Everyone's asking the wrong question.
Will AI replace us? Will jobs disappear? Will new ones be created? The truth is that nobody really knows — and anyone who claims otherwise is probably trying to sell you something.
But there's a far more interesting question, one that touches the roots of how economies actually work. Not "which job will disappear," but what happens to the structure of markets when AI enters the picture.
And the answer, in my view, starts with one word: arbitrage.
Every Transaction Is a Match
At the base of every market in the world sits a gap between seller and buyer. In a perfectly efficient theoretical world, the price accurately reflects value — and there's no profit margin for anyone. But the world is not theoretical.
Due to shipping costs, information gaps, monopolistic markets, and regulations — layers of middlemen have formed. They don't exist because they're bad. They exist because they solve a real problem: they enable transactions to happen in places where, without them, they wouldn't.
Take digital advertising as an example. Between a brand that wants to buy and a person browsing the internet, there's a complete chain: platforms holding "digital real estate," ad exchanges competing over screen time, agencies employing people who manage campaigns, and analytics tools measuring results. Every link in the chain takes a cut. By the time money gets from the brand to the consumer, it passes through five or six hands.
And that's just one example. The same thing happens in real estate, in recruiting, in supply chains, and in dozens of other markets.
Every mediation layer is the product of inefficiency and it adds up to the overall transaction cost.
AI as a Universal Arbitrage Engine
Now imagine an actor connected to every existing information source, that knows how to process that information instantly, and that also has the ability to execute actions — to make a purchase, to order services, to negotiate, to give and take.
This isn't futuristic. It's already happening. AI agents are already holding digital wallets, connecting to APIs, and executing transactions. They can access any piece of information, compare prices, and execute deals — all at a cost approaching zero.
And just like in the capital markets, where market makers identify price gaps between two identical assets in two different markets — buying cheap, selling dear, and closing the gap — AI does the same thing, but in every market. Every place where there's a gap between what something costs and what it's actually worth, AI enters and closes it.
The result? Mediation layers dissolve.
The brand doesn't need an agency — the AI manages the campaigns. The buyer doesn't need a broker — the AI finds the property, inspects it, and handles the negotiation. The company doesn't need HR — the AI filters candidates and matches interviews.
The Deflationary Pressure
This efficiency inevitably leads to price drops. Because if one player manages to deliver a service at a significantly lower price using AI, all other players in the market must match — or exit the game.
This isn't theory. You can already see it in SaaS markets. Shares of software companies are falling dramatically — not because they're bad companies, but because the market is pricing in a future where a large portion of the services they provide will cost much less, or will be entirely redundant.
But the drop in prices is just the beginning. The pressure manifests at three levels:
Three Levels of Deflation
Service and product prices. Everything that can be made more efficient — will be. And what becomes efficient — gets cheaper.
The cost of human labor. If in the past competition was with cheaper workers in developing countries, today the competition is with an AI agent that does the same work at a fraction of the cost.
Asset prices. When people earn less, they consume less and invest less. Demand for assets — real estate, equities, risk investments — weakens.
The Historical Context
To understand the implications, we need to look backward.
With the advent of the internet and technology, an enormous incentive was created to acquire hi-tech skills. Technology brought opportunities to create new value with tools that didn't exist before. This attracted investors, who attracted talent, who created more technologies, who attracted more investors — a flywheel that fed itself.
But in parallel to the growth of this sector, the gaps between classes widened. Between those who engaged with technology and those who remained in traditional occupations, a massive gap opened — in wages, in wealth, and in quality of life.
This was amplified by decades of expansive monetary policy: zero interest rates and brand expansions that injected money into the system. This money didn't truly translate into wage growth. It flowed to assets — real estate and capital markets that rose in prices. Those who held assets got richer. Those who didn't — fell behind.
Is This Time Different?
The historical argument is comforting: with every previous technology wave, destruction and creation happened in parallel. The car eliminated the power of the horse, then value moved to the road. The computer eliminated calculations, then value moved to design and manufacturing.
And there's an additional argument: we're simply not good at predicting what the new jobs will be. In 1995 nobody imagined that people would make a living from content creation or community management.
With every previous technology wave, technology eliminated one human advantage — but left the rest. AI is the first technology that attacks all floors simultaneously.
AI is the first technology that attacks all floors simultaneously. It's not just a computer — it understands, creates, judges, communicates, and learns. And for the first time, there's no clear floor it can be excluded from.
Two Futures
Possibility one — the last floor. Maybe there's something in the human experience, in taste, in will, in identity — that AI cannot and will not know how to replicate. Maybe the next floor is not cognitive but existential. Maybe value will move to places we can't yet see.
Possibility two — the end of the model. Maybe for the first time in history there is no next floor. Maybe the economic model we've known for 250 years — where productivity is tied to human labor — simply needs to change its foundation.
In both cases, we're at a turning point. Not because AI is smart — but because it makes every market more efficient. And when all the markets become efficient, the economic structure that was built on the inefficiency gap — needs to ask itself what it is when it grows up.
This is not a prediction. This is an invitation to think.