The Union of Concerned Scientists argues that building a reliable and affordable grid requires standardized, transparent operational data sharing from data centers.
Modeling our electricity system is essential for planning our energy future, but accurate models require detailed information about electricity demand. Specifically, understanding the volume of electricity consumed by households, industry, and data centers, along with the precise location and timing of that demand, is critical. Unfortunately, there is a severe shortage of data regarding the number, size, location, and energy consumption patterns of data centers. This opacity is worsened by the confidential and speculative nature of many development proposals, with some operators concealing planned site locations for “security reasons and/or for competitive advantage.”
Industry-wide transparency remains poor. Tech giants like Google, Amazon, and Microsoft frequently utilize subsidiaries to construct facilities, “masking their presence and revealing little about the resources that the facilities consume.” Currently, no central database tracks the total number of data centers in operation, under construction, or in the planning phases. Furthermore, there is little accountability for developers and utilities that may view this rapid expansion primarily as a profit opportunity.
In our recent analysis, UCS relied on electricity demand projections developed by Evolved Energy Research, which draws from Baxtel’s database of existing facilities. We supplemented this baseline with news reports on announced projects and utility filings sourced from both public records and private databases. While commercial platforms like S&P Global and BloombergNEF track individual projects, these proprietary—and costly—databases remain incomplete.
We further estimated which proposed facilities would actually break ground by cross-referencing data from Independent System Operators (ISOs) and purchasing additional datasets from S&P Global. The process was neither straightforward nor inexpensive. Despite the effort, our resulting electricity demand projections aligned with other bottom-up estimates from national laboratories and industry studies. However, because the data center market evolves so rapidly, our projections quickly became obsolete, highlighting the urgent need for consistent, transparent reporting from developers.
Without precise data on facility locations and operational profiles, utility planners, grid operators, government agencies at all levels, and researchers are forced to rely on approximations. These guesses risk flawed conclusions and could saddle ratepayers and utilities with unnecessary, expensive fossil fuel infrastructure—a particularly poor outcome when planning investments that span three decades or longer.
Inaccurate demand data directly undermines infrastructure planning. Overbuilding drives up short-term costs and increases heat-trapping emissions, while underbuilding compromises grid reliability. In the long term, both scenarios result in significantly higher expenses once corrective measures become unavoidable.
Developers are accelerating construction timelines and pressuring utilities to accommodate surging power demands. This dynamic poses significant challenges across multiple regions. In states UCS examined—including Illinois, Michigan, Wisconsin, Louisiana, and California—utility planning frameworks face additional hurdles. These include the absence of state-level Integrated Resource Planning (IRP) processes, interstate electricity trading that may increase reliance on fossil generation, and regulatory environments that fast-track data center infrastructure with minimal oversight.
This opacity and lack of accountability exacerbate planning failures, enabling utilities to overbuild infrastructure, delay coal plant retirements, and make unnecessary investments in natural gas capacity. These missteps burden consumers financially, increase pollution, and harm community health. Residents are frequently denied basic information regarding a facility’s energy consumption, water usage, and emissions, or are actively misled about its environmental impact. In some cases, proposed sites are located in communities already burdened by high pollution levels from existing energy generation, threatening to worsen those conditions. Meanwhile, the prospect of billion-dollar investments often pressures local officials to downplay legitimate risks due to inadequate planning frameworks.
Current policies at both the state and federal levels risk locking the nation into a fossil fuel-dependent energy future. Many governments view data centers as catalysts for local economic growth, job creation, and increased tax revenue, leading them to offer substantial tax breaks and financial incentives. Consequently, states with the most generous incentives have experienced explosive data center expansion.
Recognizing these risks, several states are beginning to pause or reconsider their data center policies. Texas, facing significant projected revenue losses, announced a halt on approving new data centers to allow the Public Utility Commission (PUC) and ERCOT to audit pending projects. However, the governor permitted Meta’s $10 billion, 1-gigawatt facility—which will power approximately 750,000 average U.S. households—to proceed. Minnesota recently eliminated its data center electricity exemption from the state sales tax, effectively rolling back key tax incentives. Wisconsin rescinded certain financing incentives for two municipalities, and New York became the first state to implement a temporary moratorium (through July 2027) on new data centers of 50 megawatts or larger while it evaluates potential impacts.
It is encouraging that these states are slowing development to address community concerns, but more action is required to guarantee a clean, affordable, and reliable grid for the impending surge in electricity demand. UCS analysis demonstrates that the United States can meet rising data center demand primarily through clean energy while simultaneously retiring fossil fuel capacity. Achieving this requires robust policies to accelerate the clean energy transition, alongside updated best practices and regulatory reforms at the federal, state, and regional levels to ensure accountability and strategic planning.
Policymakers at all levels must mandate transparency in contract negotiations between developers and utilities, including the elimination of non-disclosure agreements. Power purchase agreements and grid interconnection approvals should undergo public proceedings with sufficient advance notice to allow communities to participate meaningfully. Data centers must publicly disclose and be held accountable for their electricity demand, water consumption, land use, direct and induced emissions, and related metrics. They must also face strict accountability for any adverse impacts on grid reliability, environmental quality, or public health.
At the national level, efforts are underway to improve reporting standards. The North American Electric Reliability Corporation (NERC) is working to classify data centers as registered entities and is currently drafting reliability standards specifically tailored to the sector.
NERC has a critical opportunity to establish rigorous standards that safeguard both the electrical grid and the communities that depend on it daily. Submit public comments here in support of strong reliability standards for data centers.
The draft NERC standards aim to establish modeling data requirements, verification protocols, and reporting procedures for facility owners. This ensures grid operators and utility planners possess the necessary site-specific information to “plan, monitor, and assess the reliable operation” of the power system and maintain “adequate data available” for reliability and resource adequacy. While these standards represent a positive initial step, the submitted data remains confidential to NERC and focuses exclusively on grid reliability. Concurrently, the Energy Information Administration (EIA) is developing methods to collect broader energy usage data from data centers. At the state and local levels, initiatives are emerging to ban lawmakers from signing NDAs and to prevent utilities from approving data center interconnections through closed *ex parte* processes, as previously occurred in Michigan. These measures should be widely adopted and institutionalized as standard practice for all data center proposals.
States need to require utilities to create long-term integrated resource plans.