The Interest Trap: Borrowing a Trillion Dollars to Service the Debt

Watch Brian’s FY2026 budget breakdown

Video by ClearValue Tax. The analysis below expands on the fiscal accounts and the debt-interest feedback loop.

The budget in six numbers

What fiscal 2026 actually looked like

The federal government finished fiscal year 2026 on September 30. That year began on October 1, 2025. We now have the final Treasury numbers, so we can work from the completed year instead of mixing a budget proposal, a forecast and a debt-clock screenshot together. ^[1]

Revenue came in at $5.403 trillion. Spending reached $7.396 trillion. The difference was a $1.992 trillion deficit. That is the annual shortfall. The national debt is the accumulated balance left from years of borrowing and other financing transactions.

Revenue increased about 3.2%, while spending increased about 5.5%. The deficit widened about 12.2%. There was more money coming in, but spending moved faster. The word primary simply means we have taken net interest out of the spending calculation.

Where the $5.4 trillion came from

Most of the revenue came from the existing tax base: individual income taxes and social insurance and retirement receipts. Corporate income taxes, customs duties, excise taxes and the remaining categories supplied the rest. The chart below gives each category the same dollar scale, which makes the size of the revenue sources easier to compare.

Customs duties produced about $173.2 billion in net receipts, roughly 3.2% of the total. For scale, that is about one-sixteenth of individual income tax receipts. Replacing a major revenue source requires another source of comparable size; announcing a new collection mechanism does not change that arithmetic.

The customs line is also a useful example of why I want the final accounts. Treasury reports gross collections of about $317.8 billion and refunds of about $144.6 billion. The budget gets the net amount. Using the gross number while ignoring the refund side would give us the wrong picture of the money available to finance spending. ^[1]

The social insurance and retirement category is broader than just Social Security and Medicare payroll taxes. It also contains unemployment insurance and other retirement receipts. I am keeping Treasury’s label here so we do not quietly turn a broad accounting category into a narrower claim.

Where the $7.4 trillion went

Social Security was the largest budget function, followed by Medicare. Net interest sat close behind Medicare and ahead of national defense. Those are large numbers even before we start talking about the borrowing they create.

Treasury lists Medicare at about $1.073 trillion and a separate Health function at about $1.026 trillion. Together, those two functions were about $2.098 trillion. The interest bill exceeded national defense, but saying it exceeded all federal healthcare would overstate the comparison. Keeping the categories consistent gives us a stronger argument.

Social Security and Medicare together were about $2.740 trillion. Comparing that sum with the broad $1.812 trillion social insurance and retirement receipt category shows a large financing demand, but it does not calculate the actuarial shortfall of either program. Medicare’s different parts use different funding arrangements, including premiums and general revenues.

The 2026 trustees project depletion of the retirement and survivors trust fund’s reserves in late 2032, with ongoing income then covering 78% of scheduled benefits. The hypothetical combined retirement and disability funds reach depletion in 2034, with 83% payable. The disability fund itself remains positive through the projection period. Medicare’s supplementary medical insurance is financed through premiums and federal contributions that adjust with its costs. Those are different financing problems inside the same federal budget. ^[6]

We are borrowing to carry the borrowing

This is the part I want people to stop and look at. We spent about a trillion dollars more than we collected before including interest. Then we spent another trillion dollars servicing the debt. We borrowed to cover both.

The Treasury does not label one bond “for interest” and another bond “for everything else.” Money moves through the government’s financing accounts. But at the consolidated budget level, the effect is the same: with the rest of the budget already short of cash, adding interest creates additional financing needs.

That is the feedback loop. A debt discussion that stops at the latest headline total leaves out the process that keeps adding to it. Interest is a current expense, and financing that expense helps create another future expense.

There are three useful percentages here. Net interest was 14.2% of total spending, 19.5% of revenue and 52.8% of the deficit. Those percentages answer different questions. Roughly one dollar out of every five the government collected was matched by net interest expense. More than half the annual shortfall was that same interest line.

