Amazon, Microsoft, Meta and Alphabet spending projections for artificial intelligence infrastructure will boggle your mind.

The American economy hasn’t seen anything like this since the mid-19th-century railroad mania. And this might be even bigger than that.

So this had better work out.

The smartest people in the world think it will. Why? Because these corporations are laying out trillions of dollars to make AI a reality in everyday personal, business, government and military life. Would they spend that kind of money unless they were sure? Investors approve: Technology stocks are soaring. Hedge funds, pension plans and mom-and-pop investors can’t get enough of them.

But what if everybody is wrong?

AI companies are coming to dominate America’s financial markets and, increasingly, our economy. They have become the most highly valued corporations in the world. Nvidia, a company hardly anyone knew existed five years ago, is worth $5 trillion. It is the most valuable in the world.

So if things don’t work out the way everyone hopes, there won’t be many places to hide.

Ten years ago, semiconductor companies made up about 2 percent of the S&P 500. Today they account for roughly 18 percent—more than twice their weighting at the peak of the dot-com bubble in early 2000. After that bubble burst, trillions of dollars in market value were wiped out as technology stocks collapsed over the next two years.

That 18 percent doesn’t even include the tech hardware companies that have also gone parabolic in value. Or the industrial names like Caterpillar that have inflated in value because companies are buying more of their products to build the data centers. Then there are the utilities companies that are preparing to supply all the electricity.

“If you start going through all of these indexes, there is a fascinating difference today from what happened in the tech bubble peak,” says Cameron Dawson, chief investment officer at NewEdge Wealth. When that bubble popped, “you could hide out in emerging markets, value, smallcap, midcap, all indexes that effectively didn’t have the same exposure to the tech bubble … that is very different this time around.”

As Cameron notes on the Thoughtful Money Podcast with Adam Taggart, the AI trade has “infiltrated all of the indexes,” and that means it is much harder for investors to stay away from these investments even if they badly want to.

If you have a 401k, an ira or a pension, it is almost guaranteed you are an AI investor. And probably a big one.

So if this is a bubble and it explodes, there is no avoiding the shock wave.

American technology companies are set to spend over $800 billion on AI infrastructure in 2026 alone. That is about 3 percent of America’s total gdp. In one year.

They plan on spending an additional $1 trillion next year. And billions more the year after.

America’s economy grew at a 1.6 percent rate this last quarter. Take away the AI spending and our economy would have actually shrank: We would be in a recession.

Yet America’s economic reliance on AI spending might be even greater than the headline numbers suggest. That is because a lot of related infrastructure is being constructed to support the AI build-out.

Consider this statistic. For the past 20 years, America’s electricity consumption has been essentially flat. Now we are adding the equivalent of a New York City every three months, says Robert Thummel, managing director at Tortoise Capital.

Consulting firm icf says that America’s electricity demand will actually grow 25 percent by 2030. Forbes calls it an “energy frenzy.” And it is being driven largely by AI data center consumption.

The electricity utilities know this, and they are building like crazy. That means more cement, copper, silver, neodymium and a host of other industrial commodities will be mined and processed. It means more natural gas, uranium, solar panels and wind turbines will be manufactured. It means more truck drivers, crane operators, electricians, architects and welders will be needed, and all the picks, shovels and sandwiches necessary to support them.

“This is the AI boom rendered in steel and gravel,” writes cnbc. “It is the physical manifestation of a belief—that intelligence itself can be manufactured at industrial scale, and that whoever builds the biggest factory wins. Sameer Dholakia, a partner at Bessemer Venture Partners says this AI build-out has become ‘the largest market in the history of mankind.’”

The last time so much money flooded into new “transformative” infrastructure was during the railroad boom of the 1870s, says Pulitzer Prize winner Liaquat Ahamed. That boom led to an epic bust: It caused unemployment for millions of people, a stock market meltdown, two depressions and every single railroad but one to go out of business. “I am genuinely frightened,” says Ahamed.

During the late 1800s, railroads revolutionized the world as people knew it—kind of like what AI promises to do today. Railroads turned America into a truly connected, dual-ocean nation. They reduced transportation times and costs. They linked vast swaths of American geography and resources to coastal factories and markets. The funding of the railroads revolutionized finance and popularized stock investing.

The railroads helped turn America into a superpower. But they also brought a lot of debt into America’s economy. And as Ahamed noted, when railroads didn’t earn the promised returns, investors lost their investments, and a series of economic depressions racked the nation for decades.

When the railroad bubble popped, it resulted in 20 to 25 percent of America’s workforce losing their jobs.

Could America’s AI build-out lead to a similar outcome?

“The railroads were built with debt. Debt, debt, debt,” says Derek Thompson, who hosts the Plain English podcast. “The whole thing was a tottering Jenga tower of leverage, and it came crashing down.” But before the crash, that financial model was considered to be the genius of the railroad builders: They were building a transformative network, and they were even doing it with other people’s money.

Some people say that is a big difference in the AI build-out today.

It is true that the Anthropics, Oracles and OpenAIs of the world have promised to fund these massive capital expenditures out of their profits. But that is changing fast.

In January, the Bank for International Settlements warned that the AI build-out has transitioned from internal cash flows to debt-funded growth, and that is “not only reshaping corporate balance sheets, but also raises important questions about credit standards and financial stability.”

Bloomberg reported on May 19 that borrowing to finance the data center boom was quickly rising up Wall Street’s list of potential credit threats. cnbc said it has become a “historic borrowing binge.”

America’s tech behemoths are increasingly relying on borrowing to fund their growth plans. It is what the home builders and lenders did in 2007, the tech companies did in the late 1990s, the investment trusts did before the Great Depression, and railroads did in the late 1800s.

There have certainly been some creative bond offerings lately. In February, Alphabet—the parent company of Google—sold a 100-year bond. Imagine paying interest on a debt for 100 years. In May, Alphabet disclosed that it would take on debt denominated in Japanese yen for the first time. That was part of its effort to raise $37 billion, which it now has to pay back. Also in May, Amazon disclosed that it is selling debt denominated in Swiss francs for the first time. That is part of a $37 billion debt deal. In October, Meta borrowed $30 billion as part of its AI build-out.

There is a fear among investors and governments about missing out and being left behind. There is a feeling that spending huge amounts of your own money and other people’s money is worth it.

The promise of AI is huge, Joel Hilliker wrote in the January Trumpet edition. Enthusiasts say it “is bigger than the Internet, more disruptive than railroads, more transformative than electricity.” But as he wrote, there is always a catch. “The strategic and economic pressures driving this venture are profound and irresistible.”

What if these amazing promises—and actual profits—take longer to realize than investors hope?

Ask the Philadelphia and Reading Railroad workers, or the Atchison, Topeka and Santa Fe Railroad employees. Or maybe WorldCom or Global Crossing investors, who spent billions laying fiber-optic cables for anticipated Internet traffic that took longer to materialize than hoped for. Or perhaps all the land speculators and oil service employees who lost their homes and livelihoods during the 1980s oil bubble.

When bubbles pop, a lot of people get hurt. Is AI a bubble?

America is entering a dangerous time. The economy has been swelling more than it otherwise would, largely pumped up by the greatest spending binge ever known. When the the bills need to be paid, will the hoped-for AI profits materialize?

Or will America’s economy prove to be artificial?

It is a good time to prepare to reduce your standard of living. You can begin by reading “How to Survive the Coming Debt Crisis” and The Financial Law You Can’t Afford to Ignore.