Claim
The economic counterweight to AI abundance is a scarcity supercycle: as the cost of producing competent output collapses, demand for assets that are structurally hard to replicate (human judgement, named relationships, in-person service, original taste, historical brand authority, real-world certifications) doesn't fall — it rises. The pricing power is moving toward the parts of the stack that cannot be commodified by capability gains in models. Operators who keep selling the abundant thing as their wedge get squeezed; operators who identify their structurally scarce asset and lean into it command a premium.
Mechanism
Abundance and scarcity are coupled: the more cheaply you can produce one kind of output, the more valuable the kinds you cannot produce that way become — partly because of substitution pressure, partly because humans assign psychological value to the rare. AI compresses the production cost of analysis, drafting, summarisation, generic UX. Things AI cannot compress: a 17-year operator relationship, a named CSM who knows your business by heart, the founder's actual taste, a brand that's been trusted for two decades, a credential that requires a five-year cohort to acquire, a cohort experience you can't bulk-replicate. As the abundant tier saturates, willingness-to-pay for the scarce tier increases, not just relatively but absolutely.
Conditions
Holds when:
- The category has a real abundance/scarcity split — some operations can be AI-commodified, others can't.
- The buyer can perceive the difference (taste, judgement, lived experience are legible to them).
- The operator has the discipline to actually invest in the scarce asset rather than over-stretching the abundant one.
Fails when:
- The category is wholly commoditisable (where AI eats spec-to-verification end-to-end, the scarcity tier never forms).
- The buyer is purely procurement-driven and doesn't pay for scarcity premiums.
- The "scarce asset" claimed is fake — branding without history, "expert relationships" that don't exist, manufactured exclusivity. Buyers spot these eventually and the premium evaporates.
Evidence
"Humans fucking love scarce things. Always have."
— Packy McCormick, The Scarcity Supercycle, https://www.notboring.co/p/scarce-assets, 2026-05-06.
The piece names concrete examples: human-backed service tiers in B2B SaaS where competitors are AI-only, named CSMs as a paid premium, concierge migration support, founder-led sales for ACV thresholds where AI can't carry context.
Signals
- The operator's pricing has a clearly distinct tier that's not about features but about the human / scarcity element.
- The marketing language for the high tier doesn't try to compete on AI capability with the low tier — it competes on what AI doesn't do.
- Customers who pay the premium tier renew at higher rates, refer more, and cost less to retain than the low-tier base.
Counter-evidence
- "Human as the moat" can become an excuse for not improving the abundant tier. Some buyers will pick the cheaper AI-only product even if the human tier is better, and the share of the cheaper tier grows.
- McCormick's frame is most legible in B2B SaaS and luxury markets; it's less load-bearing in commodity infrastructure where humans don't add visible margin to the buying experience.
Cross-references
- Building costs collapsed; judgement didn't — the squeeze is on positioning, not production — Indig's parallel framing: judgement is the part of the workflow that doesn't compress.
- The middle is hollowing out — execution gets automated, leaving spec-writing and verification as the high-value human tasks — Yan's structural diagnosis is the production-side mirror of McCormick's demand-side argument.
- The cost of intelligence is converging toward the cost of electricity — durable advantage isn't using AI, it's parlaying AI — the price collapse on the abundant side that's driving the supercycle.