What happens if we hypothetically remove interest?

Take the thought experiment literally and hold every other budget flow constant. Start with $7.395800 trillion of spending. Subtract $1.051754 trillion of net interest. We are left with $6.344046 trillion of spending against $5.403327 trillion of revenue. The remaining gap is $940.719 billion.

So yes, hypothetically eliminating this expense eliminates about a trillion dollars of additional annual borrowing. It does not retire a trillion dollars of existing principal, and it does not produce a balanced budget. The remaining primary deficit would still add debt.

I am using net interest because it belongs in the unified-budget calculation. Treasury also reports about $1.360 trillion of gross interest on Treasury securities. That includes interest credited within the government and is offset by receipts elsewhere in the accounts. Subtracting gross interest from the unified deficit would mix two accounting measures. ^[3]

We also need to keep the debt measures straight. Treasury’s securities total at the end of September was about $40.172 trillion: $32.425 trillion held by the public and $7.747 trillion held within government accounts. Those are face-value Treasury securities figures. Other published public-debt measures adjust for discounts, premiums and agency securities.

The face-value securities total increased about $2.534 trillion over the year, more than the $1.992 trillion deficit. Cash balances, intragovernmental holdings and other financing adjustments help explain why the debt stock’s change does not equal the annual budget deficit dollar for dollar. Treasury’s financing tables show those transactions. ^[1]

The interest burden has already changed

The increase is easier to see over several years than in one month’s Treasury statement. The chart tracks net interest in dollars and then expresses it as a share of revenue. That second panel shows how much room the financing bill takes up relative to the government’s incoming money. ^[4]

In the current Treasury comparison, net interest rose from about $970.4 billion in FY2025 to $1,051.8 billion in FY2026. Its revenue share increased from about 18.5% to 19.5%. The government collected more money and still had a larger share of that money matched by interest expense.

Interest did not cause the entire deterioration. Of the $217.1 billion increase in the annual deficit, about $81.4 billion came from net interest and $135.7 billion came from the larger primary deficit. That works out to roughly 37.5% and 62.5% of the widening, respectively.

There is also a useful check on the percentages. Interest’s share of the deficit slipped from about 54.7% to 52.8%, because the primary gap grew faster that year. Interest can rise in dollars and as a share of revenue while falling as a share of the deficit. That is why I am showing the underlying amounts instead of using one percentage for every claim.

How the interest loop produces compounding

We can write the process in one line. Next year’s debt equals the debt we started with, plus the primary shortfall, plus the interest that needs financing. Leave the primary gap in place and let interest rise with the balance, and the debt path bends upward.

The multiplication is where the compounding enters. If we merely added the same $2 trillion every year, the result would be a straight line. When part of the next deficit depends on the larger balance left by the previous deficit, that year’s addition can grow.

Even balancing the budget before interest does not stop nominal debt growth if all interest is still financed. With a zero primary deficit, the simplified relationship becomes D_t = D₀(1 + r)^t. At a constant positive rate, that is exponential growth. A continuing primary deficit adds another source of borrowing on top of it.

The full constant-rate expression is D_t = D₀(1 + r)^t + P[((1 + r)^t − 1) / r], where P is the annual primary deficit. We do not need a permanently rising rate for that expression to compound. Financing the interest at a positive rate is enough.

Actual government accounts are more complicated: securities mature at different times, rates move, taxes respond to the economy, spending changes and financing adjustments occur. The equation isolates the mechanism. It gives us a way to measure the feedback without pretending it fixes the future.

Separate the original interest bill from the compounding

I built three paths from the same FY2026 starting point. Each begins with $32.425 trillion of publicly held Treasury securities and keeps the primary gap at $940.719 billion a year. The paths diverge only in how they treat interest after FY2026.

