AI’s New Economics
What Keeps Its Value — and Who Captures It
Globalization reset the market for reproducible physical labor; AI is doing the same for cognitive labor. The fear is rational. But value doesn’t vanish — it moves; where it goes, and who captures it, is the whole question. The answer runs in three parts:
- AI makes repeatable cognitive labor abundant. Work that can be procedurally reproduced has its price competed toward zero; judgment and tacit knowledge — which can’t be reproduced — retain their value.
- In an abundant-supply world, demand is the scarce thing. Value shifts to whoever owns the customer relationship — the aggregators racing to control the agent interface — while every layer beneath risks being commoditized.
- The defense is sovereignty. Own the substrate you depend on, make models contestable, and keep direct relationships with your customers — so that no supplier controls your memory, no lab makes you dependent, and no aggregator takes your market.
This is the map for where value flows; the sections below walk it.
The supply shock
The mechanism is a growing abundance of cheap cognitive labor. AI floods the market with “employees” that are effectively unlimited, cost relatively little, and get more capable with every release.
The line is reproducibility, not seniority.1 The flood takes the paralegal drafting the contract and the senior associate reviewing it, the bookkeeper closing the ledger and the analyst building the model. Wherever the work is a repeatable procedure — however senior the hands doing it — a machine can now do it more cheaply, and its price competes toward zero.2 This flood will not recede; it is the new sea level.
Two kinds of work provide safe harbor in this AI flood, both because they are hard to reproduce. Judgment and taste are the first — knowing what’s worth doing, telling good from good-enough, and standing behind the result. AI can provide the cognitive labor to do the work. However, it cannot yet decide what the work should be, when it is done, whether it is done well, or own the mistakes.
Tacit knowledge is the second form of safe harbor — the hard-won expertise you can’t quite write down, the kind that lives in the hands and habits of the people and teams doing the work. Its safety is structural, not temporary: models learn from what has been recorded, and tacit knowledge is never recorded because it can’t be. There is no dataset for the feel of a process running right, or the read of which move fits this moment. TSMC is the standing example — its chipmaking runs on public blueprints, yet no rival can copy its fabs because the knowledge lives in people, not documents.3
As AI commoditizes the reproducible and the explicit, the two things it can’t touch — judgment and tacit knowledge — grow more valuable, not less. When anyone can generate a competent draft, the scarce skill is knowing which draft is right, and standing behind it. Value flows to whatever stays scarce.4
That is where the supply story ends: two kinds of irreproducible work survive.5 But surviving a flood is not the same as getting paid for it.
The demand grab
The most valuable position in an AI economy isn’t on the supply side at all. It’s owning demand — and the story starts, as Ben Thompson’s aggregation theory did, with text.6
The internet made publishing nearly free: for the first time, anyone could reach the whole world without owning a printing press or a distribution deal, and the supply of text stopped being scarce. So the value moved from producing the writing to controlling the demand for it; Google aggregated the world’s readers by owning discovery. In a flood of free text, search was how you found anything worth reading, so every publisher had to compete for that attention on Google’s terms. The publishers were commoditized; Google captured the value. (Amazon did the same to merchants, Uber to drivers.) Once supply is abundant, whoever owns the demand owns the game.
AI is now doing to every form of cognitive work what the internet did to text: collapsing the cost of supply, and handing the prize to whoever aggregates the demand. And it goes further this time. Google still needed the publishers to produce the writing; an aggregator drawing on abundant, near-free cognitive work doesn’t need any particular supplier. Being aggregated used to mean thin margins. Now it means no leverage at all.
Who owns demand today? The incumbents: Google owns search, Apple the phone, Microsoft the enterprise. But the agent interface puts those positions in play: it’s the new discovery layer — how you find the right output amid a flood of near-free cognitive work. If and when most cognitive work routes through an agent-mediated conversation, the incumbents’ search boxes and home screens stop being where demand forms — and interaction shifts are how aggregators fall. Desktop-to-mobile cost Microsoft the consumer market; search-to-agent could cost Google the search market.
This economy-spanning opportunity is why the AI labs, already running the leading agent interfaces, have a real shot at owning demand. And they have existential reasons to take it: their models are commoditizing under them, so their escape is to stop being an input to someone else’s software, and to become the interface every cognitive task runs through.
