The same boom promising to make everything cheaper is, for now, making your gadgets and your power bill more expensive.
America’s AI building spree has become an unexpected inflation engine, lifting prices on everything from game consoles to electricity. The reason is brutal demand for the same chips, cables, and power that data centers devour. Five hyperscalers alone will spend $741 billion this year, up nearly 75% from last year.
Picture this. You walk into a store to grab a new console, half-expecting the usual “tech gets cheaper over time” discount. Instead the price went up. So did the one on your phone, and the number at the bottom of your electric bill. None of it is because you suddenly want more stuff. It is because a cluster of windowless buildings in the desert wants the exact same parts you do, and they are willing to pay more.
What Happened
Inflation found a fresh culprit. With tariff drama cooling and gas prices finally easing, the AI build-out has stepped in to keep prices stubbornly high. The Wall Street Journal reports the boom is now nudging up costs across the economy, from smartphones to power.
The numbers behind the surge are eye-watering. Analysts peg this year’s capital spending at five hyperscalers, Alphabet, Amazon, Meta, Microsoft, and Oracle, at $741 billion, according to FactSet. That is up nearly 75% in a single year.
And we are early. Columbia economist Stijn Van Nieuwerburgh estimates total spending on the build-out through 2032 could reach roughly $8 trillion, nearly five times the entire market value of New York City’s property market. The build-out, he notes, is strikingly physical.
The Backstory
Here is the uncomfortable plot twist. The technology everyone expects to eventually crush prices is, in its construction phase, doing the opposite.
Data centers are hungry beasts. They need sophisticated computing gear, cooling systems, electric and fiber-optic cabling, and backup generators. Many of those same components live inside the things you buy. Memory and storage chips, for example, show up in everything from videogame consoles to cars.
So when AI demand spikes, the spillover hits your cart. Nintendo, Microsoft, and Sony have all raised device prices. Apple is next: CEO Tim Cook told the Journal the jump in component costs was unlike anything he had seen in any area in over 40 years. When the man running the world’s most valuable consumer-tech supply chain says that, it is worth a pause.
The Plan, Sort Of
There is a hopeful version of this story, and it is not crazy. If AI delivers the productivity leap its boosters promise, it could eventually cool inflation, the way past technological revolutions did by helping businesses meet demand without hiking prices. Fed Chairman Kevin Warsh made exactly that case in November, calling AI a significant disinflationary force. His read on the boom is shaping up to be the first big test of his Fed leadership.
The catch is timing. AI infrastructure could be built far faster than railroads, electrification, or dot-com-era telecom, but it still needs time to bear fruit. Economists at UBS reckon it will be at least a couple of years before AI starts actually lowering inflation. Until then, the bill comes first and the productivity dividend comes later.
The data is already turning. In a National Association for Business Economics survey released Monday, 81% of economists said the build-out will add to inflation over the next year. Consumer prices for computer software and accessories were up about 15% from a year earlier in May, per the Labor Department, and wholesale electronic components and accessories jumped 27%.
The Business Model Angle
This is a classic input-shock pattern, and founders should study the mechanics. When a single buyer category gets large enough and impatient enough, it stops being a customer and becomes a force of nature in its supply chain. AI labs are now that force.
The strategic tell is in how the demand behaves. Strategists at Evercore ISI point out that tariffs and oil are one-time shocks that wash through prices and fade. AI is different. It is a demand shock that could persist for years. That distinction matters enormously if you build hardware, sell anything chip-dependent, or simply run a business on electricity, which is to say, all of them.
The pattern fits one of the most reliable rules in business: in the first phase of any major technological revolution, scarce resources get strained and prices climb, as EY-Parthenon chief economist Gregory Daco puts it. The companies that win are the ones that lock in supply early, hedge their input costs, and price with the spike in mind rather than pretending it is temporary. If you want to see who actually profits when everyone needs the same chips, study Nvidia’s business model, the toll booth at the center of the whole rush.
The Risk
Now the honest counterweight, because the doom version is overcooked. Nobody serious expects a repeat of the post-Covid inflation spike. Smartphones and videogames are a tiny slice of what households spend. Even electricity is only about 2.5% of consumer spending, per the Labor Department.
The real danger is subtler: not a spike, but a plateau. Most of the announced data-center money has not even been spent yet, Fed governor Lisa Cook noted last month, and looming IPOs from OpenAI and Anthropic could pour in more fuel. Power demand is the slow burn. Goldman Sachs forecasts data centers will drive nearly half of US electricity demand growth through 2030, with consumer power prices rising about 6% a year this year and next. Even chip-stock enthusiasm is jittery: the PHLX Semiconductor Index is up roughly 150% over the past year despite a sharp selloff this week.
The quiet worry is psychological. As Berkeley economist Jon Steinsson warns, the more often prices refuse to fall, the more people start treating it as a pattern and stop expecting relief. With the Fed’s preferred inflation gauge expected to read 4.1% for May, more than double its 2% target, that pattern is exactly what the central bank is trying not to let harden.
Quick Questions
Is AI actually causing inflation?
It is contributing. Tariffs and gas are calming down, but the AI build-out is now pushing prices up on electronics and electricity. In a NABE survey, 81% of economists said it will add to inflation over the next year.
Why are electronics getting more expensive in 2026?
Memory and storage chips that go into consoles, phones, and cars are in fierce demand from data centers. Nintendo, Microsoft, and Sony have raised prices, and Apple’s Tim Cook called the cost jump unlike anything in over 40 years.
Will AI eventually make prices go down?
Possibly, if the productivity boom arrives. Fed Chair Kevin Warsh thinks AI will be disinflationary long-term. But UBS economists say that relief is at least a couple of years away. Construction costs hit first.
How much could data centers raise my electric bill?
Goldman Sachs expects data centers to account for nearly half of US power-demand growth through 2030, with consumer electricity prices climbing roughly 6% a year this year and next.
The Business Model Analyst Take
The lesson here is older than AI: the company building the future usually pays for it up front, and so does everyone standing nearby in the supply chain. For founders, the move is not to wait for the productivity payoff that UBS says is years out. It is to treat input inflation as a structural condition, not a passing storm. Lock in your supply, price for the spike, and watch which buyers in your market have quietly turned into forces of nature. The cheap-tech era is paused, not canceled. The operators who plan around the pause instead of praying it ends fast are the ones who will still have margins when the productivity dividend finally shows up.
Source: The Wall Street Journal
