A More Detailed Look at How Trump and AI Could Crash the Economy

Additionally, smart money is not driving this current market rally that has painted a multi-month bearish divergence on nearly every high timeframe chart you see. All year long, we have seen headlines like “FOMO vs. Fundamentals: Retail buys the dip, institutional investors stay cautious” and “Bullish retail traders are the biggest force behind the stock market’s largest rally to all time highs.” I can point towards serious metrics like the CAPE Ratio proving this to be the second-most expensive stock market in history (second only to the height of the dot com bubble), but it doesn’t matter because the marginal buyer in the stock market probably doesn’t even know what a price to earnings ratio is anymore. It’s just moonbois and options gambling all the way down. The stock market is ripe to have the rug pulled out from under it, and unfortunately, its wealth effect has a large impact on GDP growth.

Retail Investors have been loading up on stocks this year while professionals (a.k.a. the “smart money”) have been dumping 🚨🚨

[image or embed]

— Barchart (@barchart.com) August 19, 2025 at 9:07 AM

The AI Bubble May Be Popping, Or At Least Slowing

My jaw hit the floor several times researching my article on AI’s shocking importance to the economy. It surpassing the engine of our economy, consumer spending, in its contribution to GDP so far this year is gobsmacking. It strongly suggests that if not for truly historic levels of spending on data centers by quite literally a handful of men atop a smattering of companies that smart money believes are overvalued, Trump’s tariffs would be a far larger drag on the economy than they currently seem. It would not be a stretch at all to suggest that America has bet its present economy on AI financially succeeding to the degree it claims it can.

But what if it doesn’t? A terrific article by the Financial Times last week detailed bankers’ doubts about the gargantuan financing costs of this data center buildout that is propping up the American economy. “We are in that period where the capital markets are crazy enough to throw money at almost anything,” said a banker who helps arrange financing for AI infrastructure projects to the FT. “I am curious to see the next phase and whether rationality prevails.”

The problem with AI’s massive investment in capital expenditures is that AI doesn’t currently produce enough revenue to pay for it, and all this investment is predicated on AI creating a great leap forward for society akin to the internet. All that data center-centric economic growth? A lot of it is financed by debt, as FT notes that by 2029, “global spending on data centres will hit almost $3 trillion, according to Morgan Stanley analysts. Of that, just $1.4 trillion is forecast to come from capital expenditure by Big Tech groups, leaving a mammoth $1.5 trillion of financing required from investors and developers.”

Securitized debt, that favored phrase that gives folks 2008 GFC PTSD, is increasingly at the base of this AI data center expansion. And who is financing this debt? FT says this “gap will be filled by everything from private equity, venture capital and sovereign wealth to bank loans, publicly listed debt and private credit.” Oh good, let’s get the whole global economy involved in this narrative-driven supposed gold rush, what could possibly go wrong? Private credit particularly has exploded since 2008, in part as a good development to keep risky debt off banks’ balance sheets, but like everything in America, seemingly is becoming too much of a good thing and its opacity is the increasing draw for private capital. Many have identified it as a potential risk to the economy simply due to how little we know about it and how large it is becoming.

Bankers both in this FT piece as well as everyone with access to AI companies’ financial data are beginning to question its return on investment. We are told this is a revolutionary product that will intersect with every nanometer of our lives, yet few folks really willingly pay for it. There are niche uses for it and some people will pay a subscription for its utility, but overall, as writers like Ed Zitron have documented, a lot of AI demand is just Microsoft and other giants effectively selling AI products between companies they either directly or indirectly own or control. AI is increasingly becoming a money trap, especially at Meta, who is now “overhauling” its massive AI investment.

What happens if the music stops one day and the Microsoft’s of the world begin to seriously pull back on their data center investment? Who is paying those debts? What effect will it have on all the other investments those debtors fund? What will it do to Microsoft’s stock price? What will that do to the rest of the stock market historically concentrated in just a handful of stocks driving it to new all-time highs? What will that do to the economy?

The AI narrative is changing. That may seem superfluous, but in the second-most expensive stock market in history driven almost entirely by assumptions around debt-fueled data center expansion, narrative is everything. Retail investors have bought stocks at the largest level in ten years according to JPMorgan. The immense hype around AI finding its way into earnings calls across a market reaching all-time highs proves how people are (were?) buying a narrative, and the mood in the market this week is that it is slowing down. Not only does my caveman technical analysis suggest that, but the degree of yesterday and now today’s continued selloff in the Nasdaq suggests the tides have turned to some degree (the Nasdaq is currently on pace for its worst week since Trump’s tariff announcement in April).

Bank of America’s US AI beneficiaries basket cleanly through the 50-day for first time since April.

[image or embed]

— Luke Kawa (@ljkawa.bsky.social) August 20, 2025 at 8:35 AM

You can’t just say “we did an AI thing” on an earnings call and hear your investors ooh and aah anymore. The jig is up in a sense, mainly due to the gargantuan numbers the basic math spits out around the investments being made right now and the valuations of companies like OpenAI with revenue that does not justify its market cap. This stuff is already priced like a revolutionary product, so where is the revolution? As Laura Cooper, macro credit and global investment strategist at Nuveen said to the Financial Times, “Investors are now questioning the durability of the [stock market] uptrend” due to “stretched valuations.”

