Since last year, I have been tracking U.S. Treasury yields, federal debt, commodity prices, and the Federal Reserve’s balance sheet as parts of the same system.
I began with a hypothesis. We had entered an era in which public investment and industrial policy could become crucial in the race for AI leadership. Semiconductors, data centers, power generation, transmission networks, and strategic manufacturing capacity would all require enormous investment. Higher public debt—and at least some inflationary pressure—appeared difficult to avoid.
But I also wondered whether the resulting increase in nominal GDP might create some room in the debt ratio. If the price level, incomes, and nominal output rose faster, the existing stock of fixed-rate debt could become smaller relative to the economy supporting it, even while the nominal debt total continued to climb.
That possibility contained a contradiction. The inflation that made yesterday’s debt look smaller could also push up Treasury yields and make tomorrow’s debt much more expensive.
Those broader observations led me to this analysis. It seemed like the right time to examine the issue in depth, even though the central question required bringing together an unusually wide range of data: debt, inflation, nominal GDP, Treasury maturities, refinancing costs, federal interest expense, and productivity.
Since November 2022, I have been watching one question more closely than almost any other macroeconomic development: Is the AI revolution actually raising U.S. productivity?
It may help determine whether the United States can reduce inflation without sacrificing real growth, whether the Federal Reserve can eventually justify lower policy rates, and whether the economy can grow fast enough to contain a rapidly rising federal interest burden.
In an earlier analysis, I argued that a Federal Reserve pause—or even a rate cut—would not necessarily pull down the 10-year Treasury yield. The long end responds not only to the federal funds rate, but also to expected inflation, real growth, Treasury supply, and the term premium investors demand for holding long-duration debt.
The current yield curve shows how widely short- and long-term financing conditions can diverge.
On August 21, 2026, the effective federal funds rate was 3.63%, while the 10-year Treasury yield was 4.74% and the 30-year yield was 5.27%. The 10-year inflation-protected Treasury yield—a market-based measure of the real yield available on long-term government debt—was 2.40%. The 10-year yield therefore stood 1.11 percentage points above the effective federal funds rate. Federal Reserve H.15 data confirm that the price of long-term capital remained substantially higher than the overnight policy rate.https://www.federalreserve.gov/releases/h15/
That gap demonstrates the separation between short- and long-term financing conditions. By itself, however, it does not prove that earlier Fed cuts caused long yields to remain high.
The central question is no longer whether higher yields create pressure. It is how much pressure the United States can absorb—and for how long—before the incentives facing policymakers begin to change.
Inflation Helped the Old Debt—but Made the New Debt More Dangerous
The post-pandemic inflation shock did not affect every part of America’s debt burden in the same way.
For nominal debt that had already been issued at low fixed rates, inflation provided real relief. The principal did not rise with the price level, but nominal GDP, incomes, and tax receipts did. That made the existing debt stock smaller relative to the nominal economy supporting it.
The GDP implicit price deflator increased from 104.516 in the fourth quarter of 2019 to 133.805 in the second quarter of 2026—a cumulative increase of 28.0%. The BEA series published through FRED therefore confirms that the dollar size of the economy was lifted by far more than real growth alone. https://fred.stlouisfed.org/series/gdpdef
What would have happened if the price level had instead risen at exactly 2% per year?
Over the same six-and-a-half-year period, a 2% annual path would have raised the price level by approximately 13.7%. Holding real GDP and the debt stock unchanged, nominal GDP in the second quarter of 2026 would have been about 11.2% smaller than it actually was.
For scale, applying that smaller denominator to approximate 2026 debt ratios produces the following static calculation:
This is a scaling exercise, not an alternative-history forecast. Lower inflation would also have changed interest rates, federal spending, revenues, deficits, and the eventual debt stock. The calculation isolates only the denominator effect.
On that narrow basis, the post-2019 price-level increase lowered the measured public-debt ratio by roughly 13 percentage points of GDP and the gross-debt ratio by roughly 15 percentage points relative to a steady 2% price path.