The first removes net interest altogether. The second freezes annual interest at $1.052 trillion, even as debt rises. The third allows interest to scale with the previous year’s balance. For that third path, the model’s net-interest ratio is calibrated to FY2026 net interest divided by the year-end public securities balance: about 3.244%. That is a modeling ratio, not Treasury’s reported average coupon or an auction yield.

By FY2046, freezing the interest bill produces about $72.3 trillion of debt in this experiment. Letting the bill rise with debt produces about $87.3 trillion. The roughly $15.0 trillion difference isolates the additional feedback beyond simply paying the same annual interest amount for twenty years.

Against the zero-interest path, the debt-sensitive path ends about $36.1 trillion higher. That wider gap includes both the recurring base interest cost and its subsequent financing effects. Keeping those two comparisons separate lets us see what comes from the original bill and what comes from the bill growing.

What that does to the annual deficit

The same mechanism changes the annual flow. In the central illustration, net interest reaches about $1.715 trillion in FY2036 and $2.713 trillion in FY2046. The primary gap remains $0.941 trillion throughout. Annual borrowing reaches about $2.656 trillion and $3.654 trillion at those two points.

No new spending program was added to create those rising totals. The experiment held the primary deficit constant. The widening came from financing a growing interest expense. That is the part that gets lost when people discuss the debt as though it were only the sum of new policy announcements.

There is another useful result. Set the primary deficit to zero, keep the calibrated positive interest ratio and finance the interest. The debt proxy still rises to about $61.4 trillion by FY2046, compared with the $32.4 trillion starting point. A primary balance slows the path substantially; a primary surplus large enough to cover interest would be needed to keep nominal debt flat in this simplified model.

These are dollar paths under stated assumptions. A rising nominal balance is not automatically a rising debt-to-GDP ratio. We still have to compare the financing burden with the economy and the revenue base that support it.

Refinancing puts another clock on the problem

The video also looks at the debt coming due and the cost of replacing it. Brian cites roughly $10.6 trillion of refinancing for FY2027. I treat that as the video’s maturity estimate rather than a separately verified total in this model. The model above does not use that figure. ^[2]

Rolling over a maturing security replaces a liability that already existed. A new deficit creates an additional net financing need. They both require investors to supply funding, but adding the full refinancing amount to the deficit would overstate how much new debt the government is taking on.

The interest effect depends on the rate of the debt being replaced and the rate of its replacement. A low-coupon note that matures into a higher-yield market raises the future bill. A bill that was already paying a high short-term rate might refinance more cheaply. We cannot apply the old average rate of the entire portfolio to every maturing security.

Those figures assume the full amount is repriced for a full year. Real maturities are spread across the calendar, so the first year’s effect is usually phased in. They also assume the quoted rate increase applies to the maturing obligations themselves. This is a sensitivity calculation, not a claim that all $10.6 trillion will reset upward by the same amount.

The long end belongs in this discussion. A lower overnight policy rate does not guarantee lower ten-year or thirty-year borrowing costs. Investors also price inflation, duration and the compensation they want for carrying the exposure. The refinancing bill follows the securities and rates the Treasury actually issues.

How much of revenue gets eaten depends on the path

Your concern about the interest bill consuming a growing percentage of the budget is the right next question. The percentage grows when the interest bill outpaces the revenue or spending measure we compare it with. That can continue for years, but it is a condition to test, rather than something guaranteed by the existence of interest alone.

In the upper panel, the same primary gap produces very different interest paths under 2.5%, 3.244% and 4.5% net-cost assumptions. By FY2046, annual interest is about $1.86 trillion, $2.71 trillion and $4.60 trillion, respectively. The assumptions apply to the model’s net-interest proxy; they are not forecasts of Treasury auction rates.

The lower panel keeps the central interest path unchanged and changes only revenue growth. With receipts increasing 3% a year, interest rises from 19.5% of revenue at the starting point to about 27.8% by FY2046. At 5% revenue growth, the ratio ends near 18.9%. The same rising dollar interest bill can therefore consume a larger or a smaller revenue share.

