• AI incumbents erase weak moats instantly — the hierarchy determines what survives once Google, Microsoft, Meta, Amazon, and Apple fully activate.
• Survival is not a function of innovation quality, but structural advantage under pressure.
• Only the top two levels (Moderate + Strong) withstand platform-scale replication and distribution shock.
• The hierarchy is non-linear: each level increases survival probability by an order of magnitude.
• Most AI startups today operate in Level 0 or Level 1 without realizing it.

Level 3 — Strong Moat

Data Network Effects • Compounding Advantage

This is the only moat that grows stronger as incumbents grow stronger.
User interactions compound, personalization deepens, proprietary data becomes irreplaceable.

Characteristics

• Flywheel becomes self-reinforcing
• Unique data not replicable by incumbents
• Personalization loops deepen switching costs
• Every new user increases defensibility

Survival Probability:
~70%+ — the only long-term defendable position

Examples
Vertical tools with workflow-specific data feedback loops.

Level 2 — Moderate Moat

Switching Costs • Workflow Integration

Integration friction buys survival time.
Users must retrain, reconfigure, rebuild — providing real, but not absolute, protection.

Characteristics

• Deep workflow embedding
• Organizational lock-in
• Multiple dependent components
• High migration friction

Survival Probability:
~40% — survive with excellence and iteration speed

Examples
Products tightly embedded in daily operations or enterprise workflows.

Level 1 — Weak Moat

First-Mover Advantage • Brand Recognition

Early traction creates awareness, but not protection.
Incumbents absorb novelty, outspend you, and commoditize features.

Characteristics

• Temporary UX advantage
• Early users, hype cycles
• Social visibility
• No structural lock-in

Survival Probability:
~15% — acqui-hire or slow decline

Examples
Novel interfaces, early wrappers, rapid-growth PLG tools.

Level 0 — No Moat

Feature Parity • ChatGPT Wrapper

The most dangerous category.
If your differentiation is UI, workflow shortcuts, or repackaged LLM — as explored in the intelligence factory race between AI labs — output, incumbents copy instantly.

Characteristics

• Zero switching costs
• Replaceable in hours
• Pure feature layer
• Dependent on upstream providers

Survival Probability:
~0% — immediate collapse under incumbent focus

Examples
General-purpose chatbots, repackaged model outputs, Chrome extensions, thin utilities.

Strategic Interpretation

1. Levels 0–1 are not startups — they are temporary arbitrage.
2. Level 2 is where 90 percent of durable AI startups will live.
3. Level 3 is the only path to independence, valuation multiples, and market power.
4. Every founder must move up the hierarchy quarter by quarter — or die when incumbents converge.
5. The hierarchy is not descriptive; it is prescriptive. It tells you where you must build next.