Why AI Is Not the New Internet — and Why This Is the End of Classical Economics
Most people see AI as the next internet. That is a category error. The internet changed the channel through which humans do business; AI changes the economic actor itself. All of economics since Adam Smith rests on one silent assumption — and for the first time, it is no longer mandatory.
When I talk to business owners about artificial intelligence, sooner or later one phrase almost always comes up: “Well, it’s just like the internet in the nineties — whoever moves first, wins.” I understand why this analogy is appealing. It is comforting. It means we have been through this before and know what to do.
I think it is a mistake. And not a small one — it is a category error, the kind that makes you look in the right direction through the wrong glasses.
Why the Internet Is a Bad Analogy
Consider what previous technology revolutions actually changed. The steam engine changed production — how humans make things. The internet changed the distribution channel — how humans find a customer and deliver value to them. In both cases, what changed was what humans do business with. But the grammar of economics itself remained untouched: there was still a firm, a product, a customer, and a brand. The internet did not make a single one of these concepts obsolete — it merely moved them online.
AI does not change the channel. AI changes the economic actor itself — the who that conducts economic activity in the first place.
There is an important nuance here that is often misunderstood. AI is not “the same as a human, just with superhuman abilities” — it is not the same creature with a faster processor. AI is a different dimension: it has different constraints than a human, but also entirely different possibilities.
And this brings us to something deeper. All of economic science since Adam Smith’s “Wealth of Nations” (1776) rests on one silent axiom that almost nobody ever says out loud: the agent of economic activity is a human, with human limitations. For the first time in history, this axiom is no longer mandatory. And if it falls, far more will fall with it than we think.
The Pillars of Classical Economics Are Theorems About Humans, Not Laws of Nature
Here is the thought I find most important. We treat the foundational principles of classical economics as laws of nature — something as immutable as gravity. But they are not. They are theorems about human limitations. Each of them quietly assumes something about what a human can and cannot do, what it costs, and how long it takes. Change the actor — and the theorem no longer holds.
Let us walk through six pillars in turn.
| Classical pillar | Hidden assumption about humans | What happens when the actor is AI |
|---|---|---|
| Scarcity as the foundation of the discipline | The most expensive scarcity is human thinking and time | Thinking becomes nearly free and infinitely copyable — the central scarcity moves elsewhere |
| The firm as an organization (Coase, 1937) | Market coordination is expensive, so an organization is needed | Transaction costs approach zero; the firm’s boundary dissolves into a cloud of contracts and protocols |
| Information asymmetry as a source of profit | Knowing more than the other party is expensive and slow | Verification happens in milliseconds; asymmetry-based margins collapse — only speed remains |
| Price as a carrier of information (Hayek, 1945) | Nobody can compute dispersed knowledge | Demand can be read directly from behavioral data, bypassing the price signal |
| Marketing and brand | The buyer is a human with cognitive biases | In a machine-to-machine (M2M) market there is no psychology to address — the industry that profits from biases loses its object |
| Equilibrium models | Markets adjust slowly, at a human pace | Continuous micro-arbitrage at machine speed — the economy starts to resemble ecology, not mechanics |
Let me expand each row, because the devil lives precisely here.
Scarcity. All of economics begins with the premise that resources are limited and must therefore be allocated with discipline. But the truth rarely spoken aloud is this: for the past two hundred years, the most expensive scarcity in most industries has not been raw material but human mental work — the capacity to analyze, decide, coordinate. When that work becomes nearly free, the scarcity moves elsewhere. And that changes everything, because an economy always organizes itself around whatever is rarest.
The firm. In 1937, Ronald Coase asked the question that would later earn him a Nobel Prize: why do companies exist at all? If the market is so efficient, why not organize every task as a separate transaction? The answer: because market coordination costs money — finding a partner, negotiating, drafting a contract, monitoring delivery. The firm exists to hide these transaction costs under one roof. When the actor is AI, those costs approach zero — and the boundary of the firm, that clear line where “we” ends and “the market” begins, starts dissolving into a cloud of contracts and protocols.
