The Treasury Market May Not Be Breaking... It May Be Searching for a New Price

Why 5% yields, record equities, fading foreign demand and the AI capital boom may all be part of the same feedback loop.

Subscribe
MarketsMacroeconomicsAI
Listen to this post
AI Summary The 10-year Treasury yield has risen above 5% to a nineteen-year high, yet the S&P 500 and Nasdaq have hit record closes, suggesting the bond market may be searching for a clearing price rather than signaling a crisis. …
  • The 10-year Treasury yield has risen above 5% to a nineteen-year high, yet the S&P 500 and Nasdaq have hit record closes, suggesting the bond market may be searching for a clearing price rather than signaling a crisis.
  • Major tech companies issued roughly $220 billion in bonds through August 2026 and are projected to issue a record $420 billion in 2027 to fund AI infrastructure, putting them in direct competition with the Treasury for long-duration buyers.
  • Foreign holdings of U.S. Treasuries fell to about $9.25 trillion by July, with China's holdings at their lowest since 2008 and Japan now offering domestic investors a 3% yield on 10-year bonds for the first time since 1996.
  • The bear steepening pattern, with the 30-year near 5.3% and a positive 2s10s spread, suggests the market is repricing term premium and fiscal risk rather than anticipating recession.
  • Higher yields may eventually stabilize the market by triggering capital rotation from equities back into Treasuries, creating a negative feedback loop that caps further increases rather than causing a systemic crisis.

For years, a 5% yield on the 10-year Treasury was treated as something close to a financial alarm bell. We are now well past it. The 10-year printed 5.12% this week, a nineteen-year high, while the Federal Reserve, unusually, has just begun a hiking cycle rather than a cutting one.12 The instinctive reaction is predictable: surely this breaks equities, surely something has to give, surely the bond market is about to force policymakers' hands.

I'm increasingly unconvinced that's the right frame.

The more interesting possibility is that the Treasury market isn't malfunctioning at all. It may simply be searching for the yield required to attract capital in a world where that capital has far more attractive competing uses than it did a decade ago. If that's right, then 5%, 5.5%, even 6% doesn't mark a crisis threshold.

It marks a clearing price.16

The yield may be rising not because the economy has already broken, but because everything else is competing so aggressively for the same pool of savings.

Start with the contradiction

US Treasuries Yield Curve
Consider what's happening at once. The 10-year has pushed through 5% to levels unseen since before the financial crisis, yet the S&P 500 set a record close in August and the Nasdaq has since notched its first record close since June, carried by chipmakers.1112 Credit remains open for business. AI infrastructure is consuming extraordinary amounts of capital. Foreign buyers are becoming less reliable at the margin. Sovereign yields are rising around the world simultaneously. And the curve is doing something telling: the long end is leading, with the 30-year near 5.3% and the 2s10s spread positive.6 That is a bear steepening, the signature of term-premium and fiscal repricing rather than the flat or inverted curve that accompanies a recession scare.

Important

If a 5% Treasury were already an economic wrecking ball, there would be considerably more evidence of wreckage. Instead, the risk-free rate keeps climbing while investors keep bidding up risk. That isn't necessarily irrational. It may be the very mechanism driving Treasury yields higher.

Equities aren't merely victims of higher yields

The relationship is usually described in one direction: yields rise, stocks become less attractive, stocks fall. But capital markets don't run one way. When equities are producing strong returns and investors expect the earnings and AI-investment cycle to continue, money has less incentive to migrate into long-duration government bonds. Capital stays in equities, which means less marginal demand for Treasuries, so Treasury prices have to fall and yields rise until the return becomes attractive enough to pull that capital back.

The foreign-flow data shows exactly this preference. In June, foreign inflows into Treasuries shrank to $6.8 billion while inflows into U.S. equities reached $181.4 billion.8

The world isn't avoiding America. It is choosing American stocks over American bonds.

That dynamic matters enormously when stocks trade near record highs. The market is effectively asking why it should lock up capital for ten years at 4.5% when it believes corporate earnings and AI buildout can generate substantially more.

The Treasury market's answer is simple:

  • Fine. How about 5%?
  • 5.25%?
  • 5.5%?

Eventually the price becomes irresistible to enough investors that the flow reverses. That is how a market clears.

What actually sets the equilibrium

Here is where the popular version of this story needs sharpening, including the one I hear most often: that earnings growing 20–30% justifies a much higher yield. That intuition points in the right direction but anchors on the wrong number. The neutral rate doesn't key off index EPS, which is skewed by a handful of megacaps. It keys off economy-wide nominal growth and the balance between how much the economy wants to invest and how much it's willing to save.