For this controlled comparison, primary spending moves with revenue so the nominal primary gap remains fixed. We are isolating the denominator effect, not building a complete economic forecast. In the real world, policy and demographics would alter both sides of that budget.

CBO projects the interest bill taking more room

We can compare the mechanism with an independently published baseline. CBO’s February 2026 outlook projected net interest increasing from $1.039 trillion in FY2026 to $2.144 trillion in FY2036. Over that interval, the projected deficit grew from $1.853 trillion to $3.115 trillion. These are February projections, not the final FY2026 accounts used earlier. ^[5]

Calculated from that baseline, interest rises from about 18.6% to 25.8% of receipts and accounts for about 68.8% of the FY2036 deficit. The same baseline puts publicly held debt at 120.2% of GDP in 2036. It builds in continued primary deficits as well as a growing interest burden.

The underlying fiscal-year series has jumps because the timing of certain payments shifts around weekends. The long-term increase in interest is still visible. It is useful to show the published series as it stands and identify the timing effect, instead of smoothing the chart into an invented official path.

My simple experiment and CBO’s baseline answer different questions. The experiment holds a few variables fixed to reveal the feedback. The baseline attempts to describe outcomes under a set of laws, economic assumptions and detailed program forecasts. Both show how financing costs can claim more of the future budget. Neither provides a fixed date when markets must refuse to finance the government.

Can growth, including AI, carry enough of the load?

The video examines the growth-first approach associated with Treasury Secretary Scott Bessent and the expectation that AI can lift productivity. The arithmetic behind the growth argument is understandable: a larger real economy can generate a larger tax base, making an existing debt balance easier to carry. ^[2]

The question is how much additional taxable income arrives, how quickly it arrives and how much additional spending or borrowing occurs along the way. AI investment is an expense today. Its productivity benefits can be real and still take time to spread beyond the companies building and using it.

For debt relative to GDP, a compact version of the relationship is b_t = [(1 + r) / (1 + g)]b_t−1 + p_t. Here b is debt relative to GDP, r is the effective nominal financing rate, g is nominal GDP growth, and p is the primary deficit relative to current GDP. The rate-growth comparison and the primary balance work together.

Faster nominal growth can help stabilize the debt ratio. A sufficiently large primary deficit can still overwhelm that help. A debt-to-GDP ratio crossing 100% does not create an automatic failure point, and falling below it does not by itself establish fiscal health. Funding costs, maturities, the revenue base and the direction of the ratio all deserve attention.

The IMF’s October outlook includes AI productivity gains among the upside possibilities while also identifying a reassessment of AI expectations and fiscal vulnerabilities as risks. That supports examining both outcomes. It does not establish that AI cannot help growth, or that growth alone will repair the budget. ^[7]

I have a hard time accepting future productivity as a complete answer to an interest bill being financed right now. The government needs the revenue gains to become large and durable enough to outrun the financing pressure, while also narrowing the primary gap. A technology forecast does not substitute for showing that path.

Inflation can change who carries the cost

Inflation can reduce the real value of an existing fixed nominal claim. If prices and nominal incomes rise while the amount owed stays fixed, the old obligation becomes smaller relative to that income. That is the appeal of the argument that debt can be inflated away.

Federal financing also contains bills that reprice quickly, notes and bonds that eventually mature, inflation-linked securities and spending commitments that respond to prices. Once investors expect persistent inflation, they may demand higher yields on new debt. Benefits, wages and government procurement can become more expensive as well. The budget does not freeze while the price level rises.

There is a distributional side that I think belongs in this discussion. Households whose wages lag essential costs can lose purchasing power. A fixed debt can become easier to carry for a borrower whose nominal income actually rises enough. Asset owners may receive some protection from repricing, but financing costs, valuations, taxes and the asset itself affect the result.

