Behind every credible energy-transition claim sits a capacity-expansion model: software that simulates decades of builds, retirements, and dispatch to say what the grid should become and what it will cost. Utilities file rate cases on it, developers size pipelines against it, and regulators argue inside it. The market pairs a dominant commercial platform family with a serious open-source insurgency; this guide compares both on identical criteria.
North American forecasting staple: Aurora (Energy Exemplar) — nodal price forecasting and IRP work.
Open-source framework: PyPSA — the research-to-TSO workhorse, free.
Academic expansion rigor: GenX — MIT-lineage capacity expansion, open.
Free US policy engine: NREL ReEDS — the national-lab scenario machine.
The data layer: simulation-ready datasets — where modeling budgets actually go.
Scope: power-system production-cost, price-forecasting, and capacity-expansion modeling — the long-horizon layer above our ETRM (trading operations) and PPA (price benchmarks) guides. Who uses it: utilities and IRP teams, developers and investors stress-testing revenue theses, regulators, and researchers. Six entries, identical criteria; order follows market position, not rank.
Criteria: modeling scope (dispatch, expansion, nodal detail), geographic coverage, data ecosystem, transparency and auditability, licensing, and the main tradeoff. Commercial licensing is quoted; half the category is genuinely open — labeled (checked October 7, 2026).
At a Glance
| Platform | Pricing | Best For | Link |
|---|---|---|---|
| PLEXOS (Energy Exemplar) | Enterprise licenses/cloud (quote-based) | Unified market simulation & expansion | energyexemplar.com → |
| Aurora (Energy Exemplar) | Licenses (quote-based) | Price forecasting & IRP, NA-strong | energyexemplar.com/aurora → |
| PyPSA | Free, open source | Open framework, research to TSOs | pypsa.org → |
| GenX | Free, open source | Capacity-expansion research rigor | github.com/GenXProject → |
| NREL ReEDS | Free (US focus) | Policy-grade US scenarios | nrel.gov/analysis/reeds → |
| Simulation-ready datasets | Subscriptions (quote-based) & open sources | The input layer that decides quality | power datasets → |
Pricing checked October 7, 2026. Most platforms in this category sell quote-based enterprise plans; where we cite figures they come from vendor pages or published third-party comparisons and are order-of-magnitude indications, not offers. Billing basis (per user, per MW, per site) varies by vendor — confirm current terms directly before budgeting.
The Platforms in Detail
PLEXOS
The platform of record
Best for: organizations whose modeling must span markets, horizons, and technologies in one defensible environment.
| Modeling scope | Unified production cost, capacity expansion, and co-optimized gas/power/hydrogen modeling; cloud execution at scale |
| Geographic coverage | Global — market libraries across every major power system |
| Data ecosystem | Vendor-maintained, simulation-ready market datasets sold alongside |
| Transparency | Commercial solver stack; auditable inputs, proprietary internals |
| Licensing | Enterprise licenses and PLEXOS Cloud; quote-based. Checked October 7, 2026 |
| Main tradeoff | Platform gravity — power, price, and lock-in grow together |
- The default answer to “what did the IRP run on?” in a growing share of jurisdictions — regulatory familiarity compounds like interest.
- Cross-vector co-optimization (power with gas and hydrogen) matches where 2026 planning questions actually live.
- Cloud execution converted week-long stochastic runs into same-day iterations — scenario count, not solver time, now limits analysis.
Aurora
The forecasting staple
Best for: North American utilities, developers, and consultants whose deliverable is a defensible price curve or IRP.
| Modeling scope | Nodal and zonal price forecasting, portfolio and IRP analysis |
| Geographic coverage | North America’s long-time staple; now part of Energy Exemplar’s family (Aurora 15 era, PLEXOS Cloud execution) |
| Data ecosystem | Shares the vendor’s simulation-ready dataset pipeline |
| Transparency | Commercial; widely understood by counterparties and commissions |
| Licensing | Licenses; quote-based. Checked October 7, 2026 |
| Main tradeoff | Family consolidation — the PLEXOS/Aurora boundary is now a product-line question |
- Decades of US IRP and transaction work made its curves a lingua franca — counterparties recognize the methodology before the meeting starts.