Important

The 10-year, stripped to its fundamentals, is the market's estimate of the average short rate over the next decade plus a term premium, and over long horizons it tracks trend nominal GDP.

Framed that way, the AI boom does something more powerful than compete for capital. It lifts the anchor itself. A large, rate-insensitive surge in desired investment raises the neutral rate directly: more claimants on the same savings, a higher price to balance them. So the capital buildout shows up twice, once as a rival bidder pulling money away from Treasuries and once as an upward shove on the nominal-growth rate the whole curve is priced against. A structurally higher clearing yield follows from the investment surge, not merely from a good quarter of earnings.

Financing the AI Boom

Then add the AI capital boom

This cycle has an unusual feature: the private sector itself has become a voracious consumer of capital. Alphabet, Amazon, Meta, Microsoft and Oracle issued roughly $220 billion of bonds in 2026 through August 10, according to BNP Paribas data cited by Reuters, against capital spending expected to reach about $750 billion this year.7 The pace is accelerating. Goldman Sachs projects gross hyperscaler debt issuance of a record $420 billion in 2027, some 60% above 2026 estimates, and AI-linked debt already trades around 115 basis points over Treasuries against roughly 78 for the broader investment-grade market.7

The maturity profile is what makes this a direct Treasury story rather than a credit sideshow. More than half of this year's mega-cap tech bond financings have carried maturities of 10 to 50 years, which puts hyperscaler balance sheets in head-on competition with the long end of the Treasury curve for the same duration buyers.13

So the U.S. Treasury is issuing enormous quantities of government debt at the exact moment some of the world's largest corporations are selling hundreds of billions of long-dated paper to build compute. Equities want that capital too. So do private credit funds, infrastructure projects and foreign sovereigns whose own yields have finally turned competitive again.

When that many claimants chase the same savings, the price of capital rises. It would be surprising if it didn't.

The foreign buyer isn't what it used to be

This is where the "foreigners have stopped buying Treasuries" line gets oversimplified.17 They haven't stopped. But the marginal foreign demand structure is changing, and that distinction carries the weight. Foreign holdings fell to roughly $9.3 trillion in June, with Japan, the United Kingdom and China all trimming positions. Japan remained the largest holder at about $1.12 trillion, while China's fell to roughly $633 billion, its lowest since September 2008.8 The July data extended the slide: total foreign holdings dropped another $50 billion to about $9.25 trillion, a nine-month low, with France and Canada leading the declines.9

At the same time, Japan now offers domestic investors something it hasn't in decades: yield. On September 1 the 10-year JGB hit 3% for the first time since 1996, driven by inflation, fiscal concerns and pressure on the Bank of Japan to tighten faster.10 That reshapes the calculation for Japanese institutions. For years, U.S. Treasuries carried a compelling advantage over JGBs; now a Japanese investor has to weigh whether the incremental U.S. return survives currency-hedging costs, duration exposure and foreign-market risk.

Sometimes the answer is still yes. It no longer has to be, which means Washington may need to pay a higher rate to attract the same marginal dollar.

July shows why this isn't a buyers' strike

The data demand care. The latest Treasury International Capital report doesn't show foreigners abandoning American assets. July produced $83.7 billion of total net inflows; foreign official institutions bought $44.4 billion of long-term U.S. securities, and foreign investors added $38.8 billion of Treasury bills. Private foreign investors, however, were net sellers of long-term securities, and after adjustments for stock swaps the Treasury estimates overall net foreign sales of long-term securities at $27.9 billion for the month.14

That split matters... Money is still arriving, but it is arriving short: into bills, and through official channels, while private buyers step back from duration. The claim isn't that nobody wants American debt. It's that the Treasury may need to offer steadily more attractive terms to clear an enormous quantity of long-dated debt, because the marginal buyer now has more alternatives. A very different thesis, and a more important one.

So a 5%-plus yield starts to make sense

Assemble the pieces. The government needs vast amounts of capital. AI infrastructure needs vast amounts of capital, and increasingly at long maturities. Equities remain attractive enough to hold the world's marginal dollar. Japan pays a domestic yield again, China's holdings have structurally declined, inflation remains above target, oil and geopolitical risk have injected uncertainty, and central banks have been tightening rather than flooding the system with cheap money.

Under those conditions, why wouldn't the clearing yield on a 10-year be substantially higher?

The mistake is assuming 5% is a malfunction because 5% would have been restrictive under the capital-market structure of the last decade. That structure has changed.18

Not all of this is structural

A clearing-price thesis has to survive its strongest objection: a good part of this move is cyclical, and cyclical moves reverse.