The incumbents are running the same race from the other side, wiring models into every demand surface they own. Whoever wins pairs a capable model with the dominant agent interface — the model is the commoditizing half, the interface the scarce one — to aggregate demand for every cognitive task. A model-plus-aggregator, built from either direction, will have the power to reprice every layer beneath it.7
So the supply-side threat is that AI makes your work cheap. The demand-side threat is that the winner of the interface race owns your customer — and a customer the aggregator owns is one it can cut you off from. Which raises the question the rest turns on: if owning demand is how value gets captured, how does anyone who isn’t the aggregator keep a share?
The defense: sovereignty through ownership
The answer is ownership — the path to sovereignty, the freedom to keep operating, or switch, without a supplier’s leave. You already have it over your phone number: you can carry it to any carrier, so no carrier can hold it hostage — they compete on price and service instead. Note that you don’t run your own cell towers; you depend on a carrier completely, you’re just not trapped by one. Most companies have no such freedom over the data they run on.
Ownership matters because promises don’t. Depend on any supplier — your data in its system, your team trained on it, your workflows built around it — and it can raise the price, because you can’t easily leave.8 A pledge not to exploit that lock-in is worth nothing: nothing binds tomorrow’s price to today’s promise. Even a contract holds only until leverage shifts — contracts are incomplete, and their gaps are settled by whoever has the power at renegotiation. Only structure binds. Own what you depend on, in a form you can pick up and move — your photos out of one cloud, your data out of any vendor — and there is no trap to spring. Custody is the one barrier cheap agents can’t route around, which is why the defense is ownership, not exit tooling.9
Sovereignty has three parts, one for each way an aggregator captures you:
- Own your substrate — Your substrate is your data and, now, your context — the memory, evals, and accumulated judgment your agents build up. Never let a supplier control your memory.
- Make models contestable — Models should be swappable engines pointed at your substrate, where your evals and learned behavior live under your control. Never become dependent on any one lab.
- Keep your customer — You must keep a direct relationship with your customer — the aggregator’s discovery layer can be a channel, never your only path to them. Never give the aggregator the opportunity to take your market.
Buyers at every level must fight for this sovereignty: people own their context, companies their data, vendors their judgment and their customers’ trust. Every layer buys from the one above and sells to the one below, and the same question returns at each — what must I own so the layer above can’t reprice me?
Why would any supplier go along with this? Because giving up lock-in and enabling sovereignty raises its profit — provided the value was ever the work, not the captivity. The threat of the trap is what keeps customers from committing.10 And because every company needs its data back and its models swappable, buyer demand turns “hand it over” from a concession into a market — one the old switching-cost moats (Oracle, Salesforce, Bloomberg) can’t defend, now that AI drives the price of the labor of leaving toward zero.11
Ben Thompson is the proof. He worked out aggregation theory by watching Google commoditize the publishers — then refused to be one. Stratechery runs on his own site, to subscribers he owns, not rented from the Times or Substack; he kept his substrate and his customer both. The argument of this piece is just his move, generalized: what one writer did for text, every firm must now do for cognitive work.
The AI labs exception
The AI labs are the exception to this defense; they hold the most leverage, and have the least ability to commit to buyer sovereignty. The exception is precise: sovereignty still shields you from a lab, but it can never make a lab trustworthy the way it makes an ordinary vendor trustworthy. A normal vendor can hand you your data and walk away; a lab can’t hand you a copy of a model that keeps improving, and even a frozen copy leaves the lab controlling the live model you actually use.
Open-weight models — increasingly, the strong ones from Chinese labs — soften this. Weights you host are weights no one can revoke, which is what makes them the outside option that keeps every closed lab honest.12 But the escape is partial: your copy stops improving the day the publisher stops publishing, and for a regulated buyer a Chinese model trades one government’s off-switch for another’s. So lab promises remain the least credible in the stack, right where the temptation to break them is largest.
And the exception follows the position, not the pedigree: an incumbent that adds a model inherits the same inability to commit — a franchise at risk is a reputational incentive, not a structural commitment, and the whole point of this defense is to stop accepting incentives as commitments.
This isn’t hypothetical. In June 2026, Anthropic’s Fable 5 release showed the risks posed by the labs in three ways at once:13
- mandatory thirty-day retention for Mythos- and Fable-class traffic, even for enterprises with zero-data-retention agreements;
- broad “safeguards” that could block or reroute legitimate work; and
- within days, a suspension of access while Anthropic implemented a U.S. government directive restricting foreign nationals.
