The Moat & Meter AI Value Matrix
How to decide which AI bets to fund, which to adopt cheaply, which to hold — and which to kill on sight, before the market makes the call for you.
The Moat & Meter AI Value Matrix is a two-axis framework for prioritizing AI investments. The Meter scores how far an initiative moves your P&L. The Moat scores its advantage half-life — how many quarters the edge survives once competitors rent the same models. Every initiative lands in one of four quadrants — Fortress, Commons, Vault, or AI Theater — and each quadrant has exactly one prescribed move.
Rented genius: the input everyone has
Every strategic input before this one arrived scarce. Land came with deeds. Capital came with gatekeepers. Even software — supposedly free to copy — took years and a payroll of engineers to actually wield. Foundation models broke the pattern: they are the first strategic input in history that arrives pre-commoditized. The day a frontier model ships, your fiercest competitor rents the same IQ, at the same list price, the same afternoon.
That single fact rewrites the economics of every AI business case on your desk. The model is never the moat — it's the great equalizer. Any value you create with rented intelligence and cannot fence gets competed to zero, and the surplus leaks straight through to the customer. Wonderful for them. Nothing for you.
The receipts are already in. MIT's NANDA group found that 95% of enterprise GenAI pilots deliver no measurable P&L impact (MIT NANDA — The GenAI Divide, 2025). BCG puts the share of companies actually capturing significant value from AI at roughly 5% (BCG — Are You Generating Value from AI?, 2025). Meanwhile the few who wire AI into the core are posting 10–25% EBITDA gains while everyone else stays stuck in pilots (Bain — Technology Report, 2025). Both groups rent the same models. The gap isn't the technology — it's what the technology gets plugged into.
Rented genius throws off enormous value and keeps almost none of it. The winners aren't the ones with the best AI — they're the ones who own the thing the AI has nothing without.
Moat & Meter exists to make that distinction impossible to dodge. It forces every initiative in your portfolio to answer two questions it would rather not answer.
Two questions, precisely defined
The Meter
"If AI nails this initiative, how far does the P&L actually move?"Score it in orders of magnitude, not decimals: does this shift a line item by 0.1%, 1%, or 10%? A meter measures flow — so if you can't name the specific P&L line the flow shows up on, the reading is zero, no matter how impressive the demo. "Productivity" without a line item is a feeling, not a Meter reading.
The Moat
"How many quarters does the edge survive once rivals rent the same models?"This is advantage half-life. Not whether you have an edge today — how fast it decays. Two quarters means you bought a head start, and head starts get repriced to zero. Eight-plus means you own something the model can't hand your competitor. The Moat is measured in quarters because that's how fast this market erodes an unfenced position.
The Moat is the neglected axis, and neglecting it is the single most expensive habit in enterprise AI. Underwriting an AI initiative without an advantage half-life is underwriting a bond without a maturity date — you know what it pays, you have no idea for how long.
Every AI business case models the upside. Almost none of them models the decay.
Four quadrants, four moves — no fifth option
Plot every current and planned AI initiative on the two axes and your portfolio sorts itself. Each quadrant carries exactly one prescribed move. The discipline is refusing to negotiate with the quadrant.
Fortress
AI pointed at a scarce complement only you own — proprietary data, an embedded workflow, a regulated or physical asset. The value is large and it compounds, because rivals can rent the model but not the thing it runs on.
- Yield prediction trained on your plant's proprietary sensor history
- Dynamic pricing on a decade of your own transaction and loss data
- Predictive maintenance on telemetry from a fleet only you operate
Commons
Real money, no fence. These initiatives genuinely move the P&L — and every competitor gets the identical capability the same quarter, so the edge decays on contact and the surplus flows to customers.
- Customer-support copilot answering from public docs
- AI-drafted marketing copy and sales outreach
- Code assistants for the engineering team
Vault
Defensible but dormant: built on assets only you hold, yet not wired to a P&L line — today. Vaults are where future Fortresses sleep, and where impatient CFOs mistakenly swing the axe.
- Internal LLM Q&A over your own policy and contract corpus
- Digitizing the tribal knowledge of a retiring machinist workforce
- Structured capture of field-service and warranty call notes
AI Theater
Announcements dressed as strategy. No line item moves, no fence exists — the initiative's actual product is the appearance of momentum. Theater doesn't just waste its own budget; it consumes the political capital your real bets need.
- The AI press release with no system behind it
- The innovation-lab demo that never touches an order or an invoice
- "AI-powered" rebadging of a feature that already existed
Enclose the Commons
Most portfolios are Commons-heavy, and most leaders treat that as bad news. It isn't — it's raw material. A Commons position is a Fortress that hasn't been fenced yet. The enclosure playbook is four steps, run in order:
Adopt cheaply
Take the commodity capability at commodity prices. Buy, don't build — you are not going to out-engineer the model vendors, and you don't need to.
Instrument everything
Wire the system so every interaction captures data only you can see: what was asked, what worked, what failed, what it was worth. This is the fence going up.
Close the loop
Feed that private data back so the system measurably improves on it — better answers, better routing, better pricing. Now it compounds with use, and a rival adopting the same tool starts from zero.
Re-score it
Two quarters later, plot it again. The Meter held. The Moat lengthened. The Commons has hardened into a Fortress — built from a tool anyone could rent.
This isn't theory — it's how our own AI concierge grew up. It started life as exactly the kind of commodity chat wrapper any competitor could stand up in a weekend. Then we instrumented it: every conversation captured demand, objections, and unmet searches nobody else could see. We closed the loop, and now it books, sells, and routes on private data that compounds daily. Same rented model as everyone else; entirely different asset. The full case study is here.
A Commons position isn't bad news. It's a Fortress that hasn't been fenced yet.
The 10-question rubric
You don't need us in the room to run a first pass. Take any AI initiative — live or proposed — and answer ten yes/no questions. No maybes: if you have to argue for a yes, it's a no.
The Meter — five questions
- 1Can you name the single P&L line this initiative moves?
- 2Would the movement be measurable within one quarter, not "eventually"?
- 3Is it revenue growth or hard cost-out — not soft "productivity" hours?
- 4Does it touch a process that runs daily and at volume?
- 5Would the CFO fund this with the word "AI" deleted from the proposal?
The Moat — five questions
- 1Does it run on data only you hold — not scraped, bought, or public?
- 2Is it embedded in a workflow that would hurt to rip out?
- 3Would switching away cost your customers or team something real?
- 4Is it fenced by a regulated, physical, or contractual asset rivals can't rent?
- 5Does it compound with use — measurably better every cycle it runs?
Scoring a real bet, end to end
Take the most common initiative in enterprise AI right now: a customer-support copilot answering from your public documentation. Here's the rubric, applied honestly.
Customer-support copilot on public docs
This is the framework. The Diagnostic is us applying it to your portfolio, live.
The rubric above gets you a first pass. The AI Value Diagnostic is the full instrument: a working session where we plot every current and planned AI bet you have on Moat & Meter — and you leave with a single board-ready slide showing what to fund hard, what to enclose, what to seed, and what to shut down Monday.