Is artificial intelligence blowing up our economy like we were sold into?
Yes — but “AI is blowing up the economy” is a little too simple. The more accurate story is that the AI boom is creating a huge investment bubble around the technology, while the promised economy-wide productivity revolution has not arrived yet.
And that distinction matters.
The AI economy is running on enormous spending
The biggest technology companies are pouring extraordinary amounts of money into chips, data centers, electricity, networking and computing capacity. The Minneapolis Federal Reserve estimates that capital spending by just five major AI investors — Alphabet, Amazon, Meta, Microsoft and Oracle — could approach $1 trillion by 2027, compared with roughly $5.5 trillion in total annual private investment in the U.S. economy.
Meanwhile, S&P Global has raised concerns about the financial structure supporting this buildout, particularly the increasing use of debt, leases and other financing arrangements when the eventual return on all that investment remains uncertain.
That’s where the “bubble” argument gets interesting.
We’re spending first and getting the productivity later
The Federal Reserve itself essentially describes the process as:
AI capability → business adoption → massive investment → productivity gains → labor-market effects.
The problem is that we’re very far along the investment track, while the economy-wide productivity payoff is still comparatively modest.
U.S. worker productivity increased at a 1.4% annualized rate in Q2 2026, up from 0.8% in Q1. That’s positive, but it isn’t remotely the economic explosion that some AI evangelists predicted.
And there’s another wrinkle: today’s AI spending itself is helping prop up economic activity.
We’re building data centers.
We’re buying GPUs.
We’re constructing power infrastructure.
We’re hiring construction workers.
We’re importing enormous quantities of computing equipment.
We’re borrowing money to finance it.
All of that counts as economic activity.
But building $100 billion worth of infrastructure doesn’t automatically mean that infrastructure will eventually generate $100 billion worth of additional economic value.
That’s the bet.
And we’re starting to see some weird consequences
The electricity issue is becoming particularly striking.
Texas recently halted some new data-center grid connections while regulators investigated enormous electricity-demand requests. Some proposed demand figures were so large that officials described the problem as “ghost demand” — projects being proposed that may never actually materialize.
That’s a pretty good metaphor for the AI economy:
Everybody is planning for the economy that AI might create.
The question is whether that economy actually shows up.
But here’s the important part: AI isn’t currently destroying the U.S. economy
Today’s labor market actually makes that argument difficult to sustain.
The September 4, 2026 jobs report showed 162,000 jobs added in August and unemployment at 4.1%.
There are, however, signs of AI-related disruption underneath those headline numbers. Information and financial-sector employment has weakened, and AI adoption is being cited as one factor behind employment reductions in areas such as telecommunications and customer service.
So we’re seeing something more complicated than:
AI takes everybody’s job → economy collapses.
It’s closer to:
AI eliminates some work, creates new investment and infrastructure demand, changes how other jobs are performed, and transfers enormous amounts of capital toward a relatively small number of companies.
That’s a much bigger economic story.
So were we “sold” an AI economic miracle?
To some extent, yes.
The sales pitch was basically:
AI will make workers dramatically more productive → companies become more profitable → wages rise → prices fall → economic growth accelerates.
The reality so far is considerably messier.
AI investment is already economically significant, but the Federal Reserve notes that aggregate private investment as a share of GDP has not dramatically departed from its longer-term trend.
And researchers are increasingly worried about the possibility that investment and demand arrive before productivity gains do. That can actually push up inflation and interest rates rather than immediately bringing prices down.
That’s almost the exact opposite of the original sales pitch.
My take
I wouldn’t call AI a scam.
I would call the current AI economy an enormous economic bet.
The technology is real. The productivity potential is real. The infrastructure being built is very real.
But investors, corporations and governments are behaving as though the enormous productivity payoff is already guaranteed.
It isn’t.
And that’s where things could get ugly.
If AI produces spectacular productivity gains, today’s insane spending could eventually look brilliant.
If the productivity gains disappoint, we could discover that we spent trillions of dollars building the world’s most expensive collection of very intelligent air conditioners.
The danger isn’t AI itself.
The danger is an investment cycle built on expectations that become detached from what the technology actually earns.
And right now, that’s absolutely something economists and the Federal Reserve are watching.
