Jerome Powell says the data center buildout may add inflation pressure

Jerome Powell flags a new pressure

Fed Chair Jerome Powell said the AI buildout is not yet a near-term disinflation story. He argued that the rush to build data centers is increasing demand for the goods and services needed to construct them, which can push inflation up at the margin before broader productivity gains show up.

That matters because many people hear that artificial intelligence should make life cheaper and faster. Powell’s point was simple: before AI delivers broad savings, the country first has to pay for a huge physical buildout.

Why the buildout matters first

Jerome Powell pushed back on the idea that AI should already be dragging inflation lower. He said the near-term story is not lower prices but stronger demand for construction, equipment, and power tied to data centre expansion.

In other words, the payoff may come later than people expect. The country is still in the expensive setup phase, where building capacity comes before broad productivity gains show up in everyday prices.

What the Fed is watching now

At the same press conference, Jerome Powell answered a question about the Fed’s longer-run growth estimate, which had moved from 1.8% to 2.0%. He said that reflects stronger productivity, but he also warned that AI’s near-term impact is still hard to measure.

He added that AI may raise the neutral rate in the short run rather than lower it. That is a big reason he did not frame the technology boom as an immediate case for cheaper borrowing costs.

Why concrete and steel matter

A data centre is not just software living in the cloud. It is a real building filled with servers, cooling systems, backup equipment, wiring, transformers, and other heavy infrastructure that all have to be made, shipped, and installed.

That helps explain Powell’s inflation point. When many projects move at once, they can increase demand for labor, materials, and power equipment, keeping costs firm even before AI boosts productivity across the wider economy.

Fun fact: EIA estimated computing used about 8% of commercial sector electricity consumption in 2024.

The grid is part of the story

The pressure is not only on building materials. Power is a huge part of the equation, and utilities across the country are dealing with rising demand, in part tied to large computing facilities and new data centre development.

That helps connect Powell’s comments to household concerns. When the grid needs upgrades, new generation, or extra transmission support, those costs can ripple outward long before consumers see any broad AI dividend.

Why higher bills feel personal

Analysts expect data center electricity demand to rise sharply this decade, which can increase the need for new generation, transmission, and grid equipment. Goldman Sachs Research, for example, projects a large surge in data center power demand by 2030 compared with 2023 levels.

Separately, utility filings show how those pressures can reach households. PowerLines’ year-end tracking found electric and gas utilities requested nearly $31 billion in rate increases in 2025, more than double the prior year, a sign of how quickly cost pressures are building in many regions.

Even booming demand has limits

One of the most interesting twists is that data centre development is starting to slow in some places, not because interest has disappeared, but because the power system is struggling to keep up with requests.

Wood Mackenzie said only about a third of the U.S. disclosed data center pipeline was under active development at the end of 2025. That shows how demand can stay hot even while actual construction hits bottlenecks.

Fun fact: Wood Mackenzie said the U.S. data center project pipeline reached 241 gigawatts by the end of 2025.

Productivity is real but delayed

Powell did not dismiss the upside of AI. He said the Fed has already been seeing productivity growth that is meaningfully higher for years, and he expects generative AI to contribute over time as those effects become clearer.

But he was careful not to oversell the timing. His message was that productivity gains can be real, but they still arrive too slowly to offset the near-term inflation that comes from building the systems behind them.

Demand may be outrunning supply

Powell boiled the debate down to one basic question: Is demand growing faster or slower than supply? He answered that nobody really knows yet, which makes this a tricky moment for markets, policymakers, and families alike.

That uncertainty matters because AI optimism can sound very clean in theory. In practice, the timing of construction, power availability, and productivity gains may not line up neatly, and that mismatch can keep price pressure alive for longer.

Local strain can become national

A single large data centre project may be local, but the broader buildout can have national effects. More demand for transformers, substations, transmission support, and generation can create competition for scarce equipment and longer wait times.

That is one reason this story reaches beyond tech hubs. Even people far from major projects can feel the impact if the national grid, supply chains, and rate cases start adjusting around a much larger power appetite.

Why the Fed is staying cautious

The Fed held rates steady, and Powell’s comments help explain why he is not treating AI as a reason to rush toward easier policy. In his view, the buildout phase could add inflationary pressure before any broad efficiency payoff materializes.

That does not mean AI is bad for growth. It means the central bank still has to deal with the world as it is now, where demand tied to physical expansion may be arriving faster than the savings people hope to see.

What people should watch next

The next clues will likely show up in power demand, utility rate cases, project delays, and whether productivity keeps running strong. Those are the real-world signs that can show whether AI is easing inflation or adding to it for a while longer.

For regular households, the question is straightforward. If electricity and related costs keep climbing before AI savings spread widely, Powell’s warning may end up sounding less like a theory and more like daily life.

If you want to see how the data center boom is starting to reshape daily life far from Silicon Valley, the related story explains why California desert communities express concern as data center development grows.

The AI boom may cost first

The big takeaway is not that AI has failed. America may be living through the expensive first chapter, where massive investments in data centers and power systems come before the broader benefits reach consumers.

That makes Powell’s remarks feel especially timely. People hoping for lower prices enabled by smarter technology may have to wait as the country builds the infrastructure needed to make that future possible.

If you want to see how that infrastructure strain is already turning into a political fight, the related story explains why Data centers gobble up Washington’s electricity and a bill to stop it just failed.

Do you think data centers are really pushing inflation and utility bills higher, or is something else driving the squeeze? Share your thoughts and drop a comment.

This slideshow was made with AI assistance and human editing.

Read More From This Brand:

• Higher electricity bills spark calls for U.S. data centers to share the cost
• Texans could see higher energy costs as data centers push the grid to the limit
• Behind the surge in data centers, Silicon Valley is creating a new power network across the U.S.