If you’re making calls about the AI endgame five or ten years out — this pushes the book’s deflation curve to its limit and asks the question every investor and founder eventually faces: when intelligence itself nears free and floods the world like water and power, where does value flee?

The book began with a curve: the price of a unit of intelligence fell 280-fold in eighteen months, and hasn’t stopped. Every chapter since has answered the same question — as intelligence gets cheaper, where does the money run. Now push that curve to its end: don’t stop at “cheaper,” push all the way to “near-free, near-infinite.” Imagine a world where intelligence floods like electricity — anyone, anytime, can call on near-infinite cognitive capacity at marginal cost near zero. (In reality the curve is held back by power, regulation, reliability, and organizational constraints — but pushing it to the limit is useful.)

It forces a question: when intelligence is no longer scarce, where does scarcity go?

The End of Deflation Is the Migration of Scarcity

One iron law runs through this entire book: value always flows to scarcity. Upstream earns because compute and power are scarce, downstream earns because lock-in is scarce, the model layer doesn’t earn because capability is becoming un-scarce. Every profit migration the book tracks is that one law unfolding at different moments.

So the end of “intelligence deflation” isn’t an “everything’s free” utopia — it’s the migration of scarcity. Intelligence going from scarce to abundant doesn’t make value vanish; it drives value off intelligence and forces it to find its next foothold. To ask “what’s scarce after intelligence floods” is to ask “where’s the next profit pool.” The answer has three layers, each deeper.

Scarcity 1: Judgment

When “generating an answer” becomes free and infinite, the bottleneck shifts instantly from “can we produce” to “knowing what to produce, which output to trust, what decision to make with it.” AI made execution cheap, so judgment got expensive — in a world that can generate a thousand options instantly, what’s scarce isn’t the options but knowing which one is right, why, and when to overturn it.

This is the ultimate version of two of the book’s judgments. Enterprise ROI is slow because the bottleneck is in the un-accelerated step — and the core of that bottleneck is judgment: should we, who’s accountable, is it worth it. The more intelligence floods, the more “what to do with intelligence” matters over “whether you have it.” Capacity got cheap; direction and taste got expensive. Long term, value flows from the layer that “provides cognitive capacity” to the layer that “provides judgment.”

Scarcity 2: Distribution

Infinite intelligence means its outputs are infinite too: infinite content, code, apps, plans. When supply nears infinite, the only bottleneck is demand-side scarcity — attention, entry, defaults, trust relationships. This is the ultimate form of the downstream spine: when everyone can produce, whoever can distribute takes all.

But distribution itself is being rewritten by AI — the object of aggregation slides from “human attention” to “the routing of intent.” So distribution scarcity in the flood era is no longer “who owns people’s eyeballs” but “who owns the entry to intent.” In a world where agents filter infinite supply on our behalf, the layer closest to intent is the only gate turning infinite supply into a real choice — it’s worth more than ever, and unless agent portability and mandated interoperability punch the entry through, it’s more concentrated than ever. This is the physical basis of “winner takes all” in the AI era: not because the winner’s product is better (products get infinitely copied), but because in infinite supply, the scarcity of being chosen belongs to very few entries.

Scarcity 3: Trust — the Endgame Bet

Judgment and distribution both matter, but the endgame bet this book is willing to make is the deepest third layer: trust.

When AI can infinitely, convincingly generate everything — text, image, voice, video, identity, résumé, evidence — one thing collapses faster than ever: authenticity. In a world where any content might be machine-generated and any identity might be forged, “is this real,” “did this person really say it,” “is there a human on the other side” — the answers go from free-by-default to scarce and expensive. When anything can be infinitely faked, the only thing that appreciates is the real.

This is the book’s profit formula at its deepest. Trust is becoming scarce in two ways at once: on the demand side, a world drowning in synthetic content craves “credible”; on the supply side, AI is destroying the old mechanisms that provided credibility. Demand surging, supply collapsing — by any page of this book, that means trust becomes the highest-social-value demand after the flood. It will emerge as a whole new infrastructure layer moving from back office to front: provenance and content origin, identity and “proof of human,” verifiable audit, reputation and endorsement systems.

But state one thing honestly: highest social value doesn’t mean the profit is easiest for any one company to capture. This layer differs from all the rest — it’s not just a profit pool. The flood threatens not just some business models but the entire information environment and mutual trust society runs on. So repricing trust is both the biggest commercial opportunity this book can point to and the biggest social risk it must name — the deepest scarcity AI creates is the very thing it destroys with its own hands, that someone must rebuild.

Gather the three layers: the more intelligence floods, the scarcer judgment, distribution, and trust; and trust is deepest, because AI is systematically destroying it, so rebuilding it is worth the most.

This is the book’s method, applied one last time. From start to finish it did one thing — follow the scarcity. The map goes stale: nearly every specific number and every company’s position in it will change within three years. But the method won’t: find what’s becoming scarce, the money is there; find what’s becoming abundant, the money is leaving.

The prologue said: seeing whose hands the cheap intelligence flows through is worth more than predicting intelligence itself. Here, push it to the limit: when intelligence is as cheap as air, the real question is no longer whose hands it flows through — but, in a world of infinite answers, infinite content, infinite capability, whose judgment is worth following, what can still be distributed to you, and — whose word is still worth believing.

Intelligence will flood. What’s scarce is judgment, distribution, trust. And the scarcest of all, AI can’t give you — it can only make you need it more than ever.

After the flood, which do you most fear losing — judgment, distribution, or trust? Comments open.

Sources (framework argument, June 2026): unit-intelligence price down 280x in eighteen months (Stanford AI Index and others, carried from Chapter 1); the judgment/distribution/trust layers are an era-scale extrapolation of the book’s profit formula, carrying over from Chapters 11, 8, and the spine.

— From the Epilogue of a book in progress, working title The Deflation Sandwich

#AI #AIEndgame #DeflationSandwich