I am not treating every house, business, vehicle, stock or commodity as a guaranteed inflation winner. The link between consumer prices and asset prices runs through cash flow, financing and liquidity. That is the reason my earlier liquidity work follows those channels separately. ^[10]

The PN sequence: fiscal pressure into financing and liquidity

The sequence I am watching starts with a continuing primary deficit. That deficit adds debt. Interest adds another financing requirement. Refinancing changes the cost of carrying the balance. At some point, policymakers face growing pressure to improve the primary budget, obtain better real growth, reduce financing costs or accept a different inflation outcome.

That is where the fiscal discussion meets the monetary system. Raising rates to restrain inflation can increase the government’s financing burden as debt reprices. Cutting rates may ease some of that burden, but the long end can remain expensive if inflation expectations or term premiums rise. The policy trade-off becomes harder when the fiscal side does not adjust. Federal Reserve research has examined this coordination problem for decades. ^[12]

I expect this pressure to keep pushing authorities toward measures that support financing and market liquidity when stress becomes difficult to absorb. That is my assessment of the incentives. It does not mean every deficit triggers a Federal Reserve purchase, or that a particular intervention is already scheduled.

Treasury financing and Federal Reserve asset purchases also operate differently. Treasury issues securities to investors. The Fed describes its securities purchases as independent monetary-policy decisions and buys outstanding Treasuries from the public. Bond issuance by itself is not a Federal Reserve money-creation operation. ^[8]

Buying a Treasury security does not make that government obligation disappear. Fed purchases exchange securities held outside the Fed for reserve balances, and the Fed can pay interest on those reserves. The consolidated financing cost can therefore change form rather than vanish. This is one reason “just have the Fed buy it” is an incomplete answer to the interest problem. ^[9]

In the LCI framework, I track the interaction of the Fed’s balance sheet, Treasury cash, reverse repo balances and broad money. They help explain how a policy response reaches funding conditions and collateral. The forward cycle and economic reset study asks how financing and economic stress can feed into the next stage of the cycle. This budget work gives that sequence a fiscal starting point. ^{[10] [11]}

What would change this path?

The loop can be weakened. The way to judge progress is to watch the variables that actually change its arithmetic. Announced savings and revenue plans need to show up in the primary balance, while financing conditions need to show up in the effective cost of the debt.

A recession belongs in that monitoring. Falling taxable income and profits can reduce collections, while stabilizers and policy responses can increase spending. That would put pressure on the primary balance at the same time the existing debt still needs service. It is a risk channel, not a claim that the next downturn must reproduce the pandemic budget.

The evidence that would weaken my concern is sustained improvement across the primary balance, funding costs and the revenue share consumed by interest. A lower headline yield for a few days or one strong month of receipts would not establish that change on its own.

Where I stand after the FY2026 accounts

The national debt is still growing because the government spends more than it collects. Inside that shortfall is a financing bill that is itself large enough to create another layer of borrowing. That layer then helps generate a future financing bill.

In FY2026, we borrowed almost $2 trillion on the budget measure. About $1.052 trillion of the gap was net interest. Remove that expense in a thought experiment and the annual shortfall falls to about $941 billion. Keep financing the expense at positive rates and the debt acquires a compounding component.

That is the point I want to make. We should be asking how much of the next deficit comes from carrying the last one, how quickly that cost is growing relative to revenue, and what would have to change to stop it taking more room. The completed budget already gives us enough information to ask those questions with numbers.

Questions about the interest loop

Was half of all federal spending interest?

No. Net interest was about 14.2% of total spending and 52.8% of the FY2026 deficit. It also equaled 19.5% of receipts.

Would removing interest balance the budget?

Holding everything else fixed, it would reduce the annual deficit from $1.992 trillion to $0.941 trillion. It would avoid additional borrowing; it would not retire existing principal.

Why use net interest rather than the $1.360 trillion gross figure?

