Your grocery bill isn’t the only thing creeping up this year. Check the price tag on a new laptop, or your electricity statement, and you’ll see the same pattern. There’s a reason for that, and it isn’t the usual suspects like oil prices or supply chain snags. It’s AI.
Why AI spending is pushing prices higher
The math is almost absurd once you see it laid out. Four companies alone, Alphabet, Amazon, Meta, and Microsoft, are on track to spend around $720 billion this year, and most of that is going straight into data centers. All those servers need memory chips and processors, and there simply aren’t enough to go around. JPMorgan estimates some computer memory chips have gotten up to 400% more expensive between 2024 and the end of this year. That cost doesn’t stay with the tech companies. It shows up in the price of your next laptop.
Total AI-related infrastructure spending is expected to top $700 billion in 2026, and that kind of money moving through an economy at once creates exactly the kind of demand-supply imbalance that pushes prices up. It’s not complicated economics. When everyone wants the same limited pool of chips, servers, and power capacity, prices rise for that pool, and then for everything built on top of it.
The Fed is watching, and they’re not thrilled
Kevin Warsh took over as Fed chair back in May, and he’s been fairly open about his read on this. He thinks that over the long run, AI could make the economy more efficient, which should eventually help bring inflation down. But when he spoke on July 1, he also admitted something less comfortable: AI investment is boosting demand right now, and he wouldn’t commit to how big that inflationary hit might get.
That kind of hedge from a Fed chair tells you something. He’s not dismissing the concern. He’s just not ready to put a number on it yet.
John Williams, who runs the New York Fed and also sits as vice chair on the rate-setting committee, was more direct. He’s flagged that if AI-driven demand keeps outpacing what suppliers can actually deliver, that’s not a blip the Fed can shrug off. It’s the kind of sustained pressure that usually forces a response, meaning higher interest rates to cool things down.
So here’s the tension nobody’s fully resolved: AI is supposed to be the productivity boost that eventually lowers costs across the economy. Right now, though, it’s mostly acting like a demand shock, and demand shocks against limited supply are Econ 101 for rising prices.
This isn’t 2021 all over again, but it’s not nothing either
Let’s be clear about scale. Inflation peaked at 9.1% during 2021 through 2023, and nobody serious is predicting AI spending alone will get us back there. The IMF’s current forecast has global inflation reaching 4.7% this year, up from where it stood in April, before easing to 3.9% next year. The Fund pointed to exactly this kind of tech-driven demand pressure as one reason disinflation, which had been steadily happening since early 2024, has now stalled out.
Dario Perkins, an economist at TSLombard, put it about as plainly as an economist will: right now, AI’s effect on prices is inflationary, not the deflationary story everyone hoped for.
That’s the part that catches people off guard. The pitch for years has been that AI will eventually make everything cheaper and more efficient. It might still do that. But “eventually” is doing a lot of work in that sentence, and in the meantime, the buildout itself is expensive, energy-hungry, and competing for the exact same resources as everyone else.
What this actually means for your wallet
If you’re shopping for a laptop, expect the elevated prices to stick around for a while, not just this quarter. Memory chip shortages don’t resolve overnight, and demand from AI data centers isn’t slowing down. Electricity costs in regions near major data center buildouts are also worth watching closely, since power-hungry AI infrastructure is a real, measurable draw on local grids.
If you’re watching interest rates, pay attention to what the Fed says next. Williams and Warsh are both signaling that if this demand-supply gap doesn’t close, a rate hike later this year becomes more likely, not less. That would ripple into mortgage rates, credit card APRs, and basically anything tied to borrowing costs.
There’s also a longer runway question worth sitting with. Data center construction doesn’t happen on a quarterly timeline. Once a project breaks ground, the demand for chips, cooling systems, and power capacity locks in for years, not months. That means even if AI companies slowed their spending pace tomorrow, the projects already underway would keep pulling on the same limited pool of components well into 2027. Anyone hoping this resolves quickly is probably underestimating how long these commitments already run.
None of this means you should hold off on buying a laptop or panic about your electric bill. It means the “why is everything more expensive again” question has a clearer answer this time than it did during the 2021 to 2023 stretch. Back then it was pandemic snarls and energy shocks. This time it’s a trillion-dollar infrastructure race that hasn’t figured out yet how to pay for itself without passing costs downstream.
I don’t think this is a reason to panic. It is a reason to stop treating “AI is inflationary right now” as some fringe take. It’s coming from Fed officials, the IMF, and economists who track this for a living. The productivity gains everyone’s banking on might show up eventually. Until then, the bill for building the infrastructure is landing on regular consumers, one chip shortage and one electricity statement at a time.
