Every four to eight years, U.S. energy policy seems to lurch. Administrations swap emissions targets, subsidy structures, and permitting priorities almost as soon as they take office. Yet one thing has stayed remarkably constant since the early 1990s: the National Energy Modeling System (NEMS). NEMS is the analytical engine the federal government uses to think through the consequences of it all.
That durability is not an accident. It is a design choice, and it is worth understanding if you are evaluating energy-economy models for your own organization’s planning, risk assessment, or scenario work.
A model built to outlast its builders
NEMS is developed and maintained by the U.S. Energy Information Administration (EIA), the statistical and analytical arm of the Department of Energy. EIA itself was created by the Department of Energy Organization Act of 1977, which gave it something unusual for a federal agency: statutory independence. By law, EIA’s data, analyses, and forecasts do not require approval from any other officer or employee of the U.S. government, including political appointees at DOE. Congress built EIA this way deliberately as a response to the data gaps that had hampered policymaking during the 1970s energy crises.
NEMS itself traces to the early 1990s with the first Annual Energy Outlook (AEO) released in December 1993. The AEO is a publication EIA has produced almost every year since, through Democratic and Republican administrations alike, providing continuity respected across the political spectrum. That credibility extends beyond government: organizations with widely differing policy priorities, including groups such as the Heritage Foundation and the Natural Resources Defense Council, have relied on NEMS-based analysis to inform their work, illustrating the model’s value as a trusted analytical tool rather than an ideological instrument.
That is the structural reason NEMS continues to function as a shared reference point: it is not owned by an administration’s policy agenda. It is owned by a statutory mandate to produce objective, policy-neutral projections, which EIA describes as its mission.
Evolution, not argument
NEMS was never built to litigate climate science. It is an energy-economy simulation: an evolving modeling system that, sector by sector, solves for how U.S. energy supply, conversion, and consumption will respond to a given set of assumptions (e.g., macroeconomic conditions, technology costs, resource availability, and policy constraints) through 2050.
That is precisely why it has proven so durable across a political environment where the underlying science (the attribution of recent warming to human greenhouse gas emissions, as documented by NOAA, NASA, and the IPCC) is broadly settled among scientists, even as the policy response remains contested. NEMS does not ask users to take a side in that policy debate. It asks a narrower, more useful question: given this set of policy or technology assumptions, what happens to prices, production, emissions, and consumption?
The technology stack: what “evolution” looks like inside the model
Part of what makes NEMS useful across administrations is that it does not model a single decarbonization pathway. Instead, it models the menu of technologies that any administration or researcher might want to weigh against one another. EIA’s own technical presentations describe a wide catalog of options represented across the model’s modules.
Examples of how evolving technology is incorporated into NEMS include representing natural gas and coal generation, both with and without carbon capture and storage (CCS); modeling wind, solar, geothermal, hydropower, landfill gas, and biomass generation; and tracking industry technology changes. Starting with AEO 2025, EIA also incorporated an entirely new Hydrogen Market Module (HMM) to connect hydrogen production to end users and seasonal storage.
This is what “evolution through technology” means in practice for NEMS: not a fixed bet on any single fuel or fix, but a running inventory of the options a decision-maker or researcher actually has, priced out consistently against one another, year after year.
Evolution in real time: AEO 2025 and AEO 2026
The clearest evidence that NEMS evolves rather than argues is what happened over the past two release cycles, spanning a change in administration.
EIA skipped publishing an AEO in 2024 specifically to rebuild parts of NEMS. They added the Hydrogen Market Module, upgraded the model’s representation of carbon capture and oil and gas resources, and rewrote much of the model’s connecting code from Fortran and shell scripts to modern, open-source Python. AEO 2025, released April 15, 2025, was the product of that overhaul. It reflected federal laws and regulations in effect as of roughly December 2024, including the Inflation Reduction Act’s clean-energy tax credits.
AEO 2026, released on April 8, 2026, also shows how quickly the same architecture can adapt to a change in political direction, allowing it to address a wholly different set of policy questions. EIA renamed its central projection from the “Reference case” to the “Counterfactual Baseline case” and organized the release around 11 total scenarios: the baseline plus 10 side cases, including Alternative Transportation, Alternative Electricity, a combination case, and high/low variants for economic growth, oil and gas supply, and zero-carbon technology cost. The Counterfactual Baseline reflects laws and regulations in place as of December 2025, including the early expiration of clean-vehicle and EV-charging tax credits under the 2025 reconciliation law known as the One Big Beautiful Bill Act. Because the EPA proposed repealing its April 2024 update to Clean Air Act Section 111 power-plant carbon rules in June 2025, EIA built a dedicated Alternative Electricity case to model outcomes without that regulation, alongside the baseline case that still includes it. This allowed users to see both regulatory futures side by side rather than betting the entire model on a single outcome.
