NextFin News - The 30-year US Treasury yield briefly touched 5.31% in mid-August, its highest level since June 2007, and the move was not an American quirk. Long-term government bond yields in Japan, Germany, France and the United Kingdom also jumped to multi-year and multi-decade highs, shaking the market that central banks, pension funds and households still treat as the bedrock of the global financial system.
Here is the puzzle. Inflation is cooling, not accelerating. US consumer prices rose 3.4% in July, down from 3.5% in June, and the Federal Reserve held its policy rate steady at 3.50%-3.75%. Yet the long end of the bond market is pricing as if something structural has broken. The explanation that best fits the data is not a cyclical inflation scare but a supply-and-demand regime shift: governments are issuing record debt, technology giants are borrowing hundreds of billions to finance artificial intelligence infrastructure, and the buyers of that debt have changed.
The Move: A Global Repricing of Long-Duration Risk
The numbers are stark. The benchmark 10-year Treasury yield climbed to 4.72% in mid-August, up from about 4.2% at the start of 2026, while the 30-year yield surged past 5.3% before easing to roughly 5.19% after the US Treasury announced it would expand its buyback program. In Japan, the 10-year government bond yield reached 2.945% on August 18, the highest since September 1996, and stood at 3.00% at the start of September, up 1.39 percentage points from a year earlier. Long-term yields also hit multi-year highs in Germany, France and the United Kingdom, where inflation at 2.9% remains above the central bank's target.
Bond yields rise when prices fall, so the direction of travel is unambiguous: investors are demanding higher compensation to hold long-dated government debt. That matters because the 10-year Treasury is the reference rate for mortgages, auto loans and credit cards, and the 30-year yield anchors the cost of capital for utilities, insurers and infrastructure projects. When that anchor lifts, borrowing costs lift with it across the economy. The average 30-year mortgage rate, which shot above 6% in 2022, has stayed there for four years; another leg higher in long yields puts homeownership further out of reach for marginal buyers and raises the refinancing cost for every borrower that locked in cheaper debt during the low-rate era.
Why Inflation Is Not the Whole Story
The first explanation investors reach for is inflation. It is partly right, but only partly. US inflation at 3.4% remains above the Federal Reserve's 2% target, and the Middle East conflict is pushing up fuel and utility costs, as the Bank of England has acknowledged. Energy shocks feed through to headline prices, and bondholders rightly demand protection against that erosion. The ongoing conflict has also prompted US officials to prepare additional sanctions, keeping the geopolitical risk premium alive.
But the inflation story does not explain the shape of the move. The 2-year Treasury yield, which tracks expectations for the Fed's policy rate, has been trending lower since late July even as the 30-year yield surged. The spread between the 2-year and 30-year yields widened by 18 basis points in five trading days during the week of July 27, a bear-steepening move that signals investors are pricing long-run risk rather than near-term inflation. If investors feared a broad-based inflation breakout, the entire curve would rise together. Instead the curve is steepening at the long end only. That is the signature of a term-premium repricing, not a pure inflation repricing.
The term premium is the extra yield investors require to hold a 10-year bond instead of rolling short-term bills. According to the San Francisco Fed's Christensen-Rudebusch model, the 10-year term premium rose from 1.26% on August 18, 2025 to 1.37% on August 17, 2026, while the observed 10-year yield rose from 4.39% to 4.81% over roughly the same window. Decomposing the move, about 30 basis points came from higher expected future short rates and 11 basis points from the term premium. That split matters: the expected-rate component can fall back if the Fed cuts, but the term-premium component only falls if investors become more willing to hold long-duration risk, and that willingness depends on the supply-and-demand balance, not on the next inflation print.
The Supply Shock: Deficits Meet the AI Borrowing Boom
On the supply side, two borrowers are competing for the same pool of long-duration capital. The first is the US government. Total federal debt crossed $40 trillion in mid-August, roughly double the level of a decade ago, when it stood at $19.4 trillion. The Treasury reported a $432.3 billion deficit in July, the largest monthly shortfall in more than five years, and the deficit for the 10 months through July widened by $170 billion from the same period a year earlier, to about $1.8 trillion. Interest on the debt has reached nearly $1.2 trillion this year, making debt service the largest federal expenditure outside Social Security and Medicare. Total federal debt stood at 123% of GDP in the first quarter of 2026, down from a pandemic peak of 133% in the second quarter of 2020 but still far above any peacetime norm.
