Hi friends. This week the US federal debt crossed the $40 trillion mark. Honestly, it's a number that doesn't feel real. It's very hard for a normal person (and for a not-so-normal one, too) to get your head around an order of magnitude like that, let alone to understand what it means. But as an economist who has spent years working inside the numbers, that one didn't catch my eye and didn't knock me off my chair. What stopped me was something else entirely, and it happened two days later: the US Treasury announced it is doubling the amount it spends buying back its own debt.
Think about that again, because at first glance it doesn't really add up. A government that issues bonds is a borrower. A government that buys its own bonds back off the market is a borrower chasing down his lenders to pay off the loan early (and in our case, right when his debt is at a peak and he needs the money most). Why would anyone do that at the exact moment the interest he's paying is at a nineteen-year high? The answer, as always, starts in the most boring place to look — the bond market.
The day before the Treasury's announcement, on August 18, the yield on the 30-year US government bond touched 5.33%, the highest it has been since 2007. That yield is simply the interest the US government pays to borrow money for thirty years (on a newly issued bond), and nobody sets it from above — not even the American central bank. It gets set in the market, and the ones who actually set it are the buyers. When they're calm they'll settle for less, and when they're worried they demand more before they'll agree to lend at all. Since late June the buyers of long debt — meaning the 30-year bonds — simply stopped showing up in the numbers the market was used to, and the price the government has to offer to tempt them back climbed, and then climbed some more.
This is where the Treasury steps in, announcing that starting September 9 it is doubling the size of its purchases in the 30-year bond, from $2 billion a batch to at least $4 billion, and Treasury Secretary Scott Bessent added the next day in an interview that it could run higher than that. The market reacted immediately, and the 30-year yield fell that day by almost a tenth of a percent, to 5.196%. The way a yield comes down on a bond that already trades in the market is through the bond's price: when there are more buyers than sellers the price goes up, and the yield on it goes down. Meaning, that day there were far more buyers than sellers. And then, inside of a single day, that drop was wiped out. Why? Because somebody asked the obvious question — where is the money for these purchases coming from — and the answer is that the government is issuing short-term debt (that is, going out to sell new bonds and stepping right back onto the sellers' side of the bond market) in order to buy back long-term debt.
Those of you with a mortgage already know this maneuver up close, even if nobody ever gave it a name. You have a thirty-year fixed portion, it's expensive, and you shift part of it into an adjustable-rate portion tied to the short rate — meaning a rate that resets quickly — which today, as it happens, is cheaper. The monthly payment really does come down, and that's not imagination and it's not an accounting trick. But the debt itself hasn't shrunk by a single dollar, and the risk hasn't gone anywhere; it has only moved from today to the day the short rate resets. The US government is doing exactly this, only on a scale of trillions, and if I may point it out — Bessent himself, the honorable Secretary of the Treasury, went after his predecessor pretty hard for this very maneuver, two years ago.
Now, let's climb up a floor and look at this from above, at the long-term process that's unfolding here. The average interest rate the US government pays across all of its debt stands today at just 3.39%. Not 5.3%, because most of the existing debt was raised in years when money cost almost nothing (in 2021 that average was around 1.5%). But debt, unlike a building, doesn't stay standing where you left it. About a third of the US debt comes due and gets rolled over (meaning a new bond is issued that has to find buyers in the market, against the one that just matured) within roughly a year, and every cheap bond — issued back when rates were lower — that reaches the end of its life gets replaced by a more expensive one. The implication is simple and unpleasant: even if rates froze today exactly where they stand, the US interest bill would keep climbing for years, purely from the swap. That's the slow runner in this story, and he never gets tired.
Against that slow runner, what exactly is supposed to run faster? This is where artificial intelligence comes in, along with quantum computing and the rest of advanced technology. In the first quarter of 2026, investment in artificial intelligence — data centers, hardware and software — accounted for about 74% of all the growth in the US economy. Meaning, without that investment, the largest economy in the world barely grew at all that quarter. The story the market tells itself is that this enormous investment turns into efficiency, that companies produce more with fewer people, and that profits grow faster than the interest on the country's debt.