The conclusion is limited but important: inflation materially reduced the real burden of low-rate nominal debt that had already been issued.
The problem is that the Treasury must keep issuing new debt.
Once inflation raises expected inflation and market interest rates, old low-coupon securities mature and new securities must be sold at much higher yields. The initial relief to the denominator is gradually replaced by a lasting increase in the government’s interest bill.
Inflation helped dilute yesterday’s debt. It raised the price of financing tomorrow’s debt.
The Interest Clock Is Moving Faster Than the Economy
Gross federal debt passed $40 trillion in August 2026, according to the Treasury’s Debt to the Penny data. Debt held by the public—the measure more relevant to financial markets and the federal budget—is projected by the Congressional Budget Office to reach approximately 101% of GDP in fiscal 2026. https://fiscaldata.treasury.gov/datasets/debt-to-the-penny/debt-to-the-penny
The more immediate warning comes from interest expense.
The comparison with 2019 shows how quickly the burden has changed:
The CBO’s fiscal-year 2019 review supplies the pre-pandemic reference point, while its 2026–2036 Budget and Economic Outlook provides the latest baseline and 2026 projection.
Between 2019 and 2025, net interest outlays increased by approximately 159%. Nominal GDP increased by roughly 43%, but net interest rose from about 1.76% to approximately 3.2% of GDP. In other words, the interest burden’s share of the economy increased by roughly 80% even after inflation had substantially enlarged the nominal-GDP denominator.
That is the central numerical finding of this analysis.
In 2019, roughly 11 cents of every federal revenue dollar went to net interest. By 2025, the figure was close to 19 cents.
CBO projects net interest above 3.2% of GDP in every year from 2026 through 2036—the highest sustained burden recorded since at least 1940. By 2036, the ratio reaches 4.6% of GDP and net interest outlays reach approximately $2.1 trillion.
Nominal GDP growth has prevented the debt and interest ratios from looking even worse. It has not stopped the interest bill from rising much faster than the economy and the federal revenue base.
Inflation made the denominator larger first. Refinancing is now making the numerator catch up.
Higher Rates Arrive One Maturity at a Time
The United States does not refinance its entire debt stock in a single year.
That delay is one of its greatest protections—and one of the reasons the danger can be underestimated.
Long-dated notes and bonds slow the transmission of market rates into the government’s effective interest cost. But average maturity conceals a faster reset near the center of the portfolio.
The Treasury’s August 2026 presentation to the Treasury Borrowing Advisory Committee put the weighted-median next rate reset for privately held Treasury debt at 26.9 months, using portfolio data as of July 29. Treasury bills represented 22.2% of marketable debt as of July 31. Floating-rate notes, bills, and shorter-maturity securities reset quickly, while continuing fiscal deficits require new borrowing even before existing debt matures.
Consider a simplified example. A Treasury security carrying a 1.5% coupon matures and is replaced with a new security carrying a rate near 4.5%. The annual interest cost on the same principal roughly triples—not because the government borrowed more at that moment, but because the price of carrying the principal changed.
By July 31, 2026, the Treasury’s average interest rate on total interest-bearing debt had risen to approximately 3.45%, while fiscal-year-to-date gross interest expense had reached about $1.17 trillion. The Treasury’s Interest Expense and Average Interest Rates data show how rapidly higher market rates are working through the debt stock.
Gross interest expense is not the same as CBO’s budgetary net interest outlays. Treasury’s gross measure includes interest on both publicly held and intragovernmental debt, whereas net interest subtracts interest income received by the federal government. The two measures answer different questions, but both show that carrying costs have risen sharply.
CBO provides a useful sensitivity test. If all interest rates in its forecast—including short- and long-term Treasury rates—were 0.1 percentage point higher every year, cumulative federal deficits from 2027 through 2036 would be approximately $379 billion larger. Higher rates raise interest costs, larger interest costs increase deficits, and larger deficits require still more borrowing. CBO’s sensitivity analysis quantifies that feedback loop.