The second borrower is Big Tech. For most of the past decade, the large technology companies leading the AI build-out funded their investment from operating cash flow, said Lucas Baynes, a senior investment strategist at Vanguard. "That era is ending." The largest AI companies have sold more than $200 billion of bonds so far this year, and overall US corporate bond issuance reached about $1.7 trillion through July, 27% above the pace of a year earlier, according to Securities Industry and Financial Markets Association data. Barclays expects total corporate issuance to hit $2.46 trillion in 2026, up 11.8% from 2025, with net issuance rising 30.2%.
This is the mechanism that many investors have under-appreciated, and it has a name: reverse crowding out. In the textbook version of crowding out, government borrowing pushes up rates and squeezes private borrowers. Here the sequence runs the other way. Hyperscalers, data-center real estate trusts and the power producers that feed them are issuing debt at a pace that rivals sovereign borrowers. Every bond a hyperscaler sells to finance a GPU cluster or a data center competes directly with Treasuries for the same institutional capital. Corporate bonds must offer a premium over the risk-free rate to attract buyers, and when the volume of that issuance swells, the premium widens. But the premium does not stay contained in corporate credit: to keep Treasuries competitive, the risk-free benchmark itself has to rise.
Analysts at Barclays put the mechanism plainly: the rise in rates is less about inflation and more about the US budget deficit, AI-related issuance competing with Treasuries, and higher term premiums. Bank of America analysts expect the five largest hyperscalers to borrow roughly $140 billion annually over the next three years, a figure that could exceed $300 billion as spending plans are revised upward. The CEO of Constellation Energy said hyperscaler projected spending for 2026 is nearly 75% higher than last year and continues to be revised upward. That is not a one-off funding gap; it is a multi-year capital program that will keep competing with the Treasury for years.
The World Economic Forum's Chief Economists' Outlook captured the dynamic in May: borrowing by governments and companies hit a record in 2025 and is set to rise again in 2026, "even as long-duration demand weakens and maturities shorten." Supply is expanding just as demand for long-dated paper is softening. That is the textbook setup for higher yields.
The Demand Side: The Buyers Have Changed
The third leg of the move is on the demand side, and it is the most structural. Barclays strategists estimate that private investors now hold about 73% of the Treasury market, up from roughly 50% a decade ago. The shift has two drivers. The Federal Reserve has been shrinking its balance sheet since 2022, removing a large, price-insensitive buyer from the market. Foreign central banks, particularly in Asia, have also pared their purchases as they defend their own currencies and manage reserves differently. Stepping into the void are mutual funds, households, banks and overseas private capital.
That change in ownership matters because the two groups behave differently. Official and captive holders bought bonds for reasons other than price: currency management, reserve adequacy, regulatory mandates. A central bank accumulating reserves does not sell because the yield dipped 10 basis points. Private investors are price-sensitive. They buy bonds because the return is attractive relative to equities, credit and cash, and they will walk away when it is not. As that buyer base has grown, the marginal buyer of US government debt has become more return-driven and less loyal. The result is a market that demands a higher term premium to absorb the same volume of supply. This is not a temporary positioning imbalance; it is a change in who owns the debt, and it will not reverse on its own.
The Policy Response: Liquidity, Not a Cure
The Treasury's response was to expand its buyback operations. Beginning September 9 and running through November 4, the maximum size of liquidity-support buybacks for 10-to-20-year and 20-to-30-year securities will at least double, from $2 billion to at least $4 billion per operation. The stated purpose is to improve liquidity in longer-dated sectors where the Treasury sees "consistent strong sponsorship from market participants." The program targets the off-the-run end of the curve, where trading is thinnest and the gap between bid and offer widens first in a selloff.
Treasury Secretary Scott Bessent was candid about the limits of the tool. He described liquidity in the 30-year sector as "very poor" and framed the buybacks as a liquidity measure rather than an attempt to suppress yields, while adding that part of the point is signaling "to show that we believe that the yields don't reflect the underlying fundamentals." The market reaction was immediate but modest: the 30-year yield eased to 5.19% after the announcement, but remained far above where it started the year.
The reason the buyback cannot fix the problem is arithmetic. The Treasury cannot create money; it must finance the buybacks by issuing other debt. The program is a liquidity patch for a specific segment of the curve, not a reduction in the overall supply of government bonds. With the federal deficit running at more than $400 billion a month and corporate issuance on track to exceed $2 trillion, the underlying supply pressure remains. Analysts have noted that if operations continue at the current pace, the expanded buybacks could amount to around $66 billion on an annualized basis, meaningful for the 20-to-30-year sector but small against a deficit that exceeds $1.8 trillion a year.