EIA Administrator Tristan Abbey summarized the release by describing the AEO as best understood as “… a product suite for exploring alternative futures, not a set of predictions.” This framing captures exactly why the model survives elections that its forecasts sometimes do not anticipate. The same NEMS architecture that modeled IRA tax credits under the prior administration is, one year later, modeling their early sunset under the current administration, without EIA needing to build a new model from scratch each time Congress or the White House changes hands.
A tool built for continuity, used well beyond government
This adaptability is also why NEMS’ user base extends far past EIA’s own walls. EIA’s documentation notes that, in addition to producing the AEO, the model is used to complete analytical studies requested by the White House, Congress, other DOE offices, and federal agencies. It is also used by nongovernmental users including the Electric Power Research Institute, Duke University, and the Georgia Institute of Technology. The full model source code is publicly available in EIA’s public repository, and detailed documentation for every module, from the Electricity Market Module to the new Hydrogen Market Module, is published as a matter of statutory transparency requirements. EIA has said it checks its own work and expects outside users to do the same.
For an organization evaluating long-horizon energy or climate-adjacent risk (e.g., a utility assessing generation mix under different carbon policy scenarios, a manufacturer stress-testing energy costs against CCS or hydrogen adoption, a financial institution modeling transition risk in a portfolio), that combination of features is unusual: a model that is free of partisan capture by design, transparent enough to audit line by line, stocked with the actual technology options under active policy debate, and flexible enough to run the specific scenario that matters to the user, regardless of which administration is currently setting federal targets.
Why this matters for your next analysis
The politics of energy policy will keep swinging with each presidential cycle. That is simply how U.S. governance works, and NEMS was never meant to resolve that. What it offers instead is a consistent, transparent, and technically rigorous way to ask “what if.” What if a carbon rule survives a repeal attempt, what if hydrogen tax credits expire on schedule, what if battery storage costs fall faster than expected? NEMS ensures the answer is not dependent upon who wins the next election. The last two release cycles show the model doing exactly that: absorbing a change in administration, a change in law, and a change in regulatory posture, all within the same modular framework it has used since 1994.
That is the real value proposition for energy or climate analysis: a modeling framework built to survive changing administrations, so your organization can focus on the long-term issues at hand.
Sources
Andri Rizhakov, EIA, Decarbonization Technologies in the National Energy Modeling System (NEMS) (April 2021). www.eia.gov/environment/workshop/decarbonization
Congressional Research Service, The U.S. Energy Information Administration (R46524, 2020). www.congress.gov/crs-product/R46524
EIA, National Energy Modeling System source code repository. www.github.com/EIAgov/NEMS
Energy Policy Act of 1992, Public Law 102-486. www.congress.gov/bill/102nd-congress/house-bill/776
“Executive Summary.” National Research Council. 1992. The National Energy Modeling System. Washington, DC: The National Academies Press. doi: 10.17226/1997. https://www.nationalacademies.org/read/1997/chapter/2?page_hash=1.
U.S. Department of Energy, Office of Policy National Energy Modeling System (OP-NEMS). www.energy.gov/policy/office-policy-national-energy-modeling-system-op-nems
U.S. Energy Information Administration, About EIA. Legislative Timeline. www.eia.gov/about/legislative_timeline.php
U.S. Energy Information Administration, Annual Energy Outlook 2026 Narrative. www.eia.gov/outlooks/aeo/narrative
U.S. Energy Information Administration, Assumptions to the Annual Energy Outlook 2017, Introduction. www.eia.gov/forecasts/aeo/assumptions
U.S. Energy Information Administration, Hydrogen Market Module. NEMS Documentation (August 2025). www.eia.gov/analysis/pdfpages/hmm.php
U.S. Energy Information Administration, Integrating Module — NEMS Documentation (July 2025). www.eia.gov/analysis/pdfpages/m057index.php
U.S. Energy Information Administration, Introduction to the Hydrogen Market Module, AEO2025 Working Group. www.eia.gov/outlooks/aeo/workinggroup/hydrogen
U.S. Energy Information Administration, Levelized Costs of New Generation Resources in the Annual Energy Outlook 2026. www.eia.gov/outlooks/aeo/electricity_generation
U.S. Energy Information Administration, NEMS Documentation. www.eia.gov/outlooks/aeo/nems/documentation
U.S. Energy Information Administration, Press Release, EIA releases the Annual Energy Outlook 2026 (April 8, 2026). www.eia.gov/pressroom/releases/press587.php
U.S. Energy Information Administration, Summary of Legislation and Regulations in the Annual Energy Outlook 2026 (April 2026). www.eia.gov/outlooks/aeo/assumptions/pdf/Legs_Regs_2026.pdf