI seems here to stay, its demand for increased computational power is not to be taken lightly. By creating comprehensive guidelines, best practices, and legislation, governments can help ensure that the net benefit of this tool outweighs the negative impacts. Mandatory use of renewable energy, guidelines on Power Useage Effectiveness and Water Useage Effectiveness, repurposing of data center heat, and implementation of various technologies will be key. In the coming years we can expect more regulations coming up to tackle the growth of data centers worldwide and growing environmental and community concerns. More specifically, we might see more rules and regulations related to communities to ensure that the data centers benefit to the communities and do not affect their livelihood negatively. More to come, stay tuned…
This Week in Sustainability is a weekly email from Brightest (and friends) about sustainability and climate strategy. If you’ve enjoyed this piece, please consider forwarding it to a friend or teammate. If you’re reading it for the first time, we hope you enjoyed it enough to consider subscribing. If we can be helpful to you or your organization’s sustainability journey, please be in touch.
Sources:
Great read for anyone interested in shaping digital or sustainability strategy! Smart, forward-looking regulation can help balance innovation with resilience, ensuring AI growth aligns with climate goals and public trust.