Invest1 distinct publisher3 min readPublished
The argument treats bond demand as a fixed pool, so if AI issuers can raise $1.7 trillion at barely wider spreads, the clearing price has to move somewhere, and Yardeni says it moved to the Treasury's own cost of funds.
The Investor · Invest desk
Compiled by The InvestorSomething wrong?How this is made
Divide Yardeni's $1.7 trillion by the seven months that produced it and the investment-grade market has been clearing about $243 billion a month, which straight-lines to roughly $2.9 trillion for the full year [4][1]. The $2 trillion he flags as a first is therefore a floor rather than a forecast, and landing on it exactly would require the last five months to add $300 billion in total, about $60 billion a month, a quarter of the pace set through July [2]. One of those two numbers will look silly by December.
The increment is what the crowding argument actually rests on. At 27% above last year, the comparable period a year ago was about $1.34 trillion, so the extra supply is around $360 billion, or roughly 18% of the $2 trillion deficit the Treasury has to fund this fiscal year [4][1][3] (the calendar and fiscal windows do not line up, so read that as a magnitude rather than a ratio). That is the honest size of the mechanism, and it only bites if the pool of bond demand is fixed, which is the assumption doing the work inside the line that capital flowing into corporate bonds is capital not flowing into Treasuries [5]. The same account offers federal deficits with no end in sight, oil prices lifted by the Iran war, and an economy still pressing on inflation as competing explanations for the same yield move [7]. Kevin Warsh, speaking at Jackson Hole on Friday, went no further than observing that ever-expanding pools of capital are pouring into AI-related infrastructure of all sorts [11].
What weakens the spread evidence is what sits outside it. One tally puts hidden borrowing at $1.65 trillion, which is about 97% of the visible investment-grade total through July, and that is before private credit's role in financing the build-out and Nvidia leveraging its own balance sheet to back AI deals [10][5]. A compressed spread on public paper is a price struck on perhaps half the leverage in the system, and that is a reason to distrust it in both directions.
The arithmetic a Treasury official would find duller is the one that binds. Servicing $40 trillion for $1 trillion a year implies an average cost of about 2.5%, so the stock is still rolling from an old coupon toward a new one, and 25 basis points applied across all of it is $100 billion a year [1][4]. Debt service already runs at half the deficit [6].
This is probably wrong in its strong form, but the more interesting version of the thesis is that hyperscaler funding has become a rates variable rather than an equity one, and that a yield-agnostic bid means the coupon is no longer setting the pace of the build-out [2]. What would prove it wrong is an ordering: if the next quarter's large AI deals price materially wider while Treasury yields back off, the pool was never fixed and the supply was being absorbed by credit all along.
Ranked by verification strength, evidence, and original report placement.
U.S. debt is at $40 trillion, the federal budget deficit is on track to reach $2 trillion this fiscal year, and debt servicing costs alone are $1 trillion a year.
Treasury Secretary Scott Bessent said recently that a lot of corporate issuance is "almost yield-agnostic, because the build-out for AI, the returns on that, the companies believe they're going to be so high. They don't really care what they're paying."
U.S. investment-grade corporate bond issuance totaled about $1.7 trillion in the year to date through July, about 27% above last year's pace and on track to exceed $2 trillion for the first time, according to Ed Yardeni.
The rise in Treasury yields has been attributed to a variety of factors besides AI-related debt issuance, including massive federal budget deficits with no end in sight, higher oil prices due to the Iran war, and a robust U.S. economy putting upward pressure on inflation.
S&P Global warned last month that markets are showing some signs of fatigue after absorbing the flood of debt in such a short time, saying "Market participants are growing leery of quickly rising leverage from issuers previously characterized by strong and reliable cash flow."
The same S&P Global report pointed out that hyperscalers are paying a higher premium compared with yields on risk-free bonds.
Distinct publishers with included, body-backed reporting in this cluster.
Follow any of these and your For You feed starts watching them — no settings page required.
invest
Fed disunity, not the Fed's path, is the risk in duration before Warsh speaks1 distinct publisher
invest
Hoenig reads the Fed's silence on Treasury's bond buybacks as deliberate distancing1 distinct publisher
invest
Repricing already inside the debt adds $160 billion to Washington's interest bill1 distinct publisher
invest
Bessent said the toolkit is big. Yields went up anyway.1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One note, retold once, with no spread data
The causal argument rests entirely on a Monday note from Ed Yardeni, quoted by Fortune, and the quantity that would settle it — how much spreads did or did not widen — never appears. The corroboration on offer is unquantified: capital-flow data with no figures, a hidden-borrowing tally with no author, fiscal aggregates with no source. What raises this above the floor is that the direct quotations from Bessent, Warsh and the S&P report are checkable statements of record, even if the inference drawn from them is not.
The borrowing is documented; the diagnosis is not shared
Split the question in two and it resolves cleanly. The behaviour is happening at scale and on the record: $1.7 trillion raised in seven months, foreign buyers reportedly tilting toward corporates, a rating agency writing about absorption fatigue, the Fed chairman describing pools of capital at Jackson Hole. The interpretation is another matter — Yardeni asserts crowding-out, Timmer echoes it on X, and that is the whole roster. Bessent and Warsh describe the flows without endorsing the conclusion drawn from them.
A monocausal headline over a multicausal caveat
The framing hands AI issuers responsibility for the Treasury's cost of funds, then, four paragraphs later, concedes that deficits, an oil shock from the Iran war and a hot economy are also pushing yields up — and never weighs them against each other. The overstatement is in the attribution, not the numbers. Running the other way, in one narrow spot the story undersells itself: the pace it reports annualises to something closer to $2.9 trillion than the $2 trillion record it advertises.
A market strategist, a bond salesman, and a rating agency
Every voice here has a position in the outcome. Yardeni sells market interpretation and gets the note's headline; Bessent, whom the story itself calls America's top bond salesman, benefits from an account in which rising yields reflect private borrowing appetite rather than his own deficits; S&P Global rates much of the leverage it is now describing as making participants leery; Timmer speaks for a fund manager with duration exposure either way. None of that makes the argument wrong, but the crowding-out story is convenient for the Treasury in a way Fortune notes without pressing.
Directionally interesting, mechanically unverified
We are reasonably confident about what was said and by whom — the quotations are specific and attributed. We are not confident about the claim that matters, because a single publisher relays a single note, the spread evidence pointing the other way sits unaddressed inside the same piece, and the issuance projection does not survive contact with its own inputs. A second outlet, or the note itself, would move this quickly.