Information asymmetry. A large share of profit arises simply because one party knows more than the other: the broker knows a price the client doesn’t; the insurer knows the risk better than the customer. This is profitable because acquiring information is expensive and slow. When verification happens in milliseconds and anyone can check any claim in real time, asymmetry-based profit collapses. What remains is speed — who gets there first.
Price as a carrier of information. In 1945, Friedrich Hayek wrote what I consider one of the most important economics texts of all: no individual and no planning authority can gather all the dispersed knowledge about what is needed, where, and in what quantity. That is why price is an ingenious invention — it compresses the knowledge of millions of people into a single number. But notice: price is a substitute, needed precisely because nobody can compute demand directly. A system that reads demand directly from behavioral data can partially bypass the price signal.
Marketing and brand. The entire marketing industry rests on one assumption: the buyer is a human with emotions and cognitive biases that can be addressed. We pay for brands because they create a feeling of trust and status. But in a machine-to-machine market, where the buyer is an algorithm, there is no psychology to address. An industry that has spent decades profiting from human cognitive biases is left without its object.
Equilibrium. Classical models assume that markets move toward equilibrium — that price “finds” the place where supply meets demand. This assumption hides a human tempo: markets adjust slowly because humans think and act slowly. When participants act at machine speed, continuous micro-arbitrage takes over and the system never settles into equilibrium. It resembles ecology — a constant struggle for viability — more than clean mechanics.
It Ends the Way Newtonian Physics Ended
Reading this, you might get the impression that I am predicting some crash or catastrophe. I am not. I think classical economics will not end in collapse — it will end exactly the way Newtonian physics ended.
Newton’s laws did not become wrong when relativity and quantum mechanics arrived. Bridges built on Newton still stand. Something subtler happened: Newtonian physics became a special case — correct under certain conditions (low speeds, large objects), but no longer a description of all of physics. We all learned this special case in school, believing it was the whole of physics.
I expect economics to go through exactly the same thing. Classical economics will become a special case — “economics under the condition that all participants are human.” It will not become wrong. It will continue to describe human markets accurately. But it will no longer be the whole picture, because for the first time there will be market participants to whom its hidden assumptions do not apply.
The Counterweight: Scarcity Does Not Disappear — It Moves
Here I must say something that keeps this article from becoming a utopia. It is easy to read the above and conclude: “Great, so everything becomes free and wealth becomes infinite.” It does not.
The fundamental law still holds: scarcity does not disappear, it only moves. When thinking becomes cheap, rarity shifts to what remains limited:
- Physical resources and energy. Data centers, chips, electricity — all of it is real and finite, and AI’s consumption makes it scarcer still.
- Legal accountability. A signature, a license, responsibility before the law — these cannot be delegated to an algorithm. Some person or legal entity will always stand behind a transaction.
- Human trust and attention. Precisely because machine-made content becomes infinite, human trust and human attention become rarer and more valuable.
The new economics will be a science of exactly these scarcities. And the human role in it will not disappear — it will move to where true rarity remains: oversight. Humans will set goals and ethical boundaries and bear responsibility wherever the system meets the human world. The machine will search and execute; the human will answer and decide.
What to Do With This Now
I am not writing this as a philosophical exercise. It has practical consequences for how you think about your company today:
- Check which of your profit positions rest on information asymmetry. If you earn because the customer doesn’t know something, ask: how long will they not know it, once verification becomes instant?
- Ask how much of your brand’s value is addressed to human emotions. That is real value — but it holds only as long as the buyer is human.
- Think about where systemic inefficiency lives in your industry — idle resources, disconnected data. That is exactly where the new type of profit will emerge.
Classical economics will still serve us — just as Newton still serves the engineer. But it is a map of one territory. A second one is opening up, and it has no maps yet.
This is the second article in a series on AI-native business models. In the first, I asked a latest-generation AI model what profit-making would look like without the constraints of human thinking. In the next article — the full map of human constraints: eleven principles as a practical framework for testing your business model.