That result cannot be translated one-for-one into the effect of a higher 10-year yield alone. It nevertheless demonstrates why a persistent increase across the yield curve matters far more than a brief market spike.
Where Is the Economic Pain Threshold?
There is no official ceiling for the 10-year Treasury yield.
The Federal Reserve does not target it, and the Treasury does not announce a maximum acceptable borrowing rate. The relevant threshold is therefore not a legal or mechanical cap. It is the point at which higher yields create enough simultaneous pressure on the federal budget, housing, corporate credit, and financial markets to alter the incentives facing policymakers.
CBO’s February 2026 central projection placed the 10-year Treasury yield around 4.1% in 2026 and 4.3% in 2027. CBO also estimated a roughly two-thirds probability that the 2026 annual average would fall between 3.5% and 4.8%.
The 4.74% daily yield recorded on August 21 stood near the upper end of that interval, although a single-day observation is not directly comparable with CBO’s annual-average range.
The following ranges are a GMS analytical pressure map. They are not official thresholds or forecasts of an automatic policy response.
This framework depends critically on why yields are rising.
A 5% yield caused by stronger real growth and higher productivity is not economically identical to a 5% yield caused by unanchored inflation expectations, weak Treasury demand, or a rising fiscal-risk premium. The second case is more dangerous because it raises financing costs without creating an equally strong increase in sustainable real output or tax capacity.
A brief move above 5% would affect asset prices and borrowing conditions but would not instantly reprice the entire federal debt stock. Several years near 5% would be far more consequential because each auction, new deficit, and maturing security would transmit the higher rate into actual cash interest expense.
The best current judgment is therefore not that 5% is an inviolable ceiling. It is that a 10-year yield sustained between approximately 5.0% and 5.5% would create rapidly increasing economic and fiscal pressure—and much stronger incentives for a policy response.
Why Fed Cuts May Fail to Bring Down the 10-Year Yield
The nominal 10-year Treasury yield can be simplified into three components:
the expected path of real short-term interest rates
expected inflation
a term premium for bearing duration and inflation risk.
Equivalently, it can be viewed as the expected path of nominal short-term rates plus a term premium.
Fiscal deficits, Treasury issuance, quantitative tightening, investor demand, and market risk tolerance can affect real rates and the term premium. The 10-year yield is therefore not a longer version of the federal funds rate.
If the Fed cuts because inflation is falling, labor demand is softening, and policy has become unnecessarily restrictive, the expected path of short rates may decline and the 10-year yield may follow.
But if investors interpret the cut as tolerance of above-target inflation, fiscal accommodation, or pressure on central-bank independence, expected inflation and the term premium can rise. The long yield may then fall much less than the policy rate—or even rise.
The August 2026 curve illustrates the separation, not a simple causal verdict. The effective federal funds rate was 3.63%, yet the 10-year yield was 4.74% and the 30-year yield was 5.27%. Lower short rates alone were not sufficient to make long-term capital cheap.
How America Can Push Back
No single institution controls the entire Treasury curve. The United States has several policy tools, but they operate through different channels and carry different risks.
Federal Reserve
Policy-rate cuts can lower bill yields and the expected path of short-term rates. Slowing or ending quantitative tightening can reduce the amount of duration the private market must absorb. In a severe market disruption, asset purchases can restore Treasury-market functioning.
But rate cuts cannot guarantee a lower 10-year yield. Large-scale purchases used primarily to suppress government borrowing costs could raise inflation expectations and damage policy credibility if they were deployed without a genuine market-functioning or macroeconomic justification.
U.S. Treasury
The Treasury can alter the mix between bills and longer-term coupons, conduct buybacks to improve liquidity, and manage auction sizes and maturity composition.
