On August 12, I suggested next year’s biggest macro-market story may be the interplay between Scott Bessent’s issuance strategy and a structural increase in high-grade corporate debt supply.
The hyper-scalers alone are set to sell $400 billion in new debt globally both this year and next to fund the AI buildout, and that’s just five companies. If you account for every company tapping debt markets to finance data centers, the figures are larger. If you take it a step further and include all loans earmarked for accessing compute, they’re larger still. And if you expand the definition of “AI-related,” you can conjure more or less any number you want.
Of course, not all of that borrowing’s relevant for the US Treasury Department. Or perhaps it’s more accurate to say some of it (some of the AI borrowing, I mean) is more relevant than others for Bessent. As noted here on Wednesday, about two-thirds of AI-related issuance (on a standard definition of “AI-related”) in 2025 and 2026 is 10-years and out. That’s the chunk that matters most, as it’s a lot of “extra” duration from high-quality issuers.
Even if the big AI spenders dial back capex plans and AI-related debt sales don’t ultimately become a structural feature of the high-grade credit market, this borrowing binge will continue at least into, and very likely through, 2028. That, in turn, means Treasury has no choice but to factor it in when contemplating its issuance strategy.
On Thursday, we got a reminder of why this issue, which would be pressing under any circumstances, is especially exigent now: Uncle Sam paid the most to borrow for three decades since 2001 as the 30-year refunding auction stopped at 5.216%.
There’s the chart. As a quick aside: Thursday’s auction yield was 45bps higher versus autumn of 2023, when Janet Yellen was compelled to placate the market’s oversupply concerns by tipping smaller-than-expected coupon increases. At the next quarterly refunding, Yellen inaugurated the “…for at least the next several quarters” guidance that’s been a fixture of the QRA ever since.
The sale itself went ok, all things considered. Even if there weren’t a litany of reasons to be bearish the US long end (and there most assuredly are), 30-year sales in August tail almost as a matter of course, so despite the concession inherent in the US long bond at ~5.20%, there was no guarantee the stats would be good.
In that context, I imagine the 0.4bps tail, non-dealers at 88.5% and a bid-to-cover of 2.39 constituted an “I’ll take it” moment for Bessent, if not necessarily a sigh of relief. Consider that the vast majority of Americans don’t follow this sort of thing, which means headlines shouting about the highest long-term borrowing costs in a quarter century will sound more vexing than they do to market participants, who obviously knew this would be the highest-yielding 30-year sale in quite a long time.
As the figure below reminds you, 30-year yields have been above 5% every session since the Iran ceasefire collapsed.
“The bearish momentum in the long-end hasn’t been without fundamental justification, of course,” BMO’s Vail Hartman remarked. “The energy supply shock, hyper-scaler debt issuance, concerns about a higher r-star, and [Kevin] Warsh’s elimination of forward guidance all serve[d] as an offset to aggressive bidding for the new 30-year bonds.”
When you think about all of this, don’t forget that the buyer base for Treasurys has shifted meaningfully over the past several years. That shift was part and parcel of the mini-panic that pushed long-end US yields up sharply from August of 2023 to late-October of that year.
No longer can Treasury depend on a price-agnostic bid (e.g., from the Fed and foreign investors, many of whom are irked for one reason or another at America). Instead, Treasury’s selling debt into a more or less real market, which is to say price discovery’s back.
As I wrote three years ago, America’s trying to finance growing deficits amid intractable political dysfunction inside the Beltway, where the odds of bipartisan fiscal reform are nonexistent. As yields rise, so does the country’s debt servicing bill, which makes Treasurys look even less appealing, leading to even higher yields and so on. Wars don’t help. Not only are they expensive, they can drive up inflation, making bonds less attractive as an asset class.
Consider the above a 700-word addendum to the Wednesday feature piece mentioned here at the outset. And if you, like me, are inclined to suggest Thursday’s 30-year refunding actually went well under the circumstances, it’s worth noting that, as one PM put it in remarks to Bloomberg, “a successful auction shouldn’t be confused with strong structural demand.”
Yep. No one is paying attention to this. Note that July’s $432B blowout deficit — the largest July shortfall ever recorded — pushed the fiscal trajectory sharply higher. With the year?to?date deficit already at $1.8T, FY26 is now on track to be the biggest deficit since the pandemic years, landing somewhere in the $1.9T–$2.1T range once August and September are booked.
It’s hard for me to wrap my head around who’s actually buying if not governments (foreign or domestic). These are huge numbers, and in my tiny brain, buying one asset (treasuries) requires using cash or selling another asset in order to purchase.
Not long ago the FT published an illuminating piece adding some color to this commentary. It included a chart which showed that the only major group recently increasing their buying of US Treasuries has been carry traders who borrow short and buy long. All juiced up with mega-leverage. The absolute level of interest rates matters less to them than the spread between short and long rates. I don’t see that group as being particularly long-term buy & hold investors, but what do I know?
I believe banks and pension funds trying to match the duration of their liabilities will buy TLTs merely out of necessity. Also, at nearly 4.7% on the 10-year, and 5.2% on the 30, the yields are becoming quite attractive. (Recall, treasuries pay nearly guaranteed interest that is state tax free). If the market should turn abruptly, TLTs could also benefit from a “flight to safety.” Finally, if Trump finds a way out of this war, and Warsh does manage to cut rates, bonds with these yields could eventually sell above par value. (But, as someone pointed out to me in a similar thread some time ago, “that’s a lot of ‘ifs.'”
Close parenthesis.
The ONLY way out is inflation. I don’t believe that productivity will do it. And it certainly won’t be that we actually run a budget surplus as a country and start paying it down. Soon enough, YCC will return.
Once again, the little guy is going to get effed. Inflation will run hot for years, but he won’t have the equity side or significant hard assets to offset it.
I saw the 30yr auction and the tail, although not great, is still untimely. Interestingly, I wondered if the reason Bessent sold euros in the Yen intervention last week was so he wouldn’t have to sell treasuries with this week’s auction in mind. There is some speculation that the ECB is pissed that it wasn’t given the heads up on the currency intervention and that if Bessent does the same again (as the Yen seems to be heading back towards 160), the ECB may sell some of it’s own holdings of US treasuries! (with “friends” like these eh) Still, after this weeks’ benign inflation readings along with last friday’s terrible nfp numbers, it may forestall some of the dollar appreciation (although the BOJ really needs to hike it’s own interest rate)
I’m not sure about the wording of your first sentence. It seems to reflect some confusion about auction lingo. In any case, Thursday’s small tail really needs context. There hasn’t been a stop-through at a 30-year August supply event in a dozen years. Over that period, there were nine tails, with an average of 1.7bps. More than three-quarters of August long-bond auctions have tailed since Treasury revived the 30-year bond, and by an average of more than 2bps. (h/t BMO)
Very timely and informative articles in this series. Thank you.
I’m old enough to remember when the only “crowding out” was by the Treasury in the IG market and vice-versa was inconceivable.
That’s actually a really good point. In my own economics education, the risk of “crowding out,” whether in financial or physical markets, was always one of the top arguments in favor of small government. Now to ask (largely asking myself, or at least, my college-age self): if that argument was valid for governments, does it not also apply to mega-corps?
This was published on August 13th and was one of many. All of a sudden on August 18th the mainstream financial press is picking up on this! Sell, whodathunk it?
I wish I could add a hands-clapping emoji to give our Dear Leader a round of applause for sticking with this issue.
Example from MarketWatch right now: “America’s growing debt pile will be the big focus Wednesday as global bond rout deepens”
My my, this is a black swan that came out of nowhere!!!