NPPI Methodology

Why this exists

Every hyperscaler, utility, and developer negotiating nuclear Power Purchase Agreements in 2026 is pricing deals against zero public comparables. Bespoke contracts, heavy NDAs, no clearing price. Wind and solar have LevelTen's quarterly PPA index driving billions in benchmarked transactions. Nuclear has nothing.

The Nuclear PPA Price Index (NPPI) fills that gap. It synthesizes public PPA filings, disclosed deal terms, operator SEC filings, regulatory data, and market-reported pricing into a standardized quarterly benchmark.

What we publish

Free tier (public)

• Quarterly headline $/MWh for US nuclear PPAs by contract type
• Year-over-year trend
• Top 8 market-moving deals with disclosed terms
• 4 open-source PPA contract templates (FTM, BTM, SMR, restart)

Detail tier ($35,000/year)

• Underlying component prices (capacity, energy, RECs, CFDs, escalators)
• Region-by-region breakdown (PJM, ERCOT, MISO, CAISO, NYISO, SPP)
• Contract-type decomposition: existing-fleet PPAs vs SMR offtake vs restart projects
• 5-year price curve projection with methodology
• Live deal tracker (every disclosed or NDA-cleared deal within 72 hours)
• Quarterly analyst commentary
• API access (REST, 10K calls/mo)
• Excel / PowerBI data packs

Index composition

Contract types tracked

1. Existing fleet PPAs — operating reactors (Susquehanna, Clinton, Perry, Davis-Besse, Beaver Valley, etc.)
2. Restart projects — Three Mile Island 1, Duane Arnold, Palisades, Indian Point (if commenced)
3. SMR offtake — NuScale, TerraPower, X-energy, Radiant, Oklo, Kairos (pre-operational contracts)
4. Microreactor — BWXT, Westinghouse eVinci, Radiant <20MW
5. New large build — Vogtle-style AP1000 or equivalent

Pricing components

Each NPPI publication decomposes $/MWh into:

• Capacity — $/kW-month of firm capacity
• Energy — $/MWh delivered
• REC value — Clean energy credit / ZEC pricing
• Inflation escalator — CPI-linked or fixed step
• Term premium — 10-year vs 20-year vs 30-year comparison

Geographic granularity

• ISO-level: PJM, ERCOT, MISO, CAISO, NYISO, SPP, ISO-NE
• State-level where deal volume supports (TX, PA, NY, VA, IL, SC, GA, AZ)

Data sources

Public filings

• FERC Form 1 (utility financial filings with PPA schedules)
• SEC 10-K/10-Q filings of IOUs with generation exposure
• State PUC filings and CPCN filings
• DOE Loan Programs Office award disclosures
• EIA-860 (generator inventory) and EIA-923 (generation reporting)
• NRC licensing dockets (Construction Permits, COLs, restart petitions)

Market data

• Market pricing from primary public sources (EIA, FERC filings) and the Reactor N deal ledger
• ICE Global Markets power forwards
• PJM Power Market Reports
• ERCOT Market Information System

Deal disclosures

• Corporate press releases from hyperscalers (AWS, MSFT, GOOG, META, AAPL)
• Utility investor day materials
• SMR developer investor communications
• EPRI publications

Proprietary Reactor N sources

• Anonymous operator surveys (quarterly, 40+ respondents target)
• Anonymous buyer surveys (hyperscalers + industrials, 15+ target)
• Law firm network disclosures (aggregated, anonymized)

Calculation method

Weighted average $/MWh

Each disclosed PPA weighted by:

• MW capacity (firm or expected)
• Contract tenor (weight scales with years)
• Publication clarity (public filing = 1.0, press release = 0.7, inferred = 0.4)

Outlier handling

• Prices outside 2σ are flagged and shown separately (not in main index)
• Paid-up capacity contracts (prepaid) normalized to $/MWh equivalent using 8% discount rate
• CPI-linked contracts shown at "current $" and "levelized 2026$" separately

Quality score per reported deal

Each included deal ranked 1–5 on:

• Source transparency (public filing vs anonymous)
• Pricing breakdown (full decomposition vs headline only)
• Contract execution stage (signed vs LOI vs binding term sheet)

Only quality ≥3 deals included in main index. Lower-quality deals go into the "Watch List" annex.

How the headline is built — every input, shown

The headline $/MWh is the MW-weighted average of the priced deals in the ledger. We do not run a black box: 0% of the NPPI is disclosed pricing — every $/MWh below is back-calculated from public filings and press releases, and we show the exact arithmetic and the source for each one. Anyone can reproduce the headline from this table.

Priced deals in the headline

Verification status is published per deal: verified = exact terms in a public filing or release (price may still be back-calculated from a disclosed total); estimated = real announced deal, price analyst-inferred; unverified = announced/rumored, not yet in filings — excluded from the headline. Unpriced deals (most SMR/restart announcements with no public strike) are tracked but do not enter the $/MWh average.

Headline robustness — no single row can move it

A weighted average can be fragile if a few large deals dominate the weight. We disclose that risk and engineer against it. The headline is published with a confidence band and is cross-checked under two additional lenses, computed live at /api/v1/nppi:

• Inclusive — the naive MW-weighted average of every priced deal (this is the published headline).
• Concentration-capped — the same average, but no single counterparty and no single calendar-date cluster may exceed 20% of total weight (iteratively clamped). This removes the leverage any one deal has on the number.
• Verified-only — the MW-weighted average using only verified deals, dropping every estimated or unverified row entirely.

If all three lenses bracket the published number, the headline is robust by construction. Today they do:

Contributor panel — the disclosed share, shown honestly

The decisive move for any price benchmark is the share of the index backed by directly disclosed strikes rather than inference. We publish that share honestly — including when it is zero.

As of today the NPPI is 0% directly disclosed — every number is back-calculated from public filings. The contribution panel is opening now. A confidential, real strike from someone who was in the room is the one input a competitor working only from public filings cannot reproduce — and the only thing that moves this share off 0%. Contribute confidentially → (only the blind aggregate is ever published; your individual number is never surfaced).

Our methodology vs. closed price-reporting

The established model for commodity price benchmarks is a closed one: prices are assessed behind a paywall, the underlying deals and adjustments are not shown, and you are asked to trust the brand. That works when the assessor has decades of submissions. A new benchmark in an opaque market cannot win on trust-us — so we do the opposite.

The thesis is simple: in a market with no clearing price, the benchmark whose every input is inspectable earns trust faster than the one that asks for it. That is the entire wedge. (We do not name or rank specific providers; this is a description of the closed vs. open model, not a comparison to any one firm.)

Publication schedule

Between releases: NPPI Weekly newsletter for detail-tier subscribers. Every material deal within 72 hours of disclosure.

Transparency principles

1. Methodology is public. This document. Updated quarterly.
2. All data sources are cited. Subscribers see raw source URLs in the detail tier.
3. Corrections are published. If we get a number wrong, next quarter's release lists the correction with reasoning.
4. No operator/buyer preference. We don't favor particular developers.
5. No undisclosed sponsorship. Index is subscriber-funded. Zero vendor payments influencing numbers.

Enterprise tier pricing

B2C → Public + Researcher  ·  B2B → Pro + Fleet + Strategic + Custom

First-year early-access discount: 25% off Pro and Fleet tiers for Q2–Q3 2026 signups (coupon NPPI_EARLY_2026).

Contact: intelligence@reactorn.com

Published by: OpenAgent / Reactor N Intelligence