- Under Energy Exemplar the historic PLEXOS-versus-Aurora choice became an internal roadmap — buyers should ask where their use case lands in it.
- For merchant-revenue diligence it pairs naturally with our PPA and insurance guides — the curve is the input those layers price.
PyPSA
The open framework
Best for: teams that want transparent, scriptable system modeling — from papers to real TSO studies — without license gates.
| Modeling scope | Optimal power flow, dispatch, and expansion in one Python framework; PyPSA-Eur and sector-coupled variants |
| Geographic coverage | Global by construction; strongest open-data pipelines in Europe |
| Data ecosystem | Open data workflows — reproducible from raw sources |
| Transparency | Fully open — every equation and assumption inspectable |
| Licensing | Free, open source. Checked October 7, 2026 |
| Main tradeoff | Framework, not product — engineering and data discipline are yours |
- Transparency is regulatory strategy: a model anyone can rerun ends “black box” objections before they start.
- The research-to-practice pipeline is real — methods debut in PyPSA papers and surface in commercial roadmaps later.
- For developers, a PyPSA-literate analyst is a free second opinion on every vendor curve the business buys.
GenX
The expansion specialist
Best for: modelers who want state-of-the-art capacity-expansion formulations with academic rigor — open.
| Modeling scope | Capacity expansion with advanced operational detail — unit commitment approximations, storage, policies |
| Geographic coverage | Anywhere you bring data; US research heritage (MIT/Princeton lineage) |
| Data ecosystem | Bring-your-own, with published example systems |
| Transparency | Open formulation — the methods papers are the documentation |
| Licensing | Free, open source. Checked October 7, 2026 |
| Main tradeoff | A researcher’s instrument — production workflows need wrapping |
- Formulation quality is the pitch: policy constraints, storage dynamics, and decarbonization pathways modeled the way the literature says they should be.
- The academic provenance cuts both ways commercially — unimpeachable methods, minimal hand-holding; consultancies increasingly bridge the gap.
- Ideal as the rigor check on commercial expansion results: when GenX and the platform disagree on build-out, the assumptions are hiding something.
NREL ReEDS
The policy engine
Best for: anyone arguing about the US grid’s future — because the national scenarios already run on it.
| Modeling scope | US capacity expansion with detailed renewable resource and transmission representation |
| Geographic coverage | United States, in national-lab depth |
| Data ecosystem | NREL’s own resource and cost databases (ATB) — the public assumptions everyone cites |
| Transparency | Open model, published scenarios (Standard Scenarios et al.) |
| Licensing | Free. Checked October 7, 2026 |
| Main tradeoff | US-specific and policy-oriented — not a commercial nodal forecaster |
- The Standard Scenarios are the shared baseline of US energy debate — knowing them is table stakes for policy-exposed investment theses.
- ATB cost assumptions flow through this model into everyone’s spreadsheets — tracing a claim to its ReEDS/ATB roots is basic diligence.
- Free and public means your strategy team can rerun the future under their own assumptions — few arguments are settled faster.
The data layer
Where budgets actually go
Best for: every modeler above — because in this category the dataset, not the solver, decides the answer’s quality.
| Modeling scope | Simulation-ready market datasets (vendor-maintained) and open equivalents (ISO postings, open-data pipelines) |
| Geographic coverage | Vendor libraries span major markets; open coverage varies |
| Data ecosystem | The product itself — maintained network, unit, and policy representations |
| Transparency | Vendor datasets documented but licensed; open pipelines fully inspectable |
| Licensing | Dataset subscriptions quote-based; open sources free. Checked October 7, 2026 |
| Main tradeoff | Convenience versus auditability — and the subscription often outcosts the software |
- Two teams with one platform and different datasets will defend different futures — dataset provenance belongs in every model review.
- Vendor-maintained market libraries are genuine labor savings measured in analyst-years; price them against your team honestly.
- The open counterweight (ISO data, PyPSA-Eur pipelines, ATB) keeps assumption-shopping detectable — our series’ recurring theme, one layer deeper.
