Last price
$1,073.52
The quiet monopoly on the most-reused three-digit number in American finance — and the clearest outcome-priced, AI-native franchise in the index.
Live quote sourced from Yahoo Finance. Prices cited in narrative below reflect the original memo date and may be stale.
FICO is the cleanest 'services-as-software' incumbent in public markets: it already sells an outcome (a decision), priced per use (per pull), with no human in the loop, at ~90% gross margin. The thesis upgrade path is adding more decisioning outcomes — Falcon-style fraud scoring, small-business credit, open-banking-enriched consumer risk, and prescriptive decisioning on the FICO Platform — all sold on the same pay-per-outcome rail. The quiet frontier is Falcon; the explosive frontier is applied analytics outside consumer credit (healthcare, insurance, telco churn).
The framing is: FICO has the monopoly number but limited unit-demand growth (US adult population doesn't expand quickly). The growth path is (1) price per pull, (2) new use cases per pull, and (3) selling additional models into the same pipe. (1) and (2) are working — the mortgage price hikes and the B2B2C tri-merge model are live evidence. (3) depends on execution of FICO Platform and competitive defense against hyperscaler ML platforms. If the platform lands, FICO becomes Falcon-scale across dozens of verticals. If not, the stock is still a royalty on the score — just with less optionality.
FICO has been doing services-as-software since before it was called that. Every credit pull is a cash register ring. The mortgage-score list price has gone from $0.60/pull to $4.95/pull in three years and volumes held up — a live demonstration of pricing power on an outcome nobody else can mint. Falcon's per-transaction fraud pricing is the same shape. As LLM-driven decisioning moves mainstream, FICO's decades of performance-labelled outcomes are training-data collateral that hyperscalers can't manufacture.
The flipside of monopoly pricing is political attention. CFPB, FHFA, housing advocates, and multiple senators have put FICO pricing on notice. FHFA approved VantageScore 4.0 for conforming mortgages in 2024; Fannie/Freddie are at the edge of a migration decision. Outside mortgages, open-banking-native credit models (Upstart, Pagaya, and bank-internal builds) show that the score can be bypassed for some populations.
| Segment | Approx. mix | AI posture | Services-as-software read |
|---|---|---|---|
| Scores (B2C tri-merge) | ~50% | Fully autonomous score emission | Textbook — the outcome IS the software |
| Scores (B2B — mortgage, auto) | ~20% | Price per pull, no human in loop | Core thesis, pricing power visible |
| Applied Analytics — Falcon (fraud) | ~15% | Live scoring per transaction | Core thesis; scales globally |
| Applied Analytics — FICO Platform | ~10% | Copilot + deployment of custom decisioning | The expansion lever — less proven |
| Professional services + other | ~5% | Mostly human-led | Non-thesis |
FICO has raised mortgage-score pricing roughly an order of magnitude over three years with negligible volume response. That is the empirical proof of thesis for any services-as-software argument. Investors should extrapolate the pricing-power curve into other tri-merge and B2B categories before stressing unit volume.
Falcon scores the majority of card-present fraud globally, updated in real time, with a bank-network acceptance moat analogous to the consumer score. The fraud product is less politically visible than the credit score and has long runway as instant-payment rails expand in LatAm, SEA, and the EU.
Where hyperscalers sell ML primitives, FICO sells governed, explainable, audit-trail-included decisioning. For regulated industries — banks, insurers, telcos, healthcare payors — that's a different product. Platform ARR growth implies early product-market fit.
Gross margin near 80%, operating margin in the mid-40s, FCF conversion >100%, low capex intensity, net-cash balance sheet after aggressive buybacks. If the thesis compresses software valuations toward outcome-priced models, FICO's multiple defends structurally.
FHFA has formally approved VantageScore 4.0 for conforming mortgages. A mandated two-score or single-score switch would compress list pricing and possibly volume. Even a slow migration creates narrative overhang that caps the multiple.
Multiple senators and the CFPB have referenced FICO by name. Under a different administration, a regulatory cap on mortgage-score pricing is plausible. That would not kill the business but would cap the clearest leg of the bull thesis.
FICO Platform goes head-to-head with Databricks, Snowflake's ML layer, and big-bank internal MLOps. Customer logos are real but the product's defensibility outside credit decisioning is not yet proven. Sales cycles are long; competitive win rates are uneven.
You cannot 10x score pulls the way you can 10x seats of a productivity app. Growth is price-led, which is the opposite of what bulls want when the political backdrop sharpens. The stock will always trade on whether pricing power continues.
FICO is the purest public-market expression of the thesis. It already sells outcomes, not tools; pricing is per use, not per seat; the execution layer is already embedded; there is no human in the loop for the core product; and margin structure looks like what the thesis predicts services-as-software companies should look like at maturity. The differentiator versus other 'AI-native' incumbents (Palantir, Intuit, Shopify) is that FICO arrived at that state thirty years ago and has been compounding on it. The upside case is Platform + Falcon becoming a decisioning-cloud equivalent. The downside cases (VantageScore, CFPB, open-banking disintermediation) are real but bounded.
The clearest outcome-priced franchise in public markets. Own through regulatory noise; Platform + Falcon are the multi-year optionality; the base case is already a high-quality compounder.