TRL methodology
A number is only useful if you can see what it was built from
A technology readiness level (TRL) is a 1 to 9 scale the aerospace and energy sectors have used for forty years to describe the maturity of a technology. We assess it from primary documents, publish a confidence band alongside the scoring, and cite all references to establish clarity behind the decision-making process. A subject matter expert will also verify the results.
The ladder
What each level means, and what it permits
TRL is more than a descriptive level. It represents the maturity of a technology as determined through a structured readiness assessment using predefined criteria and documented evidence. Technologies progress by demonstrating increasing levels of validation, from laboratory proof-of-concept through prototype demonstration to proven commercial operation.
For investment purposes, each TRL band defines both the financing instruments that may be offered and the proportion of capital that can be released. Earlier-stage technologies are financed through SAFEs and convertible notes, while technologies that have demonstrated operational readiness can support fixed-income securities and revenue-backed investments.
The assessment is performed at the level of a key enabling technology. This is because the overall readiness of a system is constrained by its least mature component; funding milestones are linked to independently verified evidence rather than projected development schedules.
| Band | Stage | Instrument permitted | Release | Verified by |
|---|---|---|---|---|
| TRL 1–3Tier I | Concept formulated; lab proof of principle | SAFE or Convertible note | 15% | Peer-reviewed publication; DOE award letter |
| TRL 4–5Tier I | Component validated in a relevant environment | Convertible note | 30% | Independent test report; national lab result |
| TRL 6Tier I | Prototype demonstrated in a relevant environment | Preferred equity | 50% | Third-party witness test; NRC pre-application docket |
| TRL 7Tier II | System prototype in an operational environment | Senior note | 70% | NRC construction permit; supplier contracts executed |
| TRL 8Tier II | System complete and qualified; first unit built | Fixed-rate note or Municipal mini-bond | 90% | ITAAC closure; operating license issued |
| TRL 9Tier II | Proven in commercial operation | Revenue share or Fixed-rate note or Municipal mini-bond | 100% | Commercial operation date certificate; metered output |
Evidence taxonomy
What counts as evidence, and how much
Sources are not equal. A regulator's determination and a company's own slide deck are both documents; only one of them moves a score. The ceiling is the column that gets argued with: some documents cannot support a high level however good they are.
- Operating license
weight 0.95 · supports to level 8
The strongest single class of evidence. A regulator has made a determination that cannot be re-argued by the issuer.
- ITAAC closure
weight 0.90 · supports to level 8
Acceptance criteria closure is a fact about a built object, not a projection about a design.
- Metered output
weight 1.00 · supports to level 9
Measured production. Weighted highest at level 9 and unavailable below level 8 by definition.
- NRC docket
weight 0.70 · supports to level 7
Filings and pre-application engagement. Being docketed is evidence of seriousness, not of adequacy.
- Independent test report
weight 0.75 · supports to level 6
Weighted materially higher when a third party witnessed the test than when the issuer reports its own.
- National lab result
weight 0.80 · supports to level 6
Independent facility, published protocol. Strong for component validation, weak for integrated systems.
- Supplier contract
weight 0.45 · supports to level 8
Executed and firm-priced contracts count. Letters of intent and term sheets are noted and weighted near zero.
- Peer-reviewed publication
weight 0.40 · supports to level 3
Establishes proof of principle. Rarely moves a score above level 5 on its own.
- DOE award
weight 0.35 · supports to level 4
Evidence of independent technical review and of cost-share reducing capital at risk. Milestone-gated awards are not guarantees of later tranches.
- Patent
weight 0.20 · supports to level 3
Evidence of novelty and of freedom to operate. Almost no evidence of TRL.
A review round moves a score by one level, or two where independent sources agree. That cap exists because a single document which would move a score three levels is almost always a document that has been over-read, and because an issuer who could jump from concept to commercial on one submission would have every incentive to try.
Confidence
Why a band and not a point
A score of 8 with a band of 7.6 to 8.3 is a different claim from an 8 with a band of 6.2 to 9.0. Publishing the first number without the second would misrepresent both.
The band widens when evidence is indirect, when a result is extrapolated rather than measured, when the most recent document is old, or when independent sources disagree. It narrows when a regulator has made a determination, when a test was witnessed by a third party, and when several independent sources agree.
You can filter offerings by band width. Demanding precision is a legitimate investment criterion, and an investor who will only accept tightly-evidenced claims should be able to express that.
What it does not measure
A high TRL is not a promise of a return
The scale describes how far a technology has been demonstrated. It says nothing about what the output sells for, what the plant costs to finish, or whether anyone has agreed to buy.
A station running since 1988 scores 9 because there is nothing left to demonstrate, and still carries component aging, currency and offtake risk that the scale cannot see. A design at 8 can be unfinanceable on first-of-a-kind construction cost alone. Where that is the live question, the reviewer says so in the assessment narrative on the offering page, in words, rather than inventing a second number to carry it.
Read it the other way too. A concept at 3 may not be a bad investment, it is an early one, and the ladder prices that by limiting what it may sell you. The score decides the instrument. Whether the deal is worth taking is in the offering documents and the risk factors.
Human review
No score is published unsigned
The engine proposes. A subject matter expert consults, and their name and credentials go on the record beside the number.
- 1
Collate
Primary documents are collected: dockets, licenses, test reports, national lab results, patents, executed contracts, published literature.
- 2
Assess
The engine proposes a level, a confidence band, and a weighted citation for every input. Model identifier and methodology version are recorded.
- 3
Review
A qualified reviewer examines the evidence and may adjust the level or the band. Disagreements with the model are recorded, not overwritten.
- 4
Sign and publish
The reviewer signs with name, credential and timestamp. Only then does the score become visible, and the issuer sees it when you do.
Model risk and limitations
An assessment is only as current as its most recent document. Scores are versioned and superseded rather than edited, so a historical score remains inspectable. The engine can misread an ambiguous filing, and the human review step exists precisely because it sometimes does. Where a score has changed, the history is published on the offering page.
Inference cost per assessment is trivial against the cost of the expert hours that review it. The scarce input here is nuclear expertise, not compute.