One Vendor’s Gravity, and the Open Counterweight
Energy Exemplar’s consolidation of PLEXOS and Aurora concentrated the commercial category into a single family precisely as demand exploded — IRPs, data-center load shocks, and hydrogen coupling all route through these models now. Concentration buys coherence (one platform, shared datasets, cloud scale) and carries the familiar risks: pricing power and methodological monoculture. The open stack — PyPSA, GenX, ReEDS — is the counterweight, and uniquely in enterprise software, regulators actively reward its transparency.
The mature pattern emerging in sophisticated shops is bilingual: commercial platforms where regulatory familiarity and dataset maintenance pay, open frameworks where transparency, custom questions, or budget demand it — with cross-model verification as standard practice. The worst position is monolingual faith either way; this category’s answers move markets, and single-model answers are assumptions wearing a solver.
Reading Model Output Like a Buyer
Every curve and build-out plan in this category is a conditional statement: given these gas prices, these load forecasts, these policy assumptions, this future follows. Diligence therefore starts at the inputs — whose dataset, which vintage, what sensitivity structure — before admiring the output. Our PPA guide’s benchmark discipline and our insurance guide’s trigger interrogation are the same habit applied downstream of this layer.
For renewable investors specifically, three model artifacts deserve standing skepticism: terminal-year price dynamics (where merchant tails live), curtailment and congestion treatment (where nodal detail is expensive and often simplified), and storage revenue representation (where formulations differ most, as our BESS guide’s benchmarking culture attests). Ask how each is handled, in writing, before a curve enters your capital model.
Kurums Match: Which One Fits You?
Pick the statement that sounds most like your situation.
We’re a utility or IRP team facing regulatory filing.
Commercial-platform familiarity (PLEXOS/Aurora-class) is worth real money in commission proceedings — but pair filings with open-model cross-checks (ReEDS framing, PyPSA/GenX verification) to preempt black-box challenges.
We’re a developer or investor buying price curves.
Buy the assumptions, not the brand: demand dataset provenance, sensitivity suites, and the three artifacts above. An in-house PyPSA-literate analyst pays for themselves on the first curve they catch.
We’re modeling policy or public-interest scenarios.
The open stack is built for you — ReEDS for US scenario authority, PyPSA for custom and sector-coupled questions, GenX for expansion rigor — and its transparency is your credibility.
We’re building modeling capability from zero.
Start open (PyPSA plus public data) to learn what questions you actually ask, then buy commercial seats and datasets only for the workflows regulators or counterparties demand. The reverse order buys shelf-ware.
Frequently Asked Questions
How does this differ from your ETRM and PPA guides?
Horizon and question: ETRM manages today’s positions, PPA platforms benchmark today’s contracts, and this layer simulates decades of system evolution. The curve this layer produces is the assumption those layers — and your capital model — consume.
Are open models really used for real decisions?
Increasingly, yes: PyPSA-family models run inside European TSO and government studies, ReEDS anchors US national scenarios, and GenX-grade formulations inform real IRP debates. Commercial platforms still dominate regulated filings — familiarity is a currency — but open is no longer academic-only.
What drives cost in commercial deployments?
Three stacks: licenses/cloud compute, the dataset subscriptions (often the largest line over time), and the analysts. Vendors quote all of it; negotiate on total cost per defensible scenario, and remember the free stack prices your fallback position.
Which single model is most accurate?
Unanswerable as asked — accuracy here is dominated by input assumptions and scenario design, not solver choice, and every published backcast confirms it. The buyable version of accuracy is process: documented data provenance, sensitivity breadth, and cross-model verification.
Related Comparisons & Guides
- Energy trading & risk (ETRM) software compared
- PPA price benchmark platforms compared
- Grid interconnection software compared
- Battery storage optimization software compared
- US renewable energy strategy explained
Last updated: October 7, 2026 · Reviewed by the Kurums Startup editorial team.
Disclosure: Kurums currently has no affiliate, sponsorship, or partnership relationship with any product compared on this page. If that changes, this page will say so here and affected links will carry sponsored attributes.
Part of the Kurums Renewable Energy hub — country strategies, permitting, incentives, financing, and tools across nine markets.
Discover more from Kurums | Business Intelligence
Subscribe to get the latest posts sent to your email.