Access was later restored, but two parties — Anthropic and the U.S. government — each held an off-switch the customer didn’t. And the market read it the same way: Palantir’s CEO went on CNBC describing enterprise customers “livid” at the frontier labs — demanding to know who controls the weights, the data, and the alpha of their business — while Ben Thompson called the retention change the most underrated aspect of the release.14
So the labs can’t be disciplined from inside — no architecture makes their promises credible — only checked from outside: by customers who keep their own data and stay able to switch models, by firms that keep the customer relationship out of the lab-owned interface, by the real-world inputs the labs can’t manufacture (compute above all), and by the government. The same thing that protects every other layer protects against the aggregator too: the part of the world that isn’t free to copy.
What this predicts
Value collects at the two scarce ends — judgment and tacit knowledge on one end, ownership of demand on the other — and at anything anchored to a real cost. It drains out of the free middle. Individuals are absorbed fastest. Companies split: those anchored in the real world hold, while purely digital firms — where the value is reproducible and the demand is mediated — are hollowed. Software vendors that hoard data are absorbed;15 those that hand it back and charge for the work survive, and the lock-in-heavy ones feel it first. And the labs and incumbents keep fighting for the agent interface, so their conflicts with customers and governments recur rather than settle.
Three honest limits. Timing is uncertain — the direction is set, but the speed depends on adoption, regulation, and compute. The safe ground shifts — when robots flood physical labor, the protection real-world work enjoys today erodes. And the defense is a floor, not a fortress: it protects margins and continuity, not immunity. A consumer-facing firm can do all three and still watch demand migrate to conversations it isn’t in — its customers are people, the most exposed layer. What the defense buys there is being the name the customer carries in their own context.
The bottom line: sell what can’t be reproduced — judgment, taste, and tacit knowledge; the reproducible middle is going to free. Then defend what you sell:
- Own your substrate — your data and your context, in a form you can pick up and move.
- Make models contestable — swappable engines pointed at your substrate; use any lab, depend on none.
- Keep your customer — whoever owns the demand owns the game.
And charge for the work, never for access — customer-owned substrate, production-priced services;16 leveraging the new rather than automating the old.17
This is the general argument. Forthcoming posts in this series turn it into seller-side and buyer-side playbooks, then work through concrete instances. Disclosure: I build what this argues for — judge it on the logic.
Definitions
- Customer: a relative term — every layer is a customer of its suppliers and a supplier to its own customers. Unqualified, it means the buyer in the pair under discussion; the “customer relationship” the aggregators race for is the far end of the whole chain, where demand aggregates.
- Aggregator: a player that owns demand — the direct customer relationship — and makes suppliers compete for access to it, on its terms. Google to publishers, Amazon to merchants; the model-plus-aggregator is the AI-era form.
- Substrate: the customer-owned foundation a service operates over: data, schemas, code, infrastructure, history, and context.
- Custody: control over that substrate in a way that lets a supplier deny, degrade, or reprice access.
- Access: reading or using the customer’s existing records.
- Production: creating, correcting, validating, or maintaining authoritative state.
- Sovereignty: the customer’s practical ability to keep operating, or appoint a successor, without the supplier’s permission.
- Model contestability: the ability to switch, host, fine-tune, or replace the model without losing the business behavior, evals, and accumulated judgment the firm depends on.
- Off-switch: the power to interrupt the customer’s use of an asset they depend on.
