One thing that united the polarized U.S. electorate in the November midterms was a deep distrust of data centers. Voters overwhelming chose to restrict data center development, with 44 out of 45 ballot measures breaking against data centers. A recent poll found that just 11% of people in the U.S. would support AI data center construction in their communities. Data centers seem to make manifest suspicions about big tech, job security, backroom deals, land grabs, corruption, and environmental harm. In the run-up to the 2026 U.S. midterm elections, spending on political ads mentioning data centers increased exponentially, with ads skewing overwhelmingly against data centers.
Data centers have operated in cities, suburbs, and further flung locations for two decades now. The change in public opinion is driven by the fact that today’s hyperscale data centers—often tied to AI training and cryptocurrency—are juicing electricity demand after years of relatively flat economy-wide consumption. In some areas, the increase in peak electricity demand is also driving up electricity prices at alarming rates.
The magnitude of data center energy demand is unprecedented, but there are time-tested ways to reduce peak electricity demand quickly: energy efficiency and demand management measures. Building performance standards (BPS) are a powerful policy lever driving resource efficiency. Adopting new BPS or amending existing BPS can shift data center development and operations in ways that benefit affordability and reduce greenhouse gas emissions.
BPS laws and data centers
A building performance standard is a policy that sets specific deadlines for existing buildings to achieve quantified standards of performance across one or more metrics, like energy use, water use, or greenhouse gas emissions. These standards become more ambitious over time, driving continuous, long-term improvement.
Of the jurisdictions that have adopted a BPS, seven cover data centers, as laid out in the table below. Likewise, the IMT Model Building Performance Standard includes data centers under its definition of covered buildings. Eight of the BPS policies neither address nor exempt data centers. Only one jurisdiction—St. Louis, Missouri—explicitly exempts data centers from its building performance standard on the grounds that a data center’s “energy use is not clearly reflective of building performance.”
Whether and How Each BPS Covers Data Centers
Local BPS
| Boston, Massachusetts | Not explicitly exempt |
| Cambridge, Massachusetts | Not explicitly exempt |
| Chula Vista, California | Not explicitly exempt; city currently rulemaking |
| Denver, Colorado | Covered; directed to use the manufacturing, agricultural or industrial building pathway |
| Evanston, Illinois | Not explicitly exempt; city currently rulemaking |
| Montgomery County, Maryland | Covered; performance target noted in the rule |
| Newton, Massachusetts | Not explicitly exempt |
| New York City, New York | Covered; building type designated in technical guidance |
| Seattle, Washington | Covered; target noted in the rules |
| St. Louis, Missouri | Exempt |
| Washington, District of Columbia | Covered; directed to use the standard target pathway |
| West Hollywood, California | Not explicitly exempt; city currently rulemaking |
State BPS
| Colorado | Covered; directed to use an individualized PUE score |
| Maryland | Covered; target noted in the law |
| Oregon | Not explicitly exempt |
| Washington | Not explicitly exempt |
See also IMT’s comparison of BPS laws
Many building performance standards apply solely within city borders. The high price of urban real estate keeps hyperscale data center development to a minimum within cities, pushing newer development outside urban cores. Building performance standards at the state level are better positioned to support energy efficient operation of data centers broadly.
How BPS tracking works for data centers
All 16 U.S jurisdictions with a BPS require the owners of covered buildings to annually report their energy consumption in ENERGY STAR’s free online platform (Portfolio Manager), which then automatically calculates the Energy Use Intensity (EUI) and greenhouse gas intensity (GHGI) for the building. This result is compared to a property type-specific target for EUI and/or GHGI, values set well in advance by the jurisdiction. If the building meets or does better than this target, then the building is considered in compliance with the law; if not, then the building owner faces financial consequences. Most jurisdictions also offer alternative compliance pathways or other flexibility mechanisms that building owners can apply for if meeting the target EUI and/or GHGI is infeasible by the required deadline or if the building is subject to other unusual circumstances.
Jurisdictions have opted for two general approaches to applying BPS to data centers:
(1) Performance Target: designating a broadly applicable target for data centers (as with other property types) that a data center must meet in order to comply with the law
(2) Custom Pathway: directing data centers to a custom compliance pathway that is unique to the individual data center
Performance Target pathway
Under Seattle’s Building Emissions Performance Standard Requirements Rule, data centers have a performance target to fully decarbonize by the final compliance date (2041-2045 depending on the size of the building). Montgomery County, Maryland’s BPS similarly requires that data centers consume no more energy than 206 kBtu/square foot of gross floor area by 2035; notably, data centers do not fall under Montgomery County’s process load exemption, which applies to many other buildings with large energy loads.
Custom Pathway
Data centers use energy in a fundamentally different way than other buildings, such that a custom regulatory approach may be more appropriate. As one example, Denver, Colorado has created a custom pathway for manufacturing, agricultural, and industrial (MAI) buildings to comply with its BPS (Energize Denver), defining these buildings as facilities where “energy is consumed in process loads,” a classification that, under Energize Denver, includes data centers.
