Behind the AI frenzy, an overlooked debt gamble

Written by: thiigth

Wall Street veteran and Oak Tree Capital’s Howard Marks recently made a wake-up call. The gist is: if this AI frenzy doesn’t end up as a classic bubble burst, it will be the only exception in human financial history.

But the problem is, most people are looking in the wrong place.

We are still debating whether Nvidia’s stock price is too high, or who will be the next Cisco. Everyone is obsessively watching every flicker on the K-line chart, trying to find clues of a collapse. However, the real eye of the storm isn’t on the lively stock exchanges, but in the silent, hidden corner that determines life or death — the credit market.

This isn’t a math problem about P/E ratios; it’s a high-stakes gamble built on massive debt.

01 The Disappearing “Cash Cow”

In this story, our biggest illusion is that tech giants are as rich as countries. We think they are using their idle cash to calmly reshape the world with AI.

But the numbers in their ledgers reveal a completely different reality.

Let’s turn back the clock before ChatGPT ignited the world. At that time, Microsoft was the most reassuring “cash cow” on the planet. Its cash on hand was 30% more than its debt — rock solid. But just a few years later, to feed the insatiable AI hunger, the situation reversed fundamentally: now, Microsoft’s debt exceeds its cash by 20%.

Meta (the former Facebook) is even more exaggerated. In 2022, Zuckerberg’s cash was three times his debt; by the last quarter, his debt had overtaken cash reserves by 15%.

Amazon, which already loved high leverage, now has debt half again as large as its cash. As for Oracle, the once cash-rich software empire, it now faces cash flow shortages and carries a mountain of debt.

In just four years, Silicon Valley giants have changed. To chase the elusive “Holy Grail” of AI, to avoid being left behind in the next industrial revolution, they have not only drained their reserves but also mortgaged future income, signing huge IOUs.

02 The “Real Estate” That Can’t Be Abandoned

This isn’t just about money; it’s a story of “no way out.”

Recall the 2008 subprime mortgage crisis: when house prices collapsed and pockets were empty, heavily indebted homeowners made a painful but rational decision: hand over the keys to the bank and walk away. That’s personal default — leaving the mess to the bank, life goes on.

But in the AI game, there is no option to “hand over the keys” to the bank.

Microsoft, Oracle, Meta — they are not like homeowners; they are more like developers who must complete their projects. When they sign hundreds of billions in chip contracts, or when server farms on the wasteland start piling up, they are signing a “blood oath.”

If tomorrow credit tightens and they can’t borrow money, can Oracle point to a half-built data center and say, “I’m out”? No.

Because if these semi-finished projects are not completed, they are worthless and face huge penalties, which could cause their core competitiveness to collapse instantly.

So, even if it’s a fire pit ahead, they have no choice but to keep burning money.

This is a classic debt trap: no matter whether AI can make money now, you can’t stop. Stop, and those hundreds of billions invested will instantly become zero;

Keep going, and you need continuous credit infusion.

That’s why the AI bubble is more terrifying than the stock market bubble. If the stock market crashes (like in 2000), it’s just people’s wealth shrinking and life tightening; but if credit collapses, preventing these giants from maintaining infrastructure, it’s a cardiac arrest.

03 Fear, the Highest Leverage

Why are these top CEOs collectively pushing themselves into this corner?

Because of fear.

It’s not just greed; it’s a deep-seated survival anxiety. At the crossroads of technological change, not participating means death. If you dare to sit on the sidelines and watch, you are doomed to miss the next great voyage. To stay at the table, giants have no choice but to borrow heavily.

Thus, we see this huge two-way gamble:

If AI succeeds: it might take 6 to 10 years. But during this decade, companies will carry heavy interest burdens, and money that could have been used for R&D will go to debt repayment. This will make once-light tech giants stumble.

If credit collapses first: before AI truly makes big money, if banks see too much risk and tighten the purse strings, this business cycle built on borrowing will snap like a taut string in an instant.

Today’s tech scene is like pouring gasoline on a smoldering fire. On the surface, we see stock market frenzy, but underground, accumulated debt fuel has infiltrated every inch of soil.

04 The Endgame

Most investors think: “I’m smart, I can escape before the bubble bursts.”

This is a typical survivor’s illusion. Everyone fantasizes about cashing out at the peak, buying government bonds, and then sunbathing on the beach. But history never plays out this way.

It’s like riding an escalator upstairs — smooth, comfortable, sleepy; but going down often means jumping into the elevator shaft. When the heavy door of the credit cycle suddenly closes, everyone is packed on the crowded track — in those core tech stocks — and the exit can be blocked instantly.

So, stop fixating only on whether Nvidia’s stock will rise or fall tomorrow. The real risk lies in whether the credit foundation supporting all this prosperity remains solid.

When Microsoft starts operating like a highly leveraged real estate developer, we should wake up: this is no longer just about tech dreams; it’s a brutal story of how to find a glimmer of survival in the debt swamp.

This article is compiled from Eurodollar University podcast content and is for reference only, not investment advice.