To my mind, that can absolutely happen. But it's worth remembering that as long as we're talking about investment in the future, that money is still sitting on the spending side of the equation, and (almost) not on the earning side. Every new technology in history has moved the same way: first you pay and pour money in, then it works, and the efficiency — and the big profits — show up only after that. And more than that: the spending is certain, the profits afterward are still in doubt.
What does that do to your stocks?
On the face of it there's no connection between the interest a government pays on its loans and what a share in a software company is worth. In practice, they're very much connected. The value of a stock is, in the end, the profits the company will produce in the future, translated into today's money — or in plain language, into the price you're paying for a share you're buying right now. A dollar of profit arriving ten years from now is worth less than a dollar in your hand today, and exactly how much less depends entirely on the safe alternative available to you. If you can get 4% risk-free on a long government bond, that's one thing. If the safe alternative has jumped to 5.3%, those same future profits you're counting on from the stock are already worth less against the stock's current price — because you're "giving up" a bigger pile of interest you would have earned had you not bought the share — and from there: same company, same profit forecast, lower price today. Meaning, a falling stock market.
Now, all the new guys (new to me, mind you), meaning anyone who's been trading since 2010, will tell me right away — so what. The market falls, stocks get cheaper, we buy, the market comes back. And they're right, that really is what happened every time, and fast. In March 2020 the index fell 33.9% in 33 days and was back at its previous peak within five months. In 2022 it fell 25.4% over nine months and was back at the peak within about two years (roughly — it depends which index you're looking at). Both times the pain was short and temporary, and if it taught the new investors anything, it only hardened their faith that every dip is a buying opportunity. But I, old-timer that I am (and reasonably well-read, too), remember the index buyers of the year 2000 perfectly well — the ones who sat thirteen years in the stocks they bought just to get back to what they paid for them. I've also studied the earlier history of these markets, and of other markets while I was at it (Japan, anyone?), and I know and I remember that sometimes the market does not come back so fast. And there are always reasons.
Before anyone here decides I'm a pessimist, it's important to me to say that there's a scenario that ends well, and it's a long way from fantasy. If the efficiency genuinely arrives, the economy grows, the government's tax revenue grows with it, and the debt shrinks relative to the size of the economy without anyone actually paying it down. This has happened before, and not that long ago by the way (in historical terms): after the Second World War the US carried debt worth 106% of its economy, and by 1974 it was already down to 23%.
Granted, in that case it wasn't growth alone that solved the problem, but also heavier taxation, and inflation that ate away at the debt (we'll get into that one in a separate letter sometime). But the bottom line is that government debt is a solvable problem. It's simply solvable over time rather than in one blow, and someone always pays the price. With taxes, it's whoever pays them. With inflation, it's anyone holding an asset that isn't indexed to it. But here's the important part: a shrinking debt lets the government pay less and less interest on its bonds, and lets the stock market go on realizing the upside it's capable of.
What this means for your pocket
If you have cash on the side, you're being handed an investment opportunity you haven't seen in nearly two decades: a 30-year US government bond paying 5.3% with no default risk. For years the line that "there is no alternative to stocks" was simply true, because rates were near zero and there was nowhere else to go. It isn't true anymore, and that is exactly why some of the money sitting in stocks today may want to leave, why new money may not come in at all, and why money that was supposed to finance growing companies can end up sitting in institutional pockets with nobody lending it out.
And if you stayed in stocks and passed on a guaranteed 5.3% a year for the next thirty years, what you're meant to take away from all this is that you are, in fact, betting that long-term growth in the stock market — through rising profits at the companies inside the indices — will turn out strong enough both to justify the earnings multiples you're paying today (I wrote a letter on exactly this about Nvidia, and I recommend jumping over to it here) and to pass enough tax from those companies to the government to bring the debt down. Quite a bet. And everyone buying stocks is buying that bet.
So who wins?
What would I do? If I'm being honest, to my mind what's happening in our lifetime is a revolution that isn't "once in a generation" or "once in a century," but something closer to once in the history of humanity. And personally, I wouldn't want to sit on the sidelines with a guaranteed rate and bet against that revolution. At the end of next week, on August 27, the Jackson Hole symposium opens and Fed Chair Kevin Warsh gives his first address there in the role, and that's where we'll start getting a hint of where interest rates are taking this whole story.
We shall see.