The Fed has only just begun hiking, and officials are telling us why. Chicago's Austan Goolsbee has said the central bank can't overlook persistent supply shocks,3 and St. Louis's Alberto Musalem has signaled further increases may be needed to bring inflation back to target.4 Much of that supply-shock pressure is oil, and oil is a live geopolitical variable right now. Yields have already slipped on sessions when crude fell on diplomacy hopes.5

That is inflation premium and policy expectation moving, and both can unwind.

There is also a thumb on the scale. The Treasury has been leaning on bill issuance and buybacks to manage its funding needs, and in August it doubled buybacks of 10- to 30-year debt specifically to calm the long end.1315 A market whose long-end supply is being actively managed isn't clearing at a fully free price. If anything, that suggests the unmanaged clearing yield sits somewhat above where the screen shows it.

So the useful question is how to tell structural clearing from cyclical overshoot?

The bear steepening tilts toward the structural reading, but the decomposition is the test.

Higher yields eventually solve their own problem

Suppose the 10-year grinds from 5.1% toward 5.5%. Some investors rotate. At 5.75%, more rotate. At 6%, an investor staring at a richly valued equity market faces a genuinely different decision. A stock at 20 times forward earnings carries an earnings yield near 5%; a Treasury at 6% offers a contractual nominal return with no corporate credit risk. That doesn't automatically make equities unattractive, since earnings grow, dividends compound and shareholders capture productivity gains, but the hurdle rises sharply, and some investors simply take the 6%.

When they do, they sell equities and buy Treasuries. Treasury prices rise and yields fall. Meanwhile, the equity decline tightens financial conditions, trims the wealth effect and cools investment, which softens demand and relieves upward pressure on rates from the other side. The higher yields climb, the stronger the forces that eventually pull them back.

That is a negative feedback loop, and it's why I'm skeptical of a magic 6% breaking point.

Why I doubt a magic 6% breaks anything

There's growing debate about whether equities can withstand a 5.5% or 6% 10-year. Maybe they can't at today's valuations. But that isn't the same as 6% causing a systemic crisis. It may instead be the level at which asset allocation begins doing the Fed's work for it. Capital rotates, risk assets reprice, demand cools, Treasury buying increases, yields stabilize, and the system settles into a new equilibrium. The relevant question isn't at what yield does everything break but at what yield does enough capital rotate that the Treasury market finally clears. Those are radically different questions.

A real bond-market crisis would look different

The genuinely dangerous scenario still exists, and it's worth naming precisely so we can recognize it. Picture yields climbing rapidly while equities fall, credit spreads blow out, auctions tail badly, liquidity evaporates and foreign investors cut exposure all at once. That's no longer orderly price discovery. That's a funding shock, and if the move ran fast enough, the damage could arrive well before yields reached some round number like 8%.

That is not what the current configuration shows. The September selloff has coincided with resilient equities: the S&P 500's record close came in August with the 10-year already near 4.7%, and the Nasdaq returned to record territory this month on a semiconductor rally even with yields around 5%.1112

That is not what capitulation looks like. It's what competition for capital looks like.

The Search for a Clearing Price

What the 5% may actually be telling us

The prevailing read is that bonds are warning equities: rates are too high, you're going down. Maybe. But another read deserves far more attention: equities, AI investment, government borrowing and shifting global capital flows are telling Treasuries that yields aren't high enough yet.

Those two interpretations lead to opposite expectations. In the first, the bond market is the attacker and equities are waiting to break. In the second, yields are simply rising until they reach a price capable of pulling capital away from everything competing with them, and once that rotation begins, the process itself generates the demand that caps further increases. The second view explains why 5% hasn't produced the catastrophe so many expected, and why 5.5% or 6% might look frightening without being structurally unstable.

The number to watch isn't an arbitrary threshold. Watch the flows:

  • When equities begin losing meaningful capital to Treasuries
  • When hyperscaler issuance gets harder to place
  • When foreign demand finally responds to higher yields
  • When financial conditions start slowing private investment.

And watch the decomposition... whether real yields and the term premium, or breakevens, are carrying the move. When the flows turn and the real-yield story is intact, the bond market will have found its buyer.

Until then, the yield may simply keep searching.

Note

The question isn't "How high can Treasury yields go before something breaks?" It may be "How high must Treasury yields go before enough capital finally chooses them over everything else?"

