Insurance prices distributions. Banks price points.
Banks underwrite a single borrower against equipment residual and personal guarantees. Insurers must believe a population of borrowers behaves like the population already in their book. That is why insurance is the hardest tier to unlock — and the most valuable. With 100+ operators in the cohort and 24 months of attested data, Allometry becomes the actuarial table no carrier has access to today. Munich Re. Swiss Re. Lloyd's. Eventually: Allometry IS the carrier.
Thirteen named carriers. From global reinsurers to parametric MGAs.
Three layers: global reinsurers (Munich Re, Swiss Re, Hannover Re) who provide capacity to everyone else; Lloyd's market syndicates (Beazley, Hiscox, Atrium, Nephila) writing specialty subscription business; and emerging parametric MGAs (Parametrix, Descartes) who measure events independently and pay fast. Lloyd's and Hannover Re backed Parametrix specifically because Parametrix could independently measure cloud outages. The same template applies to field-service operators — independent operator-side measurement is what makes the cohort insurable.
Five product types. One operator data substrate underneath.
Insurance products relevant to asset-heavy operators all share one structural requirement: independent, verifiable measurement of the underlying risk. The carrier (or its reinsurer) needs to know whether the SLA actually breached, whether the asset actually failed, whether the contracted delivery actually happened. Today most of that measurement is operator-reported, which is uninsurable at scale because of adverse selection. With ZKP-attested operator data, the same products become writable at cohort scale — and at materially tighter combined ratios.
Parametric SLA wraps
Payout triggered by measured performance drop (uptime, throughput, response time). Parametrix is the canonical example: 15-business-day payment, no adjuster. The trigger is the data — which is exactly what the vault produces.
Equipment warranty wraps
OEM warranty extensions backed by insurance paper. Blink × EVSTAR's 5-year EV-charger protection plan is the template. The carrier needs failure-rate distributions; the vault produces them per SKU and vintage.
Business interruption
AXA XL, Allianz Commercial cover physical damage + revenue loss from asset failure. Rated against replacement cost, typically 0.3–1.5% of insured value. Loss-cost ratio improves when historical loss data is hash-chained.
Performance bonds
Tokio Marine HCC up to $50M per bond on solar O&M contracts. 1–3% of bonded amount per annum. The vault evidences delivery history — the substrate for premium reduction.
Cyber-OT towers
Coalition (IoT origin), Beazley Cyber Risks, Allianz/Coalition JV. Covers operational disruption from connected-equipment compromise. $1M–$5M retention, $5M–$25M limits. The vault's IoT-sourced data is the underwriting evidence.
"Performance Insurance"
AXA XL Structured Risk Solutions writes guarantee paper on new / renewable tech performance. Emerging hybrid bridging warranty and insurance — the vault is the data layer that lets it scale.
A four-subject sample cannot fit a distribution.
Banks underwrite a point estimate of probability-of-default times loss-given-default. They can fall back on equipment residual, personal guarantees, and AR collateral when the model is uncertain. Insurance has no such fallback. The carrier must fit the entire shape of the loss curve — mean, variance, tail — and price the tail at 99th or 99.5th percentile. That requires a population. Until the population exists, the math doesn't.
The credibility math everyone underweights.
~1,082 claims ~27,000 exposure units 100+ comparable risks
Actuarial Standard of Practice 25 ties full credibility to claim counts. At standard parameters, full credibility requires roughly 1,082 claims. If expected frequency is 4 claims per 100 units, that's ~27,000 exposure units. Partial credibility kicks in much earlier — but a cohort below 100 comparable risks is rarely credible enough for a new program. Banks lose principal; insurers lose multiples of premium when tails fire. The asymmetry justifies the data appetite.
The Parametrix precedent is the model. Hannover Re, Lloyd's syndicates, and Lockton backed Parametrix specifically because Parametrix could independently measure cloud outages with sensor-grade fidelity. Operator self-report would have been uninsurable. The same template applies to field services — independent measurement is what unlocks the cohort.
Phase 1: MGA. Phase 4: own the syndicate.