AI is partially a debt bomb waiting to go off if it doesn’t get paid down first, and the mounting problem for AI is that its debt burden is based on valuations that a market may be looking at as overheated. If you’re wondering what happens when the value of your assets decline as the value of your debt stays the same, read more about how I lost a million dollars in crypto. As 2008 taught us, the nature of debt bombs is that most don’t know they’ve exploded until it’s too late, and the notion of suggesting in 2006 that Lehman Brothers would soon go bankrupt is about as preposterous as suggesting now that Meta and its pedophile chatbot will go bust soon. The whole point of an economic crisis is that a surprise catalyzes it, and businesses and people adjust off of it as the range of potential outcomes become worse and worse. Very often that catalyst is debt because it is a classic bill that comes due that people need to raise cash for. If AI does not become a serious revenue generator for the economy without completely destroying the labor market, some extremely heavy debt payments will weigh down some very large companies, and that tends to be when skeletons come out of people’s closets.

Speaking of the labor market…

Stagflation

The divergence between a good CPI report and a bad PPI report last week is proof that producers are eating the cost of Trump’s tariffs right now, and it has not bled down to the consumer yet, but GM taking a 35 percent hit to their second quarter net income strongly suggests it eventually will. Tariffs have clearly had a chilling effect on hiring, as well as Trump’s racist deportation raids combining to jack up the prices of goods like vegetables by 39 percent. Trumpflation is already here, the question is in what products and to what degree.

And this Trumpflation is hurting economic growth. No one wants to make long-term investments when president deals wakes up every day and farts out a new tariff rate for our allies on TruthSocial (or when he can’t even produce trade agreements proving the deals he’s touting actually exist). Trump can fire all the BLS statisticians he wants, but it won’t change the dynamic these companies are reporting in their BLS surveys and on their own balance sheets. The labor market keeps getting weaker, not stronger, and it’s because Uncertainty Season is the only season under Trump’s trade war economy.

The latest PPI data shows that inflationary forces are now making their way through the economy (so do rising energy prices), all while GDP figures reveal that growth is slowing relative to last year. If you add persistently rising unemployment to slowing growth and rising inflation, that is called stagflation, the worst thing that can happen to an economy that was once thought to be impossible until the 1970s came along. This is the explicit fear that Fed Chair Jerome Powell has communicated throughout the year, only becoming more and more direct in his critiques of Trump’s economic policy making his job to tame inflation more difficult.

Trump and the GOP have also established an overall policy aimed at jacking up interest rates through the entity that actually sets them: the bond market. American debt was downgraded to send a message to the GOP that their murder bill will explode the deficit and throw the bond market into chaos, and long-term rates have been up ever since it passed. Firing the nonpartisan BLS statistician overwhelmingly confirmed by Republicans and replacing her with a Nazi sympathizer who struggles with basic math will only further harm bond investors’ confidence in our debt, which translates to everyday folks through higher interest rates. You think it’s hard to buy a house now? Just wait until the BLS produces some funny numbers and the JPMorgan’s of the world can tell and institutions stop buying our debt as the 30-year yield runs to the moon.

So here’s how this all comes together to create disaster. The labor market is very clearly on a downtrend, and no hack statistician will change that. Firms can’t take on debt to hire people or invest in the American factories Trump wants them to build because Trump’s trade war is injecting inflationary forces that threaten to jack up interest rates and make debt more expensive. This all is leading to a slow spiral around the drain where economic growth gets slower and slower while inflation slowly rises until something breaks.

Crises are unpredictable by definition, but let’s just take a non-Trumpy example where an AI giant delays or misses a debt payment, and a market becoming more skittish this week looks at that as a canary in the coal mine moment. Regardless of whether it is or not, perception is reality when buy and sell buttons are involved, and given the bear divergences over long timeframes that any trader can see for themselves in the stock market, should the euphoric AI sentiment seriously turn among the moonbois helping to prop it up, it would be easy to talk yourself into selling the second-most expensive market in history. Have enough people do this, and it can create a panic. If that hypothetical delayed or missed debt payment was indeed a canary in the coal mine moment for AI, then I suggest that folks brush up on their Lehman Brothers history. When someone misses a debt payment, quite often that leads to more people missing debt payments, as the money they were counting on to pay other people does not come in, etc…etc…etc…and the government eventually has to step in and backstop everyone before the economy collapses. That is the short story of the 2008 Great Financial Crisis.

The amount of AI infrastructure debt this economy is carrying is extremely dangerous. That Morgan Stanley $1.5T debt figure between now and 2029 is equivalent to roughly five percent of annual US GDP. It’s not hard to see how this house of cards not making any serious revenue relative to its investment could collapse, taking the rest of the increasingly stagflationary economy with it. The AI bubble popping would leave us alone with no artificial debt-driven infrastructure boost, just Trump’s 19th century trade war whose logical conclusion is lower growth and higher inflation. Every day we are closer to stagflation than the last, and I feel strongly that if the trade war is still ongoing when I write my next Cassandra update in six months, that will still be true.