Under the Denver MAI Alternate Compliance Option, a data center’s energy improvement is measured against its individual baseline. Instead of aiming to achieve a performance target based on use type, each data center has until 2032 to achieve a percent reduction in its energy usage against its 2022 baseline (the reduction amount is tiered in stringency depending on the starting point). Denver provides flexibility by offering several additional compliance options to buildings with an MAI designation.
The challenges of measuring data center resource use
Most BPS that cover data centers set requirements measured in energy use intensity (EUI) or greenhouse gas emissions intensity (GHGI). But, BPS regulate EUI (and to a lesser degree GHGI) with the goal of driving improvements in energy efficiency, and these metrics do not always accurately reflect a data center’s efficiency.
ENERGY STAR Portfolio Manager, which uses EUI as its primary metric for building energy performance, uses power usage effectiveness as its featured metric to assess data centers. Power usage effectiveness (PUE) is equal to total facility energy consumption divided by IT equipment energy usage. PUE offers insights into operational efficiency, providing a means of comparing across data centers and a design target for new data centers.
While EUI is a measure of energy consumption relative to building size, PUE measures the proportion of total energy consumption used to power IT equipment. According to the National Laboratory of the Rockies—home to highly efficient data centers of its own—an energy efficient data center would likely achieve a PUE of 1.2 or less; a score of 1 would be perfect, indicating that all energy consumed is used for IT equipment. However, a perfect score is likely impossible in practice because PUE includes in the numerator physical building systems that are not perfectly efficient and experience energy losses during normal operation.
PUE is considered a better metric for assessing the energy performance of data centers than EUI, as PUE focuses on the energy consumption of power supply and cooling systems in a data center instead of the energy consumption of the IT equipment. Many other entities also employ PUE to understand energy efficiency (National Laboratory of the Rockies, the Federal Energy Management Program, and the U.S. Green Building Council, for example).
Other metrics provide a more comprehensive look at a data center’s sustainability. One example paralleling PUE is Water Usage Effectiveness (WUE). Water usage effectiveness (WUE) assesses the water used for data center operation, including equipment cooling, and is defined as annual water usage divided by IT equipment energy and reported in liters per kilowatt-hour. WUE provides a window into the use of scarce freshwater resources and allows energy and water concerns to be considered together.
For example, the commonly used technology of evaporative cooling would likely result in a better PUE but a much worse WUE. This insight makes clear that jurisdictions should consider banning the new installation of evaporative cooling, which is energy efficient but water intensive. BPS should also hold facilities using evaporative cooling to a more ambitious energy requirement (e.g. a lower PUE) to compensate for the freshwater consumption of evaporative cooling.
Combining metrics like PUE and WUE offers a holistic look into overall sustainability performance of data centers. This is key as a diverse range of locations weigh the trade-offs among energy efficiency, water efficiency, carbon footprint, or other elements. Looking at a range of metrics allows a jurisdiction to better understand the impacts of a particular data center—and to demand better on behalf of its constituents. In the coming months, IMT will update the Model BPS Law with greater specificity regarding performance metrics for data centers.
What’s next for data centers
There are three ways that governments can address data centers. The first is to ensure existing data centers are covered by BPS policies. BPS requirements should be based on performance metrics specifically crafted for data centers.
The second avenue is to increase transparency. Despite the largest data centers producing an estimated 25.7 million metric tons of carbon dioxide emissions on an annual basis—almost as much as an entire country—many publish virtually nothing about their energy use, a level of secrecy that has raised public ire.
To provide much-needed transparency for data center resource consumption, IMT additionally recommends that benchmarking and transparency laws, which typically cover most large buildings, also cover data centers. Benchmarking refers to measuring a building’s energy use and comparing it to the average for similar buildings. Benchmarking and transparency laws require that buildings annually publish their energy use and energy efficiency, and in most cases their water use. These measures have precedent: governors in Massachusetts, Maryland, New Jersey, Pennsylvania, and Virginia have signed bills or executive orders requiring energy disclosure, and the bipartisan Data Center Transparency Act was introduced in the Senate in 2026.
This week, IMT published its second Model Benchmarking and Transparency Law, which features greater specificity regarding compliance by data centers. We’ll provide a deeper dive into the topic of benchmarking and data centers in the coming weeks.
The third is to leverage broader data center policy frameworks to inform decision-making, especially around metrics. Technical resources include the New Building Institute’s Data Center Energy Performance Evaluation Framework, which provides a structure for understanding data center metrics as well as the relationship between data centers and the grid. PNNL, ASHRAE, and NEMA’s AI Data Center Energy Performance Framework addresses the design, construction and operation of AI data centers, including guidance on retrofits for existing data centers to improve sustainability.
From a legislative and regulatory lens, Climate Mayors, the Sabin Center for Climate Change Law at Columbia Law School, and Climate Xchange have created a model local law for responsible data center development that includes language for siting and permitting, operational standards, transparency, and community benefits, and a model state law focused on aspects of grid-level energy planning. Additionally, the Regulatory Assistance Project’s Data Centers and the Next Era of Energy Regulation series offers regulators resources to prompt decision-making as it relates to large-load energy growth.
In an era of volatile energy prices, concern about environmental damage, and public pushback against data centers, it is not only possible to design and operate data centers for efficiency; it is essential. Building performance policies offer a path forward that meets the urgency of the moment and provides a structure for successful coexistence.