Net interest belongs in the unified-budget deficit calculation. Gross Treasury interest includes intragovernmental amounts and is offset elsewhere in the accounts.

Does a primary balance stop debt growth?

If interest is still financed, nominal debt can continue growing even with no primary deficit. A large enough primary surplus can offset interest. The debt-to-GDP ratio also depends on growth.

Are the PN charts a forecast of $87 trillion of national debt?

No. That endpoint is a controlled publicly held debt illustration under a fixed nominal primary gap and calibrated interest ratio. It is not a forecast of gross national debt.

Does every maturing Treasury refinance at a higher rate?

No. The relevant comparison is the old financing cost of the specific maturing debt versus its replacement cost. Some obligations can refinance more cheaply.

Does crossing 100% debt-to-GDP mean a country immediately fails?

No. There is no universal automatic failure point at that ratio. The debt measure, rates, maturity structure, growth, primary balance and investor demand all matter.

Does Federal Reserve buying cancel the government’s debt?

No. The Treasury obligation remains outstanding, and reserve liabilities created in the purchase can themselves carry interest. The financing structure changes.

The records behind this analysis

1. U.S. Treasury — Final Monthly Treasury Statement, September 2026
   Tables 1–4: budget totals, revenue and gross interest; Table 6: financing and securities balances; Table 9: budget functions and net interest. FY2025 comparisons use this statement’s prior-year column. Amounts are published in millions and independently rounded.

2. ClearValue Tax — source video, October 22, 2026
   Brian’s fiscal-year overview prompted this article. The full auto-generated transcript was reviewed. Budget results were checked against Treasury. The approximately $10.6T maturity estimate is attributed to the video and is not an input to the PN debt model.

3. Congressional Budget Office — Federal Net Interest Costs: A Primer
   Used for the accounting definition, intragovernmental offsets, rate transmission and maturity structure. Its old numerical forecasts are not used as current projections.

4. Federal Reserve Bank of St. Louis, FRED — OMB annual fiscal series
   FYOINT: net interest; FYFR: receipts; FYFSD: surplus/deficit. Downloaded October 22, 2026. FY2000–2024 is retained in the history chart. FY2025–2026 is taken from the current Treasury statement to keep the year-over-year comparison consistent.

5. Congressional Budget Office — The Budget and Economic Outlook: 2026 to 2036
   February 11, 2026 baseline; Table 1-1, printed pages 12–13. Figure 8 reproduces the dated projected series. Laws and economic assumptions predate the final FY2026 accounts.

6. Social Security and Medicare Boards of Trustees — 2026 report summary
   Trust-fund reserve depletion projections, continuing benefit financing, and the distinct funding arrangements for Medicare’s supplementary medical insurance.

7. International Monetary Fund — October 2026 World Economic Outlook
   Used for the balance of AI productivity opportunities, AI expectation risks and fiscal vulnerabilities; it is not cited as proof that AI cannot improve growth.

8. Federal Reserve Board — Securities purchases and government borrowing
   Explains the separation of Treasury issuance and Federal Reserve monetary-policy purchases of outstanding securities.

9. Federal Reserve Board — Interest on Reserve Balances FAQ
   Explains reserves, the asset exchange in Fed securities purchases, and interest on reserve balances.

10. Pattern Nexus — Hard Assets Follow Liquidity: The LCI Framework White Paper
    Prior PN framework linking the Fed balance sheet, Treasury cash, reverse repo balances, broad money and collateral conditions.

11. Pattern Nexus — Recession Risk and the Economic Reset
    Related PN work on the forward cycle, financing stress and a technology-driven economic reset.

12. Federal Reserve Bank of St. Louis — Is It Time for Some Unpleasant Monetarist Arithmetic?
    David Andolfatto, 2021. Research context for the interaction of monetary tightening, fiscal adjustment and debt-service costs; it is not an official prediction of current policy.

Back to top