Footnotes

  1. 10-Year Treasury Yield Rockets to 19-Year High. Here's What's Driving the Spike — CNBC, September 23, 2026 — The dated account of the move this piece opens on: the 10-year popping more than 13 basis points to 5.104%, a level not seen since July 2007, in the biggest one-day move in nearly 18 months. The drivers it names are stronger-than-expected U.S. activity surveys — manufacturing especially — hawkish commentary from a senior Fed official, a five-year note auction met with poor demand, and WTI crude up 2%. Live quote references for the same level: MacroMicro (https://en.macromicro.me/series/354/10year-bond-yield) and CNBC's US10Y page (https://www.cnbc.com/quotes/US10Y), both of which will show a different number in a week. https://www.cnbc.com/2026/09/23/treasury-yields-oil-inflation-fed.html ↩
  2. Fed Rate Decision September 2026: Rates Rise to 3.75%-4% — CNBC, September 16, 2026 — The hiking cycle the sentence refers to, and the reason the word is italicised: the FOMC's first increase since 2023, taken unanimously, lifting the target range to 3.75%–4%, with several policymakers projecting further increases before the year ends. This is the unusual configuration the whole piece is written against — a long end repricing while the short end is being raised rather than cut. https://www.cnbc.com/2026/09/16/fed-rate-decision-september-2026.html ↩
  3. Fed's Goolsbee Says Road to 2% Inflation May Not Be Painless — Bloomberg, September 21, 2026 — Goolsbee's remarks in London on September 21, 2026, and they are stronger than the sentence they support. His argument is that supply shocks have come more frequently, hit harder and lasted longer, so the Fed can no longer 'look through' supply-driven inflation as it has since the 1970s, and that forcing inflation back to target in the short run means pushing employment below target. He names oil near $100 after the war in Iran and repeated tariff escalation as the shocks in question. Open-access account of the same remarks: FXStreet, https://www.fxstreet.com/news/feds-goolsbee-demand-tariffs-energy-supply-shocks-all-are-fuelling-inflation-202609211113 https://www.bloomberg.com/news/articles/2026-09-21/fed-s-goolsbee-says-road-to-2-inflation-may-not-be-painless ↩
  4. Fed's Musalem Says More Rate Hikes May Be Needed to Tame Inflation — Bloomberg, September 21, 2026 — The St. Louis president's signal, in his own framing: both persistent demand and recurring supply forces are keeping inflation risks elevated, and further increases may be needed to return inflation to target. Delivered the same day as Goolsbee's London remarks, two days before the yield move this piece opens on. https://www.bloomberg.com/news/articles/2026-09-21/fed-s-musalem-says-more-rate-hikes-likely-needed-to-cool-prices ↩
  5. United States Government Bond 10Y — Trading Economics — The running market commentary this sentence draws on, for the sessions in which yields eased as crude fell on diplomacy headlines. A live page rather than a dated article: it reflects the most recent session whenever it is opened, so it supports the pattern described rather than any particular day's move. https://tradingeconomics.com/united-states/government-bond-yield ↩
  6. Treasury Rates Today — Forbes Advisor; U.S. Treasury Yield Curve — StreetStats — The curve levels behind the bear-steepening claim: the 2-year near 4.7% against the 30-year near 5.3%, leaving 2s10s positive while the long end leads. Both references are live curve pages and will move; the shape claim, not the levels, is what the argument rests on. Forbes Advisor's version: https://www.forbes.com/advisor/investing/treasury-rates/ https://streetstats.finance/rates/treasuries ↩
  7. Corporate Bond Buyers Get Picky With Flood of AI Debt — Reuters via U.S. News, September 22, 2026 — The forward-looking half of this section: Goldman Sachs projecting record gross hyperscaler issuance of $420 billion in 2027, roughly 60% above 2026 estimates, with AI-linked debt trading around 115 basis points over Treasuries against roughly 78 for the broader investment-grade market. The backward-looking figures — about $220 billion issued by Alphabet, Amazon, Meta, Microsoft and Oracle in 2026 through August 10, on BNP Paribas data cited by Reuters, against roughly $750 billion of expected 2026 capex on S&P figures — are at MLQ, https://mlq.ai/news/hyperscaler-bond-issuance-reaches-about-220-billion-as-ai-financing-costs-rise/ https://money.usnews.com/investing/news/articles/2026-09-22/corporate-bond-buyers-get-picky-with-flood-of-ai-debt ↩
  8. Foreign Holdings of US Treasuries Fall in June — Reuters via AOL — The June TIC data carrying both claims this marker supports: total foreign holdings at $9.299 trillion with Japan the largest holder at about $1.116 trillion and China down to $633.4 billion, its lowest since September 2008; and the flow split that is the section's point — $6.8 billion into Treasuries against $181.4 billion into U.S. equities in the same month. Secondary account: Traders Union, https://tradersunion.com/news/financial-news/show/3015937-foreign-us-treasuries-decline-june/ https://www.aol.com/articles/foreign-holdings-us-treasuries-fall-210553000.html ↩
