For two years, the AI skeptics had a pretty solid argument: “Yeah, sure, companies are spending $2 trillion on AI infrastructure, but where’s the revenue?” It was the financial equivalent of building a massive stadium and hoping someone would eventually buy tickets.
Well, someone’s buying tickets now. And the bears just lost their best excuse.
Here’s the thing that just flipped: the AI economy is now generating enough cash to actually cover the cost of all that infrastructure. Not with champagne-popping margins, but the math is working. For every dollar of AI infrastructure that depreciates, roughly $1.19 in revenue is coming in to cover it. A year ago, that ratio was below 1.0. Now it’s above it.
That’s not a small detail. That’s the moment when a “bubble” becomes a “business.”
The numbers are wild when you zoom out. The global AI economy is producing $175 billion in annualized revenue—and that’s excluding chips, ad uplift, and legacy software features. It’s only counting real customer demand. That’s 0.5% of U.S. GDP, which sounds tiny until you realize it’s growing 3x faster than the internet and mobile booms did. The AI economy needed 180 days to add $1 billion in revenue back in 2023. Today? Less than two days. That’s a 90x acceleration.
But here’s where it gets interesting: the bears are pointing to falling token prices as proof the whole thing is collapsing. Cheaper tokens mean lower margins, right? Wrong. This is where the Jevons Paradox enters the chat. When technology gets cheaper, people use more of it. Cheaper tokens unlock more use cases, more agents, more inference, more everything. The very thing the bears are waving as a red flag is actually the accelerant for the next leg of growth.
Think about it: better AI models can do more complex work, which means they consume more tokens. So the price compression that scares the bears is actually creating more demand. It’s like saying “gasoline got cheaper, so nobody will drive anymore.” The opposite happens.
Here’s another thing worth noting: most of the AI revenue today isn’t coming from flashy new business lines. It’s coming from doing the same work faster, cheaper, and better. That’s boring, but it’s also how every major platform shift starts. The internet’s first decade was all about cost reduction and efficiency. Revenue came later—and when it came, it was enormous. AI is following the same playbook.
The macro data is bullish. The micro data—real company revenues, utilization trends, capex payback—is inflecting positively. And yet AI stocks have been choppy and volatile. That combination—improving fundamentals, weak stock prices—is textbook buying opportunity territory.
The infrastructure names best positioned to benefit span the full stack: chips and semiconductors, memory, networking and optics, servers, and power and cooling. These aren’t sexy picks, but they’re the picks that actually make money when the infrastructure buildout accelerates.
For the past two years, the central question was whether AI capex would ever be justified by revenue. This quarter, for the first time, the revenue side pulled ahead. The race isn’t over, but the direction changed. And in markets, direction matters more than destination.
The AI trade is alive. The fundamentals are inflecting. And the market is handing you a discount on one of the most compelling long-term growth stories in history.