Footnotes
Geoffrey Hinton predicted in 2016 that AI would make radiologists obsolete within five years; radiologist employment rose instead. Part of the job (reading the scan) is a repeatable procedure; the rest — judgment, accountability, the production system around it — is not. The reproducibility line runs through jobs, not just between them.↩︎
“Toward zero” is the generation step, not the delivered cost. Verification, integration, security, accountability, and distribution remain real costs — and they are where the scarce complement, and the margin, relocate. The reproducible portion loses pricing power; the parts that decide, validate, and stand behind the result keep it.↩︎
“Tacit knowledge” is Michael Polanyi’s “we know more than we can tell.” Its industrial form — the accumulated operating know-how that lets TSMC, or a Shenzhen supply chain, build what others can’t from the same blueprints — is Dan Wang’s “process knowledge,” the term the later essays use for the operational asset a vendor actually sells.↩︎
Economists call this the “smiling curve,” after Stan Shih of Acer (1990s): value high at design and at the customer end, low in the manufacturing between.↩︎
Physical work is also safe for now — but only because this flood is cognitive. When robots flood the market for hands, the same logic arrives, so treat that safety as borrowed.↩︎
Ben Thompson, “Aggregation Theory” (Stratechery, 2015) and “Defining Aggregators” (2017): once the internet drove distribution costs to zero, power shifted from controlling supply to owning demand — and aggregators capture demand by owning discovery over an abundance of supply.↩︎
Not everything collapses into one interface. The grab bites hardest on standardized tasks a buyer will delegate, where demand flows through the interface and nothing about the supplier’s brand, regulation, or operational standing pulls the customer back. Regulated, relationship-heavy, operationally embedded buying stays multi-homed. The claim isn’t that every customer relationship moves to the agent interface — only that enough task demand does that vendors whose value is reproducible and whose customers are mediated lose leverage.↩︎
The “hold-up problem” (Klein, Crawford & Alchian 1978; Williamson). Ownership as the fix: Grossman, Hart & Moore; Oliver Hart shared the 2016 Nobel for the idea.↩︎
More precisely, three switching costs. Labor (migration, schema translation, integration rewrites) is what AI attacks first. Custody (the vendor denies, throttles, or reprices access) is what ownership attacks. Operational costs (risk, downtime, compliance, retraining) remain legitimate friction. Exit never becomes free; vendors just can’t hide custody rent inside labor friction and call the whole bundle inevitable.↩︎
Farrell & Gallini (1988): a supplier can raise its own profit by deliberately creating the customer’s escape route. Open-source software runs on the same logic (Lerner & Tirole 2002).↩︎
“Economic moat” is Warren Buffett’s coinage; switching costs were canonized as one of the seven durable powers in Hamilton Helmer’s 7 Powers (2016) and became standard venture diligence.↩︎
DeepSeek, Qwen, and their successors have kept open weights within months of the frontier — the engine of the commoditization the labs are fleeing. For a regulated U.S. buyer they discipline more than they deploy: provenance, security review, and possible restrictions on Chinese models complicate adoption, and continued publication is a publisher’s choice, not a commitment. The structural fact survives either way: weights you hold cannot be taken back.↩︎
Anthropic announced Claude Fable 5 and Claude Mythos 5 on June 9, 2026; its Mythos-class data-retention policy required thirty-day retention for prompts and outputs; its Claude Mythos page described Fable 5 as sharing Mythos 5’s underlying model with additional safeguards; Anthropic’s June 12 statement said it had received a U.S. government directive to suspend access to Fable 5 and Mythos 5 for foreign nationals, leading it to suspend access while implementing controls; and Anthropic’s redeployment statement said the export controls were lifted on June 30, 2026, and access restored July 1.↩︎
Alex Karp on CNBC, discussing Palantir’s partnership with Nvidia to bring the open-source Nemotron model to the U.S. government: enterprises want “control over their compute, their models, their data stack and their alpha — they want to know they own the means of production.” Ben Thompson (Stratechery) on the Fable retention change: Anthropic “didn’t put in any sort of safeguards to guarantee they wouldn’t” train on the retained data — and the labs have “an economic imperative to move up the stack and own the customer directly.”↩︎
How absorption works, spelled out. The aggregator sits at the interface, so it sees the demand for what you do; software production is near-free, so it can rebuild your function natively on the surface the customer already works in — and a vendor whose revenue rests on custody is, almost by definition, one whose function is mostly reproducible. The only thing still holding your customer is their data in your silo, a barrier hostile to the customer, who therefore helps break it: demanding portability at renewal, or letting agents reconstruct the state. Revenue priced on captivity breaks all at once when the wall does. The hand-it-back vendor is spared not out of mercy but because there is nothing to seize: its price was production, its defense is work interface traffic doesn’t teach, and agents compose with it instead of tunneling through it.↩︎
A broad consensus points the same way — usage-, agent-, and outcome-based pricing displacing per-seat — argued by investors (Bessemer, a16z) and tracked by the analyst houses (Gartner, IDC). This essay takes the direction as given and argues about belief, not timing.↩︎
The distinction is Alan Kay’s, by way of Arthur Koestler: a new technology first automates the old form — the horseless carriage, the word processor as a faster typewriter — before someone uses it to reach a plane that wasn’t possible before. Hamilton Ulmer frames the working choice crisply as “automate the old, or leverage the new.”↩︎