Everybody wants to know “who wins” AI.
OpenAI. Anthropic. Google. xAI. Maybe it’s a company nobody’s even heard of yet.
I don’t know.
More importantly, I don’t think that’s the most interesting question.
I’m certainly not an AI expert. I know enough to be dangerous, which is usually the most dangerous amount.
What I do know is markets. I’ve spent most of my adult life studying booms, busts, bubbles, and technological shifts. I’ve had a front-row seat to some of them.
And one lesson keeps showing up over and over again: The biggest opportunities are usually one or two steps removed from the thing everybody’s focused on.
That’s why I’ve been thinking a lot about what people are calling the “agentic economy.”
An “AI agent” is an autonomous software program that can be designed to perform specific tasks. The idea is that businesses and individuals are increasingly building AI agents that perform tasks on their behalf.
Some will be useful. Some will be terrible. Some will save enormous amounts of time. Others will disappear as quickly as they arrived.
That’s how every technological revolution works.
But let’s assume for a minute that this trend continues. Let’s assume that a few years from now businesses are relying on AI agents every day.
Consumers are relying on AI agents every day. Software is communicating with software. Agents are working alongside other agents.
What happens then?
At some point, these agents are going to start transacting with one another. They’ll exchange information, perform services, negotiate tasks, and eventually transfer value.
The first question everyone asks is: What currency will they use? Bitcoin? Stablecoins? Some token that hasn’t been invented yet?
I think that’s the wrong question.
The Rails
The way I see it, everybody’s arguing over what the trucks will be carrying while ignoring the interstate highway system being built underneath them.
Because whether these agents pay each other in Bitcoin, dollars, stablecoins, bottle caps, or Chuck E. Cheese tokens, they’re all going to need the same things.
They’re going to need computing power. They’re going to need storage. They’re going to need networking. They’re going to need electricity. They’re going to need data centers.
They’re going to need all the infrastructure that makes the entire ecosystem function.
The AI evangelists are selling a very specific dream. The barriers are coming down. Building software is getting easier. Creating businesses is getting easier. Automating workflows is getting easier.
The promise is that you’ll be able to build what you want, create what you want, automate what you want.
As Biggie put it, “The sky is the limit and you know that you can have what you want, be what you want.”
If they’re right, we’re heading toward a world where billions of software agents are working on behalf of businesses and individuals.
Some will be creating content. Some will be managing schedules. Some will be writing code. Some will be buying and selling goods and services.
Some will be doing things we haven’t even imagined yet.
And every single one of them will consume resources. That’s the part I keep coming back to.
It doesn’t actually matter who wins.
OpenAI can win. Anthropic can win. Google can win. All of them can win. None of them can win.
The trend remains the same.
Follow the Demand
The cost of intelligence is collapsing.
Just as the Industrial Revolution dramatically reduced the cost of physical labor, artificial intelligence is reducing the cost of cognitive labor.
That’s a profound shift.
Whether the technology ultimately exceeds expectations or falls short of them, the direction seems pretty clear.
The more intelligence gets embedded into software, the more demand there will be for the infrastructure that powers it.
That’s why I keep focusing on the chips, the power generation, the networking equipment, the data centers, and the companies building the digital highways that all of these agents will travel across.
And if you really want to take it one step further, where do all those things come from?
You don’t build data centers out of thin air. You need copper. You need rare earths. You need steel. You need energy.
The further down the supply chain you go, the more this starts looking less like a software story and more like an industrial boom.
Markets have a way of rewarding investors who can identify the second- and third-order effects before everyone else.
Sometimes those opportunities come from technological shifts. Sometimes they come from policy shifts.
Either way, the exercise is the same: Follow the money and follow the incentives.
When I zoom out and look at what’s happening, I keep arriving at the same conclusion.
The opportunity may not be in figuring out which AI application wins. It may not be in guessing which token agents use to pay one another.
The bigger opportunity might be owning the picks and axes being sold to participants in the rush.
Everybody wants to know “which AI” wins. It’s really what I’ve been calling “AI mining” all along.
I’m more interested in who gets paid regardless.
Stay sharp,
JC Parets, CMT
Founder, TrendLabs