Issuing more bills can reduce near-term pressure on long-duration markets, but it shortens the refinancing clock and makes the government more sensitive to future policy rates. Treasury buybacks can improve liquidity in older securities; they are not equivalent to the Fed creating reserves to purchase assets and should not be described as quantitative easing.
Fiscal Policy
Credible reduction of primary deficits is the most direct durable response to a fiscal term premium.
Lower primary deficits reduce the amount of new debt the market must absorb, slow the compounding of interest expense, and improve confidence in the future debt path. It is also the most politically difficult option.
The response to sustained yield pressure would therefore probably be layered rather than dramatic: lower short rates when inflation and employment conditions permit, changes to balance-sheet policy if appropriate, careful debt management, and a more credible fiscal path.
That would be policy resistance—not necessarily formal yield-curve control.
AI Productivity Can Strengthen the Fed’s Reason to Cut
This is where the productivity question becomes decisive.
The healthiest way to improve debt dynamics is not to generate more inflation. It is to increase real output.
Two economies can both produce 5% nominal growth: one through 1% real growth and 4% inflation, the other through 3% real growth and 2% inflation.
The second path is more sustainable. The tax base and GDP denominator still grow, but real incomes are stronger, inflation expectations are easier to anchor, and the Fed has more room to normalize restrictive policy without relying entirely on economic weakness.
The productivity evidence since late 2022 is increasingly important, but it must be interpreted carefully.
Using the fourth quarter of 2022 as a benchmark—not as proof of causation—U.S. nonfarm-business real output increased by approximately 10% through the second quarter of 2026, while hours worked rose by less than 1%. Output per hour increased by 9.1%, equivalent to roughly 2.5% per year. These calculations use the latest preliminary BLS productivity indexes.
Over the current business cycle, from the fourth quarter of 2019 through the second quarter of 2026, nonfarm-business productivity grew at an annualized rate of 2.1%, compared with 1.5% during the previous cycle from 2007 through 2019.
In the second quarter of 2026, productivity was 2.2% higher than a year earlier while unit labor costs rose only 1.4%. The BLS Productivity and Costs release explains why that combination is important: productivity growth offsets increases in hourly compensation when calculating the labor cost of producing one unit of output.
There is also strong causal evidence that generative AI raises productivity in specific tasks. A large customer-support study found that access to an AI assistant increased issues resolved per hour by an average of 14%, with substantially larger gains among less-experienced workers. The NBER study, Generative AI at Work, demonstrates that the productivity mechanism is real at the task level.
The bridge from task-level gains to the national economy is less certain.
St. Louis Fed researchers estimated that reported AI-related time savings were equivalent to 1.6% of all work hours in 2025. Their production-model calculation suggested that generative AI may have raised aggregate labor productivity by as much as 1.3% since ChatGPT’s release. Industries reporting greater AI-related time savings also tended to record stronger productivity growth relative to their pre-pandemic trends. The researchers explicitly cautioned that the industry relationship was correlational, not causal. Their State of Generative AI Adoption in 2025 nevertheless provides one of the clearest links yet between workplace adoption, reported time savings, and aggregate productivity.
CBO has also begun incorporating AI into its macroeconomic projections. Its February 2026 central estimate assumed that AI-related productivity gains would add approximately 0.1 percentage point per year to economic growth on average. But CBO also emphasized that faster productivity can raise investment demand and interest rates.
In a separate high-growth scenario—not an AI-only scenario—CBO estimated that much stronger real GDP growth could reduce cumulative deficits by approximately $1.3 trillion over a decade because revenues would be about $2.5 trillion higher. At the same time, net interest outlays would be about $1.0 trillion higher, partly because stronger growth and investment would lift interest rates. CBO’s analysis of alternative outcomes captures the central tension: productivity improves the fiscal denominator and revenue base, but it does not automatically produce lower bond yields.
The evidence does not prove that AI caused all of America’s post-2022 productivity acceleration. Capital investment, labor reallocation, business formation, and pandemic-related distortions are crucial.