For AI and economic impact, here’s one digital transformation option
Is artificial intelligence blowing up our economy like we were sold into?
Yes — but “AI is blowing up the economy” is a little too simple. The more accurate story is that the AI boom is creating a huge investment bubble around the technology, while the promised economy-wide productivity revolution has not arrived yet.
And that distinction matters.
The AI economy is running on enormous spending
The biggest technology companies are pouring extraordinary amounts of money into chips, data centers, electricity, networking and computing capacity. The Minneapolis Federal Reserve estimates that capital spending by just five major AI investors — Alphabet, Amazon, Meta, Microsoft and Oracle — could approach $1 trillion by 2027, compared with roughly $5.5 trillion in total annual private investment in the U.S. economy.
Meanwhile, S&P Global has raised concerns about the financial structure supporting this buildout, particularly the increasing use of debt, leases and other financing arrangements when the eventual return on all that investment remains uncertain.
That’s where the “bubble” argument gets interesting.
We’re spending first and getting the productivity later
The Federal Reserve itself essentially describes the process as:
AI capability → business adoption → massive investment → productivity gains → labor-market effects.
The problem is that we’re very far along the investment track, while the economy-wide productivity payoff is still comparatively modest.
U.S. worker productivity increased at a 1.4% annualized rate in Q2 2026, up from 0.8% in Q1. That’s positive, but it isn’t remotely the economic explosion that some AI evangelists predicted.
And there’s another wrinkle: today’s AI spending itself is helping prop up economic activity.
We’re building data centers.
We’re buying GPUs.
We’re constructing power infrastructure.
We’re hiring construction workers.
We’re importing enormous quantities of computing equipment.
We’re borrowing money to finance it.
All of that counts as economic activity.
But building $100 billion worth of infrastructure doesn’t automatically mean that infrastructure will eventually generate $100 billion worth of additional economic value.
That’s the bet.
And we’re starting to see some weird consequences
The electricity issue is becoming particularly striking.
Texas recently halted some new data-center grid connections while regulators investigated enormous electricity-demand requests. Some proposed demand figures were so large that officials described the problem as “ghost demand” — projects being proposed that may never actually materialize.
That’s a pretty good metaphor for the AI economy:
Everybody is planning for the economy that AI might create.
The question is whether that economy actually shows up.
But here’s the important part: AI isn’t currently destroying the U.S. economy
Today’s labor market actually makes that argument difficult to sustain.
The September 4, 2026 jobs report showed 162,000 jobs added in August and unemployment at 4.1%.
There are, however, signs of AI-related disruption underneath those headline numbers. Information and financial-sector employment has weakened, and AI adoption is being cited as one factor behind employment reductions in areas such as telecommunications and customer service.
So we’re seeing something more complicated than:
AI takes everybody’s job → economy collapses.
It’s closer to:
AI eliminates some work, creates new investment and infrastructure demand, changes how other jobs are performed, and transfers enormous amounts of capital toward a relatively small number of companies.
That’s a much bigger economic story.
So were we “sold” an AI economic miracle?
To some extent, yes.
The sales pitch was basically:
AI will make workers dramatically more productive → companies become more profitable → wages rise → prices fall → economic growth accelerates.
The reality so far is considerably messier.
AI investment is already economically significant, but the Federal Reserve notes that aggregate private investment as a share of GDP has not dramatically departed from its longer-term trend.
And researchers are increasingly worried about the possibility that investment and demand arrive before productivity gains do. That can actually push up inflation and interest rates rather than immediately bringing prices down.
That’s almost the exact opposite of the original sales pitch.
My take
I wouldn’t call AI a scam.
I would call the current AI economy an enormous economic bet.
The technology is real. The productivity potential is real. The infrastructure being built is very real.
But investors, corporations and governments are behaving as though the enormous productivity payoff is already guaranteed.
It isn’t.
And that’s where things could get ugly.
If AI produces spectacular productivity gains, today’s insane spending could eventually look brilliant.
If the productivity gains disappoint, we could discover that we spent trillions of dollars building the world’s most expensive collection of very intelligent air conditioners.
The danger isn’t AI itself.
The danger is an investment cycle built on expectations that become detached from what the technology actually earns.
And right now, that’s absolutely something economists and the Federal Reserve are watching.