The American Economy Is Shockingly Dependent on AI

This was a *very* bad report. Employment down, implies ~4% annualized core PCE, and demand is clearly weak given new orders.

*US JULY ISM SERVICES PMI FALLS TO 50.1 FROM 50.8; EST. 51.5

[image or embed]

— George Pearkes (@peark.es) August 5, 2025 at 8:04 AM

Growth would be weaker without the AI capex boom. Probably a lot weaker. Economic changes are not linear, so removing one variable affects all the others, but needless to say, if the greatest consumer spending economy in the history of mankind now sees more growth from AI capex than consumer spending, AI capex is vital to our current economic health.

The simple answer to how this economic regime change seemed to unfold so quickly under our noses is that the largest companies ever started spending more than ever very quickly. Large Language Models (LLMs) require massive amounts of computing power and energy, necessitating a gargantuan investment in physical infrastructure, as the Wall Street Journal’s Christopher Mims excellently detailed in “Silicon Valley’s New Strategy: Move Slow and Build Things.” Meta has gone from renaming their company around a fake world no one will ever visit to bragging about building out a data center across an area the size of Manhattan. The world has changed pretty dramatically in the last few years.

The logical question is what if this AI bubble pops, and it likely means “watch out below?” as Kedorsky so aptly states. The math is simple enough and Trump’s 1934-esque tariffs are restrictive enough that if something that comprises a larger share of the economy than freakin’ consumer spending is removed, the economy will likely tank.

However, the magnificent seven companies (Microsoft, Nvidia, Amazon, Meta, Tesla, Alphabet, Apple) investing in all this infrastructure that is driving the stock market to all-time highs and creating a wealth effect that also helps prop up consumer spending would like you to think about what if it works out! I was taught in finance school that if you pay for basic infrastructure for your business, that these huge investments should produce proportionally sizeable returns in the future, growing the economy in a much more organic fashion than trillion-dollar companies spamming the “build more data centers” button. We should be thinking about the beginning of a boom, not a bust!

But AI has a demand problem, and the amount of people truly using it often is very much in question. As Ed Zitron has documented, the subscription as a service (SaaS) products offered by companies like Microsoft tie into their AI products, creating a much higher floor of demand than would likely exist in a world where companies weren’t propping themselves up. LLMs also have a basic economic problem where they function like the opposite of the code that builds Facebook, where it becomes more expensive to scale the product alongside more users. Despite its helpful abilities in specific areas, AI is still very much a product searching for a market. OpenAI loses money on its users, and a lot of this capex spending is financed by debt on the backs of companies whose valuations have risen by trillions over the past year because of the AI boom. If this doesn’t work out, it will get ugly for the magnificent seven, as Morgan Stanley calculated a “financing gap” of $1.5 trillion between capex and company cash flows over the next three years. It’s already getting ugly for one of them.

“During the first half of the year investment-grade borrowing by tech firms was 70 percent higher than in the first six months of 2024,” wrote The Economist. “In April Alphabet issued bonds for the first time since 2020. Microsoft has reduced its cash pile but its finance leases—a type of debt mostly related to data centres—nearly tripled since 2023, to $46bn (a further $93bn of such liabilities are not yet on its balance-sheet). Meta is in talks to borrow around $30bn from private-credit lenders including Apollo, Brookfield and Carlyle. The market for debt securities backed by borrowing related to data centres, where liabilities are pooled and sliced up in a way similar to mortgage bonds, has grown from almost nothing in 2018 to around $50bn today.”

Relentlessly hyping up a debt-financed product that doesn’t make money while resting atop 2008-style collateral debt obligations, what could go wrong!

Investment in information processing equipment and software rose at a 25 percent annual rate in the first six months of 2025, while GDP growth was 1.2 percent. America’s technically negative first quarter growth may have been actually negative without AI capex, and we might already be in a recession. Whether we want it to or not, the United States needs AI to financially succeed, we have literally bet our economy on it.