  9. Foreign Holdings of US Treasuries Fell to Nine-Month Low in July — Bloomberg, September 16, 2026 — The July continuation: another $50 billion decline to roughly $9.25 trillion, a nine-month low, with France and Canada leading the reductions rather than the usual suspects — which is part of why this piece reads the change as compositional rather than as a retreat from American assets. https://www.bloomberg.com/news/articles/2026-09-16/foreign-holdings-of-us-treasuries-fell-to-nine-month-low-in-july ↩
  10. Japan's Benchmark Bond Yield Rises to 3% for First Time Since 1996 — Reuters via Yahoo Finance, September 1, 2026 — The fact that changes the Japanese institution's calculation, and with it the price Washington has to pay for the marginal dollar: the 10-year JGB at 3% for the first time since September 1996, on inflation, fiscal concern and pressure on the Bank of Japan to tighten faster. https://finance.yahoo.com/economy/policy/articles/japans-benchmark-bond-yield-rises-063338944.html ↩
  11. S&P 500 Closes at a Record — CNBC, August 6, 2026 — The record close of 7,757.64 on August 7, 2026, which the piece later pairs with the 10-year already near 4.7% — the configuration it reads as competition for capital rather than capitulation. https://www.cnbc.com/2026/08/06/stock-market-today-live-updates.html ↩
  12. Nasdaq Composite Notches First Record Close Since June — CNBC, September 20, 2026 — The second half of the contradiction: the Nasdaq returning to record territory on a semiconductor rally with the 10-year around 5%. Equities setting records into a rising risk-free rate is the observation the entire clearing-price reading is built to explain. https://www.cnbc.com/2026/09/20/stock-market-today-live-updates.html ↩
  13. August 2026 Review and Outlook — Nasdaq — The maturity profile that makes hyperscaler issuance a Treasury story rather than a credit sideshow: more than half of 2026's mega-cap tech bond financings carrying maturities of 10 to 50 years, competing for the same duration buyers as the long end. The same source covers the Treasury's reliance on bill issuance and buybacks to manage funding, which the caveat section returns to. https://www.nasdaq.com/articles/august-2026-review-and-outlook ↩
  14. Treasury International Capital Data for July 2026 — U.S. Department of the Treasury — The primary release behind the split this section calls the whole argument: $83.7 billion of total net inflows, foreign official institutions buying $44.4 billion of long-term securities against private foreign net sales, and, after adjusting for stock swaps associated with acquisitions, estimated overall net foreign sales of long-term securities of $27.9 billion. The $38.8 billion increase in foreign holdings of Treasury bills is reported at FX.co, https://www.fx.co/en/forex-news/3169790 https://home.treasury.gov/news/press-releases/sb0631 ↩
  15. Treasury Doubles Buybacks of 10- to 30-Year Debt — CNBC, August 20, 2026 — The thumb on the scale this section concedes: the Treasury doubling buybacks specifically at the 10- to 30-year part of the curve to calm the long end. It is the reason the piece argues the unmanaged clearing yield probably sits above the quoted one. https://www.cnbc.com/2026/08/20/stock-market-today-live-updates.html ↩
  16. The Clearing Price — The same move, run on a different asset class. That piece argued private marks were not breaking but searching — Airtable growing revenue 20% and selling at 89% off was price discovery, not failure, and the only live questions were how fast and who absorbed it. This piece applies the identical reading to the most liquid market in the world: a price that looks like an alarm is a market still looking for its buyer. ↩
  17. The Marginal Buyer Is a Carry Trade — The dedicated treatment of exactly the oversimplification this sentence is objecting to. That piece established the compositional handoff — who stopped buying matters more than the headline total, and the buyer who replaced them holds for different reasons and on a different horizon. The July TIC split described below is that argument's next data point. ↩
  18. Valuation Gravity: Why 2022–2028 Could Mirror Historic Compression Cycles — The Compression Series entry that named the structural change this sentence is pointing at. Its argument is that once capital has a price again, multiples compress — and the clearing yield described here is that price, arrived at from the supply side rather than the valuation side. The last decade's capital-market structure is precisely what that piece said was ending. ↩
Back to the Journal