Picture a woman named Denise who runs a small catering business out of her kitchen. Last month she sat across from her bank manager, asking for a modest loan to buy a second delivery van. The number on the page made her wince: with the 10-year Treasury sitting at 5.04 percent, her small-business loan priced out far above what she had imagined. She did the responsible thing — she waited.
Now imagine, in the same week, some of the biggest technology companies on earth borrowing billions of dollars essentially for free.
That’s the strange financial weather we’re living in, and this week it caught the attention of the credit-rating world. S&P Global is warning that the financing behind the artificial-intelligence boom is getting bigger and more complicated — and that the pristine credit ratings of Big Tech’s hyperscalers are at risk.
What S&P Global is actually saying
The warning is worth unpacking, because it isn’t the usual doom-mongering. S&P Global’s concern centers on how the true debt load of the AI buildout is measured. When you look at a hyperscaler’s balance sheet the old-fashioned way, the numbers still look fine — these are among the most cash-rich companies in history. But S&P is pushing analysts to use an adjusted debt metric, one that pulls in the things that function like debt even if they don’t carry the word “loan” in their name: cash positions netted the other way, future lease obligations, and — crucially — power purchase agreements.
That last one matters more than it sounds. The AI data centers being built across the country are power-hungry in a way previous computing waves never were. To secure electricity, tech companies are signing long-term power purchase agreements that lock in huge, fixed financial commitments stretching years into the future. Functionally, those are obligations. If you treat them honestly, the leverage picture changes. S&P’s message is simple: the headline numbers flatter; the fine print is where the risk lives.
And here’s the part that should make everyday savers and investors sit up. While S&P waves this flag, Wall Street is simultaneously witnessing a record surge in zero-coupon convertible bonds, according to data from Dealogic. These are instruments that let companies borrow billions essentially for free — no regular interest payments at all — in a rate environment where the 10-year Treasury just touched 5.04 percent, which Axios Markets notes is the highest since 2007. The bond converts into equity if the stock rises enough, so investors accept the zero coupon in exchange for upside. It’s clever financing. It’s also a sign of just how fevered the demand for AI-linked paper has become: lenders are so eager to get a piece of the AI story that they’re handing over cash and asking for nothing in return but a lottery ticket.
It’s not enough to just marvel at the innovation or fear the complexity — we must listen to what the numbers are telling us, learn how this new financing really works, and contribute our own clear-eyed judgment as participants in these markets.
The echo boom nobody should ignore
There’s a wonderful detail in this week’s market coverage that tells you everything about where we are in the cycle. Axios Markets contributor Matt Phillips notes that Wall Street banks are reaping a windfall from stock and debt sales — the AI-driven capital-markets dash from companies like Alphabet, Nvidia, and SpaceX amounts to an echo boom for investment bankers. When the bankers are feasting, the financing cycle is running hot. That has historically been a reliable, if uncomfortable, signal.
Phillips also describes what he calls “the great AI wealth transfer”: a generational movement of money from the hyperscalers — Amazon, Alphabet, Meta, Microsoft, Oracle — to the chipmakers and equipment suppliers, names like Nvidia, Broadcom, Applied Materials, and Micron. Nvidia’s own week reinforced the point: the company reported blowout earnings for the second quarter of fiscal 2027, beating Wall Street expectations so convincingly that even jaded investors were won over. Meanwhile, software stocks staged a comeback — the State Street Software and Services ETF, tracking roughly 130 software and IT names, rose 5.2 percent Thursday to a new all-time high, as investors concluded AI won’t obliterate the software sector after all.
So the picture is genuinely two-sided, and honest writing should say so. The AI trade is producing real earnings, real revenue, and real optimism in parts of the market. Nobody is claiming the data centers are empty.
Why the credit rating warning still matters
But here’s where the S&P warning earns its place in the conversation: credit ratings are the plumbing of the financial system, and they matter most precisely when everyone assumes they don’t need to think about them. A pristine rating — the kind the hyperscalers enjoy — is what lets a company borrow cheaply and on flexible terms. If that rating comes under pressure, future borrowing gets more expensive, covenants get tighter, and the cost ripples outward. And those costs eventually touch Denise and her delivery van, because the corporate bond market and the small-business loan market both drink from the same well of prevailing rates.
There is also a quieter risk in the zero-coupon convertible boom. “Borrowing for free” is a lovely phrase, but convertibles carry their own price: if the stock soars, the bonds convert and existing shareholders get diluted. If the stock stalls, the company still owes the principal at maturity — a bill that arrives years from now, when the AI cycle may look very different than it does today. Free money is rarely free; it’s usually just deferred accounting.
My take: the healthiest way to read this week is as a maturation moment, not a meltdown signal. S&P isn’t predicting defaults. It’s asking the market to price the AI buildout with its eyes open — to count the leases, the power contracts, and the clever zero-coupon structures as part of the real obligation stack. That’s what good credit analysis is supposed to do, and investors should welcome it. The danger would be if everyone decided the warnings were just noise. They aren’t noise; they’re the adults in the room doing their job.
For those of us watching from the sidelines — savers, 401(k) holders, small-business owners — the practical lesson is about humility toward leverage. When borrowing is easy and bankers are celebrating, that’s exactly when discipline matters most. Denise waiting on her van until the math works is not a failure of ambition. It’s the same prudence S&P is asking of the biggest companies on earth. The scale is different. The principle isn’t.
The AI story is still being written, and it may yet be one of the great wealth-creation chapters of our lifetimes. But every great chapter has a financing footnote. This week, we got to read it. Let’s not skip it.
