How We Score 28 Industries Across 5 Time Horizons

Eight dimensions per industry. 167 cross-industry effects. Five time horizons. All running simultaneously, updated weekly with sourced data. Here's what that looks like from the inside.

Here's the situation. You've got 28 industries. Every one of them is being reshaped by AI at a different speed, in a different direction, on a different timeline. And they're all connected to each other in ways that nobody tracks.

So we built an engine that tracks all of it. Simultaneously. Every industry, every dimension, every cross-effect, every time horizon — updated every two weeks with sourced data. Not hunches. Not vibes. Specific, named evidence that you can check yourself.

This page shows you exactly how it works. No black boxes. No "trust the algorithm." Just the architecture, in plain English, with real examples.

Fair warning: once you see how the pieces connect, you won't be able to look at a single-industry analyst report the same way again.

Step 1: Four Numbers That Define Every Industry's AI Trajectory

Before we touch dimensions, cascades, or scenarios, every industry gets four foundational scores. We call it the CCVR framework, and once you understand it, you'll see why a single "AI impact rating" is basically useless.

The relationship between these four numbers tells the story. High ceiling + low current + high velocity + dropping resistance = coiled spring. High ceiling + low current + high resistance = stuck potential. Every industry has a different pattern. And it changes as real-world events unfold.

Quick example from this month. In early March, insurance data showed that 82% of insurers say AI will dominate their industry, but only 14% have actually operationalized it. That's a huge ceiling-to-current gap with moderate velocity — the coiled spring pattern. Meanwhile, enterprise SaaS just reclassified from "AI assists the worker" to "AI replaces the workflow" after C3 AI and OpenAI launched autonomous agent platforms. That's a velocity spike that changes the whole trajectory. The CCVR framework captures both of these in real time.

Step 2: Eight Dimensions That Shape the Score

CCVR tells you where an industry is headed. But it doesn't tell you what happens to stock prices along the way. For that, you need context. Eight different kinds of it, all running at once.

Each dimension has a different time profile. Regulatory drag might ease over 3-5 years as new frameworks emerge. Competitive moat effects compound. Geopolitics can shift overnight. The engine weights each dimension differently at each of the five time horizons (1, 2, 3, 5, and 10 years).

Here's why this matters right now: Broadcom just posted $8.4 billion in AI chip sales, up 106% in a year, with guidance that inference will be 70% of all AI compute by late 2026. That single data point touches five of these dimensions simultaneously — capital efficiency (custom ASICs vs general GPUs), competitive moat (Broadcom gaining on Nvidia for dedicated inference), geopolitics (US chip policy favoring domestic ASIC production), cross-industry cascade (inference cost drops accelerate adoption across every other industry), and historical pattern (this looks exactly like the training-to-application transition in previous tech revolutions). One earnings call. Five dimensions updated. That's what the engine does every two weeks.

Step 3: 167 Cross-Industry Effects

OK. This is the part that makes people's eyes go wide.

Most analysis looks at one sector at a time. "How will AI affect pharma?" "What happens to banking?" Each question gets an answer. Fine. Except industries don't exist in isolation. They're connected. And nobody tracks the connections.

We do.

167 directional relationships connect the 28 industries in our model. When one industry moves, others feel it. And when those industries move, the ripple continues.

Every one of those 167 connections has three properties:

• Strength — how powerful is the connection? A karaoke company's press release crashing the logistics sector was a 0.90+ strength event.
• Direction — is it positive (one industry's gain lifts another) or negative (one industry's gain is another's loss)?
• Timeframe — does the effect hit immediately, or does it take years to propagate?

The engine doesn't just track first-order effects. Second-order propagation follows the ripples as they spread. When pharma shifts, it touches healthcare, cloud, insurance, chips, SaaS, and foundry equipment — and each of those industries touches others in turn.