Routing premium to Munich Re is the obvious play. Becoming the carrier is the better one — because the proprietary cohort data is the underwriting moat, and the underwriting moat is the rating-agency capital adequacy that lets a carrier exist at all. The path is well-trodden: Lemonade, Hippo, Coalition all walked it. The binding constraint is not regulatory capital. It's cohort data + reinsurer willingness to cede capacity. The vault produces both.
MGA on rated paper
Program-business MGA on Markel / AXA / Lloyd's paper. Markel targets $10–15M annual programs — achievable on a few hundred operators at $100–300K each. Allometry handles distribution, underwriting, and claims; the rated carrier holds the policy.
Rented cell / PCC
Vermont, Bermuda SAC, or Guernsey protected-cell company. Setup in weeks. Minimum participation premium ~$500K. Capital often <$250K per cell vs $5M+ standalone captive. Allometry takes 10–30% quota share alongside fronting carrier.
Own captive · fronted
Own captive or sponsored captive (Green Mountain Vermont model). Fronted by AIG or Marsh-managed facility. Reinsured by Munich Re / Hannover Re / Nephila against tail risk. Allometry holds proportionally more of the underwriting profit.
Lloyd's syndicate · scale
Lloyd's syndicate or full carrier license. Bermuda Class 3A typically $5M minimum capital. US admitted carrier $5–10M+ plus RBC. The proprietary cohort data is the asset that unlocks rating-agency capital adequacy.
Adverse selection collapses when the data can't be cherry-picked.
The fundamental problem with insuring field-service operators today is adverse selection: operators with hidden churn, hidden downtime, hidden warranty exclusions self-select into coverage. Carriers respond by loading premium with an "ignorance premium" — pricing the worst-case operator into every quote. Cryptographically attested operator data eliminates adverse selection structurally. Operators can't hide what's in the hash chain.
| Vault evidence (continuously attested) | What it unlocks · insurance economics |
|---|---|
| Real-time loss-event streams vs annual loss-run snapshotsIBNR uncertainty eliminated | Replaces the annual "as-of" loss run that's already stale at submission. Reserve calculations tighten; carrier confidence in reserves rises; combined ratio target tightens. |
| Hash-chained operational metrics uptime, MTBF, MTTR per SKU and vintageAdverse selection priced out | Operators cannot cherry-pick which events to report. Carriers can fit distributions from full population data, not curated submissions. Loss-cost ratio improves systematically. |
| ZKP-attested SLA compliance cohort-level without exposing per-operator dataCohort priced as a cohort, not worst-of | Letting carriers verify proofs without seeing the underlying operator data lets the cohort be priced collectively — the price reflects cohort risk, not the worst observed operator. |
| Equipment manufacturer warranty data OEM warranty terms · recall history · residualsTail risk priced from data, not from theory | Carriers can fit failure-rate distributions per OEM-vintage cohort. PML calculations tighten. Reinsurance attachment points move favorably. |
| Contract terms parsed SLA definitions · liquidated-damages clauses · concentration exposureWording disputes eliminated | Parametric triggers can be set against parsed-contract definitions, not interpreted ones. Basis-risk on parametric products tightens. |
| Workforce data technician certifications · training hours · supervisor ratiosOperational risk underwriting | Workforce quality becomes a measured underwriting input. Carriers can price the human-execution component of operational risk, not assume worst-case. |
| Geographic / climatic location data COPE per site · nat-cat exposure · supplier concentrationPer-site precision · not per-portfolio averaging | Nat-cat reinsurance attachments price more accurately. Supplier-concentration coverage becomes underwritable for the first time. |
Twenty-four months is where the distribution stabilizes.
The 24-month threshold is anchored to insurance-specific math. Actuarial credibility builds with both claim count and exposure-year accumulation. Six months produces a single seasonal slice; twelve captures one annual cycle; twenty-four captures two — which is the threshold where failure-rate distributions, SLA compliance variance, and tail behavior begin to stabilize against expected parameters. Below 24 months, partial credibility ratios are so low that carriers price the cohort at "ignorance premium" — usually 3–5× rational pricing — or decline.
For operators today: T1 unlocks parametric SLA wraps, equipment warranty wraps, and performance bonds priced from real distributions rather than theoretical worst-case. For Allometry as a future carrier: T1 vintage is the point where MGA programs (Phase 1) become economically viable at scale. The 24-month threshold is where the actuarial table goes from theoretical to real.