Everything About AI
NextFin News - The world's government bond market is staring at a 4% yield for the first time since 2007, and the U.S. 30-year Treasury has climbed to its highest level since 2004, as a synchronized global selloff forces investors to reprice the cost of sovereign debt across every major economy. The 10-year U.S. Treasury yield reached 5.23% on September 28, up 1.09 percentage points from a year earlier, while Japan's benchmark yield touched 3.10%, its highest since 1996, and Germany's 10-year bund sat at 3.62%. This is not an American problem with foreign echoes. It is a global repricing of the price of money, and it is happening because three forces have collided at once: a Federal Reserve that has turned back to hiking, governments that cannot stop borrowing, and an artificial-intelligence investment boom that is absorbing a swelling share of the world's savings.
The selloff gathered pace after the Federal Reserve raised its policy rate by 25 basis points on September 16 to a target range of 3.75%-4.00%, the first increase since 2023, and signaled that inflation running near 3.7% would not be tolerated. Within days, the 30-year Treasury yield pushed to 5.50%, its highest level since 2004, while the 10-year note touched a 19-year high during the same week. The move rippled outward: Japan's 10-year government bond yield jumped 8 basis points to 3.055% on September 24, the highest since August 1996, and U.K. gilts also climbed to their highest levels since the global financial crisis. The combination matters because it marks the first time in nearly two decades that the three largest advanced-economy bond markets — the United States, the euro area, and Japan — have all been pushed higher together by something other than a short-lived inflation scare.
What Is Actually Moving Bonds: Supply and Demand, Not Just Inflation
The first-order explanation for any bond selloff is inflation, and the inflation story is real. Energy prices surged after Middle East tensions threatened the Strait of Hormuz, U.S. consumer prices were revised higher, and the Fed's own September projections put 2026 inflation at 3.7% — well above its 2% target. But stopping at inflation misses the deeper mechanism. Bond yields are the price of capital, and the price of capital is set by the balance between the supply of savings and the demand for investment. On both sides of that equation, the structure has changed.
On the supply side, governments are flooding the market with debt. The U.S. Treasury reported a year-to-date fiscal deficit of $1.97 trillion through August 2026, and the nonpartisan Congressional Budget Office projects the full-year shortfall near $2.1 trillion. Net interest payments alone are on track to exceed $1 trillion in fiscal 2026, roughly 3.3% of GDP and close to 14% of federal spending — money the government must borrow before it can spend anything else. Total federal debt is approaching $40 trillion. When a borrower that size must continuously refinance and add to its stock of bonds, investors demand a higher return for the risk that the debt path is not sustainable. That is not a cyclical quirk; it is a risk premium baked into every new auction.
On the demand side, the AI buildout is competing for the same pool of capital. The Federal Reserve Bank of Minneapolis calculated that capital spending by the five largest AI data-center investors — Alphabet, Amazon, Meta, Microsoft, and Oracle — is projected to approach $1 trillion by 2027, up from $200 billion in 2024. Alisdair McKay, a monetary advisor at the Minneapolis Fed, put the scale plainly:
"For reference, total private investment in the economy is about $5.5 trillion dollars. We're talking about 20 percent of investment coming from this one category."
JPMorgan has estimated that $4.1 trillion of AI-related debt will be issued through 2030. That is not a rounding error in the global savings pool; it is a structural claim on it. When the private sector's appetite for capital swells at the same moment the public sector's borrowing accelerates, the price of capital rises. That is the mechanism, and it is why this selloff is broader and more durable than the inflation scares of 2013 or 2022.
The result is visible in the neutral rate — the r-star that policymakers use to judge whether policy is tight or loose. The New York Fed's Laubach-Williams model put r-star at 1.65% in the second quarter of 2026, up from 1.36% at the start of 2025. Chip Hughey, managing director of fixed income at Truist Wealth, connected the dots:
"The likely culprits for higher R-star could be the heavy investment around the AI build-out and higher government debt levels, both of which can boost demand for capital and lift real yields."
When the neutral rate rises, every bond yield that was anchored to the old, lower r-star has to reprice upward. That repricing is what 2026 is. The 10-year Treasury's long-term average since the 1960s sits near 4.26%; today's 5.23% is not an aberration from that history so much as a return to a world in which capital is scarce again.
The Transmission Channel: Why Long Bonds Lead the Move
Notice which part of the curve is moving fastest. The 30-year Treasury, not the 2-year note, is setting the pace. The 2-year yield, which tracks the Fed's policy path most closely, sits near 4.93%. The 30-year is at 5.50%. That gap — a steepening curve — tells you the market is not just pricing a few more rate hikes. It is pricing a higher term premium: the extra return investors require to hold long-duration risk when the future path of inflation, deficits, and debt supply is uncertain.
Term premium is best understood as a fear tax on duration. When investors are confident that deficits will be tamed, that debt will be rolled over smoothly, and that the central bank will keep inflation anchored, they accept a low premium for locking money away for 30 years. When any of those three promises looks shaky, the premium expands, and long yields rise faster than short yields. That is exactly the pattern in 2026: the curve is steepening because the long end is re-rating on fiscal and supply concerns, while the short end is constrained by the Fed's measured hiking path.