Here's one that showed up this month. Roche launched a 3,500-GPU "AI factory" for drug discovery. Earendil Labs raised $787 million to push 40+ AI-generated drug programs into clinical trials. That's biotech creating a brand new category — dedicated bio-supercomputing data centers — which sustains demand for enterprise GPUs even if general AI compute growth moderates. Which affects chip designers. Which affects foundry equipment orders. Which affects energy demand for manufacturing. One biotech announcement, four industries downstream. That's a cascade.

Step 4: Five Timeframes, Not One

"AI will transform healthcare." Sure. Great. When? That's the question that actually matters if you're allocating capital. An industry that scores beautifully at 10 years might score terribly at 1 year. An industry with a strong 1-year score might be fully priced in by year 3.

So we score at 1, 2, 3, 5, and 10 years. Each additional year widens the outcome range by roughly 12%. The 1-3 year scores are positioning signals. The 10-year scores show trajectory across a broader confidence band. Each timeframe weights the dimensions differently:

• 1-2 years: Dominated by market sentiment, speculation cycles, and multiple re-rating. What the market believes matters more than what's actually deployed.
• 3 years: The transition zone. Hype fades, deployment accelerates, early cross-effects start hitting. This is where the most mispricings live.
• 5-10 years: Fundamentals dominate. Revenue expansion, labor restructuring, and full cascade effects determine winners. Sentiment is noise.

The most valuable insight often isn't which industry scores highest overall. It's which industries have the steepest trajectory change between timeframes — the ones where the 1-year story and the 5-year story are completely different.

What the Engine Actually Produces

Here's what you get. Every industry gets a Relative Valuation Score (RVS) at each timeframe. An RVS of 1.0 means "no relative change expected." Above 1.0 means AI is likely to increase relative value. Below 1.0 means AI is likely to decrease it.

The key word is relative. We don't predict stock prices. (Nobody can, and anyone who says otherwise is selling you something.) What we model is which industries are likely to gain or lose value relative to each other as AI restructures the economy. If you're making sector allocation decisions — where to overweight, where to underweight — that's exactly what you need.

Illustrative scores. Actual industry scores available to members.

Example C is the interesting one. It looks bad at 1 year, neutral at 2 years, and then accelerates. A traditional analyst looking at the current picture might avoid it. The engine sees the coiled spring.

The Human Layer: Why This Isn't a Black Box

Every score in the engine traces back to sourced data you can verify. Expert predictions with named sources. Published research. Regulatory filings. Earnings calls. Historical precedent from five previous technology revolutions. The knowledge base currently contains hundreds of individually sourced data points across 9 specialized databases.

When a score changes, the engine logs why. "Broadcom Q1 FY26 earnings: AI semiconductor revenue $8.4B, up 106% YoY. Guidance: inference 70% of AI compute by late 2026. Impact: chip-design-fabless competitive moat increased, cloud-platforms capital efficiency adjusted." Named companies. Specific numbers. Dates. No black boxes. No "trust the algorithm."

This matters because the engine is a framework, not an oracle. It's a systematic way to hold more variables in view than any analyst can — weight them against each other across time horizons, trace the cross-industry effects nobody tracks manually — and hand you the output in plain English with receipts. The scores are the starting point for thinking, not a replacement for it.

Why We Built a Dedicated Engine

It's not the math. The math isn't complicated. It's the scope.

Here's what the engine holds in view simultaneously:

1. Research and calibrate four core scores for 28 industries
2. Score each industry across 8 additional dimensions
3. Map 167 directional relationships between industries, each with strength, direction, and timeframe
4. Model second-order cascade propagation
5. Weight everything differently across 5 time horizons
6. Update it weekly as new evidence arrives
7. Maintain 9 specialized knowledge bases with sourced data
8. Trace every score change back to specific evidence

Any one of those is straightforward on its own. All of them running simultaneously, updated weekly, cross-referencing each other — that's what the engine was built to do.

Related reports: What Wall Street's $40 Trillion Runs On  |  Why Math-Only AI Predictors Fail  |  The Human Bottleneck