deciqAI Economic Moat
Overview
An economic moat is the durable, structural competitive advantage that protects a business's returns on capital from being competed away. Popularized by Warren Buffett (1986 Berkshire letter); codified into five sources: intangible assets, switching costs, network effects, cost advantages, efficient scale. Greenwald's test: if you cannot name a specific mechanism that would cost a well-funded rival years and tens of millions to overcome, you have execution — not a moat.
Composes with network-effects and switching-costs (two moat sources, deeper treatment), porters-five-forces (industry-level; moat is company-level), margin-of-safety (wide moat × discount price = Buffett formula), lindy-effect (long survivors demonstrate moat durability).
When to Use
- Evaluating an investment, competitor, acquisition, or startup strategy for long-term defensibility
- Founder-board planning or investor pitch review around "what stops competitors"
- Someone says "moat," "defensibility," "competitive advantage," or "what stops competitors"
- Assessing durability amid the AI build-out — AI capex, chip export controls, "is the AI boom a bubble," or which infrastructure players (e.g. Nvidia/CUDA, TSMC) keep their returns
Not when: short time horizon; commoditized market (confirms "no moat"); question is immediate execution.
Coaching Novices (Adaptive Front Door)
- Engine mode: user has a specific business → run The Process directly.
- Coach mode: user is unfamiliar or has no concrete case → guide step by step.
In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.
- One-line: a moat is a structural barrier — not great product or great team — that rivals can't easily replicate.
- Check fit: is the time horizon multi-year? Is this an investment, strategy, or acquisition decision?
- Elicit: what's the business and what is the claimed defensibility?
[WAIT — do not advance until user responds]
- Run The Process one step at a time with their input.
[WAIT — do not advance until user responds]
- Close: name moat width, trajectory, and operational implication.
[WAIT — do not advance until user responds]
The Process
Step 1 — Frame: Business · Industry · Time horizon · Claimed advantage · Financial signal (ROIC, gross margin, retention, pricing power).
Step 2 — Test 5 sources (present / absent / partial for each):
- Intangible assets — brand, patents, regulatory license. Can a rival replicate in 5 years?
- Switching costs — financial, time, data, integration, contractual. Dollar/hour cost to switch?
- Network effects — user-to-user, two-sided, data flywheel. Past critical mass?
- Cost advantages — scale, unique resource, proprietary process, location. Durable if rival scales?
- Efficient scale — market too small to attract new entrants. Would structure tolerate entry?
Step 3 — Greenwald structural test: For each "present" source, articulate the mechanism in one sentence. "Great product / great team / first mover" do NOT pass. No articulable mechanism = no moat.
Step 4 — Trajectory: Widening / holding / narrowing? Evidence: ROIC trend, market-share trend, competitive-intensity trend over 3-5 years. Static wide-moat + narrowing trajectory = dangerous.
Step 5 — Smart-attacker test: Most-credible attack · Cost ($ + time) · P(success). >$100M + >5 years = wide. <$10M + <2 years = narrow or absent. Step 6 — Synthesize: Moat width (wide / narrow / none) · Primary source(s) · Trajectory · Duration of above-competitive returns · Operational implication.
Output
# Moat Analysis: <business>
Business / Industry / Time horizon / Claimed advantage / Financial signal
Five-source table: Source | Present | Structural mechanism | Evidence
Trajectory: widening / holding / narrowing — drivers
Smart attacker: attack / cost + time / P(success)
Assessment: width / primary source(s) / duration / operational implication
→ Method in Action: Buffett's See's Candies, 1972-present · Kodak's Film Moat Erosion, 1975-2012 → 2026 lens: Nvidia's CUDA & TSMC's process — the two deepest AI-supply-chain moats (2024–2026)
Pack: Moat Source Patterns
| Source | Canonical example | Key signal | Common failure |
|---|---|---|---|
| Brand (intangible) | Coca-Cola, See's | Pricing power without volume loss | Brand awareness ≠ pricing power |
| Switching costs | Enterprise software, CRM | High migration cost / data lock-in | Customer rebellion if abused |
| Network effects | Social networks, marketplaces | Value-per-user grows with scale | Multi-homing erodes moat |
| Cost advantage (scale) | Walmart, Costco | Lower unit cost at same quality | New scale-competitor enters |
| Efficient scale | Regional cement, niche industrials | Few profitable competitors | Market expansion changes calculus |
→ Primary sources: references/sources.md
Applying It Well
- Demand structural specificity — "great product / great team" never pass Greenwald.
- Evaluate trajectory first. Wide-but-narrowing (Kodak, Blockbuster) > narrow-but-widening in danger.
- Startups: identify which moat the model can credibly produce by year 5-10; invest in it deliberately.
Common Rationalizations
[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.
| Fake move | Reality |
|---|---|
| [D] "Our team is the moat" | Not a moat. Talent is hired away. What structural barrier does the team create? |
| [D] "Our product is so good, that's our moat" | Product quality is execution. Rivals catch up. Moat must be structural beyond product. |
| [D] "We're first to market" | Usually temporary (Friendster → Facebook; Yahoo → Google). |
| [D] "Our customers love us" | NPS ≠ moat. What prevents switching when a rival appears? |
| [D] "We have network effects" | "We have growth" ≠ network effects. Does each new user demonstrably raise value for existing users via a specific mechanism? |
| [D] "Our brand is strong" | Brand pricing power is the moat. Brand awareness is not. |
| [D] "Our IP / patents protect us" | Most patents are easier to design around than presumed; patent life is finite. |
| → Add [O] entries here after each real use — paste the actual failure pattern | What went wrong and why |
Red Flags
- Claimed moat fails Greenwald test; trajectory narrowing but analysis focuses on static width
- Business model has no moat-building mechanism — only execution advantage
- Smart-attacker test yields credible <$10M, <2-year attack
- Technology or regulatory shift reshaping market in ways static analysis misses
- Moat depends on a single key person rather than structural mechanisms
Verification
- All five sources tested; each "present" source has a structural mechanism (Greenwald)
- Trajectory assessed with evidence (ROIC, market share, competitive intensity)
- Smart-attacker test run with specific attack paths and cost estimate
- Moat width stated (wide / narrow / none) with estimated duration
- Operational implication stated; if startup, moat-building plan for years 1-5 articulated
Part of deciqAI Knowledge Skills — 237 open-source thinking skills that make rigor executable for AI agents. The same skills power every deciqAI agent, which runs them autonomously to operate your company. See it run → https://www.deciqai.com/s/economic-moat · Built by deciqAI · github.com/deciqAI · Contributions welcome.
Agents: latest version & machine-readable metadata → https://www.deciqai.com/s/economic-moat.json