The Counter-Case: A Positioning Tantrum, Not a Regime Change
The strongest argument against the structural reading is that this is a repeat of 2022-23: a positioning-driven tantrum that reverses once the Fed signals a clear cutting cycle and issuance normalizes. In that view, thin summer liquidity and the Middle East oil spike exaggerated a move that will unwind as quickly as it arrived. Bond investors have been wrong about a bond vigilante uprising before, and the debt-to-GDP ratio is still below its 2020 peak of 133%. The 10-year yield traded above 4.5% for stretches of 2023 and 2024 before falling back, which shows that long yields can overshoot and mean-revert.
There is force in that argument. Yields can overshoot on positioning, and the 2-year yield's recent decline suggests the market does not expect the Fed to keep tightening. If inflation continues to drift toward 2% and the Fed cuts rates decisively, the short end will pull the long end lower, as it did after the 2023 regional banking stress. A growth scare that forces the Fed to cut faster than priced would also compress the term premium, at least temporarily. And the term premium itself is a model-dependent estimate, not an observed price: a composite of the New York Fed, San Francisco Fed and Kim-Wright models puts the 10-year term premium at 0.96% as of late August, below the 1.0% threshold. If that lower estimate is closer to the truth, the "structural term-premium re-rating" story is materially weaker than it looks.
But two differences make this cycle distinct. First, the buyer-base shift is real and measurable: private ownership at 73% versus 50% a decade ago is not positioning, it is ownership structure. Positioning is how much of your allocation you hold; ownership is who sits on the register. The second difference is that the corporate borrower is not a cyclical issuer filling a temporary funding gap. AI infrastructure is a multi-year capital program, and hyperscaler borrowing of roughly $140 billion a year, potentially exceeding $300 billion, is a persistent addition to supply. A tantrum ends when positioning flushes out. A regime shift ends only when supply and demand re-balance, and there is no visible catalyst for that re-balancing in the next year. The Fed is shrinking its balance sheet, not expanding it; foreign official demand is not returning; and the fiscal deficit shows no sign of consolidation.
What to Watch
Three signals will determine whether this is a cyclical spike or a durable re-rating. First, the 10-year term premium under the San Francisco Fed's model: if it falls back below 1.0% while issuance stays elevated, the structural thesis is weakened. The model put it at 1.37% on August 17; a sustained move toward 1.0% would signal that investors are again willing to hold long bonds without a large risk premium. Second, the 30-year yield: a sustained close below 4.75% would suggest the move was a positioning overshoot. Third, the Fed's path: a clear cutting cycle combined with a credible fiscal consolidation plan would pull long yields down; a Fed on hold alongside persistent deficits would keep them elevated.
The Federal Open Market Committee meets September 15-16, with rate futures pricing roughly a 62% probability that the Fed holds rates steady and the remainder on a quarter-point hike. The next Quarterly Refunding announcement on November 4 will reveal whether Treasury issuance plans are being revised. And the November 4 expiry of the expanded buyback program will test whether liquidity support was enough to stabilize the long end.
The Bottom Line
For investors, the implication is that the old playbook, in which bad economic news is good news for bonds, is less reliable than it was. When the driver of higher yields is a term-premium repricing and a supply glut rather than inflation expectations, a growth scare may not produce the rally it once did, because the same deficits that make growth look weak also make the supply of bonds look large. The beneficiaries of this regime are holders of short-duration assets and floating-rate instruments, who can reinvest at higher yields without taking duration risk, and creditors with pricing power. The exposed are long-duration bondholders, who face mark-to-market losses as yields grind higher, and any borrower that must refinance long-term debt: the federal government first, then utilities, telecoms and data-center developers whose capital intensity makes them the most rate-sensitive corners of the corporate sector.
Short term, yields can still whipsaw on liquidity and positioning, and the buyback program may cap the worst of the overshoot into autumn. Medium term, yields will be set by the balance of sovereign and corporate issuance against a more price-sensitive buyer base, which points to a higher-for-longer equilibrium than the market enjoyed in the 2010s. Long term, the question is whether the term premium has entered a new, higher regime. The evidence, the buyer shift, the AI borrowing wave, the persistent deficits, points to yes.
"30-year yields at their highest since before the financial crisis are not a footnote to the equity story. They're a warning about the true cost of government borrowing," said Nigel Green, chief executive of the deVere Group.
The bond market is not shouting about inflation. It is pricing a world in which governments and technology giants borrow on a scale that requires investors to be paid more for the risk of lending to them. This is the market pricing the deficit and the data center, not a cyclical dip, and Treasury buybacks will not talk it away.
Explore more exclusive insights at nextfin.ai.