The second-order effect is where this becomes a cross-asset story, not just a bond story. Higher long-term yields raise the discount rate applied to every future cash flow in the economy. Growth stocks, commercial real estate, and highly leveraged companies feel it first — their valuations compress because their earnings are furthest in the future. But the transmission does not stop there. Mortgage rates, corporate bond issuance costs, and the interest expense on existing floating-rate debt all reset higher, which slows hiring, capex outside the AI winners, and consumer spending on credit. The AI buildout is driving yields up; those same higher yields then make the rest of the economy's investment more expensive. One sector's boom becomes every other sector's headwind.
There is also a cross-border channel that most investors underweight. Japan spent three decades as the world's cheapest source of capital, with yields pinned near zero and Japanese institutions and the carry trade exporting cheap funding globally. A 10-year JGB yield above 3% — the highest since 1996 — begins to reverse that flow. Japanese insurers and pension funds, which once hunted yield abroad, now find it at home. The global pool of marginal bond buyers shrinks at the very moment supply is expanding. That is a mechanical, not a psychological, pressure on yields worldwide.
The Counter-Thesis: Positioning Overshoot or Regime Change?
The strongest case against the structural view is that the selloff has been amplified by positioning and term-premium overshoot rather than a durable shift in fundamentals. Analysts have noted that governments, seeking to avoid paying a high term premium, have been shortening the maturity of new issuance — evidence that the premium itself, not just the deficit, is the market's complaint. Mizuho economist Alex Pelle observed that the Treasury's buyback response came in smaller than some investors had hoped, implying that the market's reaction may have outrun the policy response. And history offers comfort: long yields have repeatedly spiked on inflation scares — the 2013 taper tantrum, the 2022 post-pandemic shock — only to fall back once the scare passed and growth slowed.
There is also a genuine productivity argument on the other side. If AI delivers the productivity gains its boosters promise, the same investment boom that is lifting rates today could expand the economy's supply capacity tomorrow, pushing inflation and rates lower. Griffiths at CreditSights framed it directly:
"In the near term, the impact of AI is going to be a higher R-star before potentially flipping to a lower nominal policy rate in the longer term depending on the disinflationary or even potentially deflationary effects of it."
On that reading, today's 5%-plus long yields are the cyclical peak of a capex cycle, not the new normal.
Both points deserve weight, but they do not overturn the structural call. The positioning argument explains the timing and the amplitude of the move, not its direction: positioning can overshoot in either direction, but it overshoots around a level set by fundamentals, and the fundamentals — a $40 trillion debt stock, a $2 trillion annual deficit, and a $1 trillion AI capex program — have not reverted. The productivity argument is real but operates on a longer lag than the debt math. Productivity gains take years to show up in the data; interest payments on newly issued 30-year bonds are contractually due every six months starting now. A government cannot wait for AI-driven growth to arrive before it refinances its debt. The burden comes first, the payoff later — and the bond market prices the burden today.
So the judgment is this: the cyclical leg — the inflation scare, the energy spike, the positioning squeeze — will fade, and yields may drift lower from their peaks. But the structural leg — higher r-star driven by debt supply and AI capital demand — will not revert on its own. The era of near-zero global yields that defined the years after the global financial crisis is over not because of one inflation episode, but because the balance of savings and investment has shifted. Bonds are not broken; the world they priced is gone.
Who Benefits, Who Is Exposed, and What to Watch
The impact is asymmetric, and it splits cleanly by time horizon. In the short term, the losers are the most duration-sensitive assets: long-dated growth equities, commercial real estate, and leveraged borrowers who must refinance within the next 12 to 18 months. Banks and money-market funds benefit from wider net interest margins as short rates stay elevated. In the medium term, the winners are investors who can lock in yields above 5% on high-quality sovereign bonds before the cycle turns — the opportunity that major asset managers have been highlighting as Treasury yields through the 10-year sector remain above 4%. In the long term, the beneficiaries are economies and companies that run fiscal surpluses or generate the AI productivity gains; the exposed are sovereigns with large deficits and no reserve-currency privilege.
Three scenarios frame the path ahead. The base case is higher-for-longer with volatility: the 10-year Treasury trades in a 4.75%-5.50% range as inflation grinds down slowly and deficits remain wide. The upside case for bonds — yields falling back toward 4.5% — requires two triggers: a decisive disinflation print and a credible fiscal consolidation plan from Washington. The downside case — the 10-year pushing toward 6% — triggers if a long-dated debt auction fails to clear at acceptable yields or if energy prices spike again on renewed Middle East escalation.
The falsifying signal is specific: if core PCE inflation prints below 0.2% month-over-month for two consecutive months and the 10-year Treasury yield simultaneously falls back below 4.5%, the structural-higher-r-star thesis is wrong, and this is a cyclical overshoot after all. Until then, the burden of proof lies with the bulls.
The Treasury has tried to steady the market. Secretary Scott Bessent doubled long-dated debt buybacks to at least $4 billion per operation, effective September 9 through November 4, covering the 10-to-20-year and 20-to-30-year sectors. But the program repurchases only a fraction of the outstanding debt stock, and the market read it as a signal rather than a solution. That distinction captures the whole story: no amount of buyback engineering can substitute for a credible path back to fiscal balance, and no central bank can hold long-term yields down when the neutral rate itself has moved.
Here is the uncomfortable conclusion for the 4% bond trade: yields are not high because investors are panicked. They are high because the world is finally paying the real price of capital after more than a decade of paying less than the price. Whether that price settles at 4%, 5%, or higher depends less on the next inflation print than on whether governments treat their debt as a problem to solve rather than a bill to roll.
Data as of September 28, 2026.
Explore more exclusive insights at nextfin.ai.
