04Standing analysis · A worked reading
Six indicators, one composite reading. Two are anchored to reported data and cite their source on the card; the other four, and the composite itself, are editorial synthesis — directional, not a live index. The point isn’t to predict the market. It’s to keep the distance between conviction and evidence visible while it’s still cheap to correct.
Heated 79/100
Conviction is running ahead of evidence. Expect louder claims, thinner proof, and a widening gap between pilots and production.
Composite of six weighted indicators, 0 (frozen) to 100 (overheated).
Each indicator is normalized to a heat score from 0 to 100, where 100 reads as maximally overheated. Two indicators are inverted: for those, low raw values heat the reading.
The share of S&P 500 companies citing “AI” on earnings calls — narrative pressure in its rawest form. It hit a ten-year high, around two-thirds of the index, in 2025.
The share of generative-AI pilots delivering a measurable return. MIT’s 2025 study found roughly 95% see none; the lower it goes while spend rises, the hotter the market.
AI-attributed staff cuts followed by rehiring for the same function within a year — conviction spent before evidence.
Enterprises with an AI budget but no named owner or written strategy. Money moving faster than accountability.
Median revenue-multiple premium of AI-labeled firms over comparable peers. What the label alone is worth right now.
Firms with no AI capability plan whatsoever — the frozen end of the market. High values cool the composite; that failure just books later.
Frozen 0–15
Activity has stalled well below what the technology already reliably does. The failure here is quiet: lost bids, lost optionality.
Cool 15–40
Adoption is deliberate and evidence is keeping pace with conviction. Underinvestment is the risk to watch.
Balanced 40–60
Conviction and evidence are roughly in step. Decisions made in this band tend to survive the next cycle.
Heated 60–85
Conviction is running ahead of evidence. Expect louder claims, thinner proof, and a widening gap between pilots and production.
Overheated 85–100
The narrative is carrying spend that the operating results no longer support. Corrections from this zone are rarely gentle.
Two indicators are anchored to reported figures and cite their source on the card: AI mentions in earnings calls (FactSet, which put the 2025 share at a ten-year high) and the pilot-to-return rate (MIT and Gartner, on the ~95% of pilots that see no return). The other four are editorial estimates, labeled as such. The composite is an editorial synthesis of all six — not a measured index.
Every indicator is normalized against a fixed floor and ceiling to a heat score between 0 and 100, where 100 reads as maximally overheated. Indicators where a low raw value signals heat — the pilot-to-return rate, the no-plan share — are inverted before weighting.
The composite is a weighted mean. Weights favor indicators that measure the gap between stated conviction and operating evidence (pilot conversion, reversals, ownerless budgets) over pure narrative volume.
Zones are fixed, not fitted: 0–15 frozen, 15–40 cool, 40–60 balanced, 60–85 heated, 85–100 overheated. Both tails are failure zones by construction — that is the editorial thesis, stated upfront rather than hidden in the scale.
The four editorial indicators and the composite are directional estimates compiled from surveys and press review — built to compare the shape of quarters, not to trade on. Even the two anchored figures are point-in-time and will drift. Not investment advice.