Nor does faster productivity independently trigger a Fed rate cut. If productivity raises the economy’s equilibrium real interest rate, the neutral policy rate may also be higher.
The narrower conclusion is stronger and more defensible:
If sustained productivity growth allows wages and real output to rise without generating equivalent unit-cost inflation—and if monetary policy remains restrictive—it can strengthen the Federal Reserve’s economic justification for lowering policy rates.
AI does not need to control the bond market. It needs to help create a combination of stronger real growth and lower inflation pressure that permits monetary normalization.
The Policy Routes Are Not Equally Sustainable
America has four broad routes through its debt-and-interest-rate problem.
Route One: Inflation
Inflation reduces the real value of old nominal debt and enlarges nominal GDP.
The post-pandemic period demonstrates that the denominator effect can be substantial. But it is self-limiting. Persistent inflation raises yields, increases the cost of new borrowing, lifts inflation-indexed federal spending, and eventually causes interest expense to catch up with the larger denominator.
Inflation provided temporary relief. It is not a durable exit strategy.
Route Two: Monetary and Debt-Management Support
Lower policy rates, balance-sheet policies, and changes to debt management can reduce borrowing costs, improve market functioning, or shift refinancing risk through time.
Used carefully, these tools can prevent unnecessary financial tightening or repair a dysfunctional Treasury market. Used primarily to finance the government cheaply, they become a form of financial repression and risk unanchoring inflation expectations—potentially increasing the term premium they are intended to suppress.
Route Three: Productivity and Real Growth
Faster productivity raises potential output, supports real incomes, and enlarges the tax base without requiring permanently higher inflation.
This is the healthiest economic route, but it is not a complete solution. Even a powerful AI productivity cycle cannot indefinitely offset large primary deficits, rising mandatory spending, and compounding interest costs.
Route Four: Fiscal Adjustment
Productivity can enlarge the denominator, but fiscal policy determines how quickly the numerator grows.
A credible reduction in primary deficits is the only route that directly slows both new borrowing and the compounding of future interest costs. The arithmetic is straightforward; the politics are not.
The most sustainable path therefore combines stronger real productivity with a fiscal trajectory that does not consume the resulting gains.
America Is Already Under Pressure - Perhaps 5% Is Not a Trigger
The United States does not face a single yield at which the Treasury market suddenly becomes impossible to finance.
It faces a moving threshold determined by inflation, real growth, debt maturity, fiscal deficits, investor demand, and the length of time yields remain elevated.
At 4.74% for the 10-year and 5.27% for the 30-year on August 21, the market was already delivering meaningful fiscal and economic pressure. Those rates were not evidence of a funding crisis. But they were high enough to keep long-term borrowing restrictive and to accelerate the refinancing of federal debt at rates well above the low coupons issued before and during the pandemic.
A sustained 10-year yield between approximately 5.0% and 5.5% would move the system into a more severe pressure zone. It would not force automatic Fed cuts, Treasury intervention, or yield-curve control. It would materially strengthen the incentives for the Fed, Treasury, and elected government to pursue policies that reduce long-term financing pressure.
The central constraint is credibility.
America cannot sustainably inflate away its debt if the bond market responds by demanding a still higher yield on every new dollar of borrowing. The Fed can lower short-term rates when inflation, labor demand, and unit-cost data permit. The Treasury can manage maturity composition and market liquidity. Elected officials can reduce primary deficits.
And if the AI revolution produces durable productivity growth, it may strengthen the Fed’s justification for lowering restrictive policy rates without relying on a severe recession.
That is a possible part of the escape route—not the entire solution.
Not more inflation.
Not permanent financial repression.
More real output from each hour worked, combined with a fiscal path that does not consume the gain.
The question is no longer whether the debt clock is moving.
It is whether America’s productivity and fiscal clocks can move fast enough to keep the interest clock from taking control.