Corporate buybacks rise as long-term bond buyers disappear from market
Companies are buying back record amounts of their own stock while the government struggles to attract buyers for its long-dated debt, exposing opposing forces in U.S. markets.
On Tuesday, August 18, the Treasury Department ran a scheduled operation to buy back $2 billion of its own twenty- and thirty-year bonds, an effort meant to keep a quiet corner of the market trading.
The thirty-year yield rose anyway, touching 5.34 percent, the highest for the long bond since 2007, according to US News on August 19. Two days later, Secretary Scott Bessent announced plans to double the size of those buybacks to at least $4 billion each and to run four of them a quarter starting September 9, as 24/7 Wall St. reported on August 21. Yields fell for exactly one session.
By Thursday, the ten-year was back up to 4.70 percent, AP wrote on August 20, and when Bessent told investors the selloff was a temporary mispricing, the thirty-year climbed from 5.19 to 5.23 percent during his remarks, as reported by 24/7 Wall St. on August 21. A government bidding for its own bonds in size and watching the price fall anyway is not a liquidity problem. It is a buyer shortage.
On the equity side of the same corporate ledger, companies in the S&P 500 are repurchasing their own shares at an annualized pace above $1 trillion in 2026, with full-year authorizations tracking toward roughly $1.2 trillion—a record, reported Cardano Capital on July 7 and Financial Content on March 12. Salesforce alone authorized $50 billion of buybacks in February, according to Financial Content on March 30. So the same economy produces two opposite facts at once: corporate America is the largest reliable buyer of U.S. equities on record, and the buyers of long-dated U.S. debt have thinned out enough that the Treasury itself had to step in.
Bessent pointed to part of the cause in a CNBC interview, noting that the long end was thinly traded, especially in August, while Treasuries compete with heavy corporate bond sales, including borrowing for AI data centers, according to Business Times on august 20. That competition is substantial. US investment-grade issuance hit a record $1.68 trillion this year, Mezha Media reported in August 2026, and AI-related corporate borrowing is approaching $400 billion of that, according to Economy.ac on July 10.
Amazon, Alphabet, Meta, Nvidia, Oracle, and SpaceX sold $182 billion of bonds in 2026 against $13 billion last year, as recorded by 24/7 Wall St. on July 10.
Each of those deals pays investors more than a Treasury of comparable maturity, so every deal pulls a pension fund or insurer away from the long bond.
Underneath the week's drama is slower arithmetic. Federal debt crossed $40 trillion this month, LiveNewsChat reported in August 2026. Inflation has run above the Federal Reserve's target for about five years, according to Yahoo Finance and Bloomberg in August 2026. Washington must keep selling long-dated paper into a market where the traditional big holders—foreign central banks and commercial banks—own less than they used to. The trigger this week was a poor auction backdrop and a sudden announcement. The pressure is that whoever is asked to lock up money for thirty years now demands close to 5.3 percent to do it, and no announcement changes what they demand.
A British comparison
The comparison that cuts is the Bank of England in September 2022. A fiscal announcement broke the gilt market; the Bank pledged to buy long-dated gilts, and the purchases halted the spiral only because the government simultaneously reversed the budget that caused it. Intervention bought time; policy changed the price. Bessent has announced intervention without anything resembling a policy change, which is why the market treated his program as diagnosis rather than cure.
The honest counterargument
One honest counterargument says otherwise: the Fed's Operation Twist in 2011 also swapped short instruments for long ones without adding debt, and long rates measurably fell. On paper Bessent's program resembles Twist. The difference is what each operation was compensating for. Twist smoothed an allocation problem among abundant willing buyers; this program stands in front of missing ones. Same mechanics, opposite diagnosis, and the market spent two sessions pricing the second diagnosis over the first.
Follow the mechanism forward and someone always pays. First order. The Treasury locks in today's long rates as maturing coupons get refinanced, converting a temporary yield spike into permanent interest expense on a $40 trillion stack.
Second order. Companies that borrowed cheaply to repurchase stock face costlier refinancing, and the arithmetic that made buying back shares a bargain at 3 percent stops working near 5.5 percent, which would drain the largest standing bid under the equity market precisely when it is most relied upon. Third order. Mortgage rates and commercial-property loans reset off these long yields, pushing the bill toward households and regional lenders who never traded a bond in their lives.
Who profits meanwhile? The holders of new-issue corporate credit. Record supply plus reluctant demand lets investors dictate terms, reporting shows bond buyers already pushing back on issuer-friendly covenants and pricing (Remio, Aug 2026). Insurers and pension funds locking decades of guaranteed returns above 5.5 percent from AI-capex borrowers benefit quietly from this repricing, paid to do what they were doing anyway. The losers are the issuers of everything long-dated, sovereign first and equity-heavy corporations second.
The August objection
The honest objection runs the other way. Maybe Bessent is right and this is just thin August trading amplified by algorithms, with September bringing a reopening of full desks, restoring demand and pulling the thirty-year back toward 5 percent. If corporate issuance tapers after Labor Day and auctions tighten, this piece ages badly within a month. That possibility is real, and the next few auction calendars will settle it.
Two September outcomes
So watch the sequence. If the read is right, September and October long-bond auctions go badly even with the doubled buybacks. Foreign buyers keep skipping, and the price still must be discounted to clear, because $4 billion per operation is rounding error against a refinancing calendar measured in trillions. If instead the doubled operations coincide with steadily falling long yields through October, the liquidity explanation wins and the buyer-shortage story was wrong.
When the last dependable buyer of American long-term debt is the American government itself, the price of that discovery does not stay inside the bond market. It travels, at coupon speed, to everyone who borrows over five years.