AI Isn’t Free. American Workers Are Picking Up the Huge Tab.

The costs to Americans from AI go beyond lost jobs. (Dreamstime)

Key Points

• The Federal Reserve’s “Beige Book” indicates firms anticipate job cuts due to early generative AI implementation results, especially if demand slows.
• AI investments by megacap tech companies benefit from 100% depreciation under the One Big Beautiful Bill Act, potentially increasing national debt by $3 trillion.
• The market’s top seven AI-committed stocks constitute 40% of the S&P 500, creating significant concentration risk for 70 million 401(k) investors.

The launch of OpenAI’s ChatGPT nearly two years ago wasn’t the first iteration of artificial intelligence technology, but it was without question the most significant.

The chatbot not only fired the starting gun on an AI investment arms race that has attracted trillions in capital and is transforming the economy, it has also changed the public’s perception of what the new technology can achieve.

What it is likely to cost beyond the spending spree now under way is starting to become clear. American taxpayers are paying to have their jobs taken away as their utility bills rise and risks to retirement savings mount.

Now that ChatGPT has a host of competitors, and companies from fast food restaurants to healthcare benefit managers are vowing to deploy AI across their day-to-businesses, the likely scale of potential job losses is coming into focus.

The Federal Reserve’s regular compendium of commentary about economic activity around its various regions, a typically dry report that befits its nickname, the “Beige Book,” included an unusually eye-catching phrase when it was published late Wednesday.

The Atlanta Fed noted that job cut plans “were cited more frequently…and some firms expect further cuts based on early generative AI implementation results. Particularly if demand slows further.”

In essence, company bosses are saying that they need to keep people working right now, but that they will replace them with AI as soon as they can.

“It’s very clear that AI is going to change literally every job,” Walmart’s Doug McMillon told The Wall Street Journal last month. As CEO of the country’s biggest private employer, he might be worth listening to.

But it gets worse. The megacap tech companies are pouring billions into AI investments based not only on the promise of future profits, but also because of the tax efficiency of doing so today. The One Big Beautiful Bill Act allows for the 100% depreciation of new assets and research and development projects.

That is an enormous tax break for the country’s biggest corporations. The Tax Foundation estimates it will increase the national debt by around $3 trillion over the next decade. Who ultimately pays for that is no mystery.

AI projects are also power-consuming monsters. A Bloomberg study found that wholesale electricity prices in Baltimore, which sits near the “data center alley” of northern Virginia, have risen more than 125% over the past five years.

Competition for power from strained grids is likely to push prices even higher. The International Energy Agency, in fact, sees data centers in advanced economies accounting for 20% of the growth in power demand over the next five years.

In other words, your home energy bill is likely to increase as a result of the power and water needed to run, and cool, the thousands of data centers already in place, not to mention the several thousand more that are planned.

Jed Dorsheimer, group head of energy and power technologies at William Blair, calls it “pain at the plug.” Higher costs are “hitting those who can’t afford it the most,” he said in a CNBC interview on Thursday.

An equally critical longer-term risk is the danger the market’s enthusiasm for AI poses to retirement savings. The market’s seven largest stocks, all of them committed to the AI investment thesis, comprise around 40% of the S&P 500.

That means the average passive investor in a 401(k) plan is plowing nearly 40% of his or her savings, around $2,300 a year, into just a handful of stocks, “regardless of whether their outlook is good or bad,” said Torsten Sløk, chief economist at Apollo Global Management.

That creates tremendous concentration risk for the 70 million Americans who invest in 401(k) plans to fund their retirement. If there is an AI bubble, and stocks come crashing to the ground as a result, the pain won’t be distributed evenly.

AI is likely an unstoppable force, and many of the country’s brightest minds are betting heavily on its success. The only question now is what kind of economy it will leave for the rest of us.

Write to Martin Baccardax at martin.baccardax@barrons.com