The US Now Pays More in Interest Than It Spends on Its Military

Last month, the United States crossed a line.

Annual interest payments on the national debt passed one trillion dollars for the first time in history. The figure appeared in the Treasury's monthly statement for August 2026, the eleventh month of the fiscal year.

In the same year, the federal government will spend about $885 billion on national defense.

The United States now spends more servicing its past borrowing than it spends on the most powerful military ever assembled. More than it spends on Medicaid, which will cost about $708 billion. Among all federal expenditures, only Social Security is larger.

Here is how fast it happened. Interest cost $352 billion in 2021. It reached $476 billion in 2022. By 2024 it was $881 billion. In 2025, $970 billion. This year, one trillion.

Roughly tripled in five years.

The Congressional Budget Office projects it will double again, to $2.1 trillion a year, by 2036. Over the next decade, interest payments alone will total $16.2 trillion.

No legislation created this. No president campaigned on it. Nobody in Congress voted for a trillion-dollar interest bill.

It is the automatic consequence of two numbers multiplied together, and it is now the fastest-growing item in the budget of the United States.

Here is the full file.

Interest expense is the simplest calculation in government finance. It is the size of the debt multiplied by the rate paid on it.

Both numbers moved at once, and that is the entire story.

Federal debt passed forty trillion dollars in August. Debt held by the public now stands at about 101 percent of GDP, and the CBO projects it reaching 120 percent by 2036 and 175 percent within two decades. That is higher than at any point in American history, including the peak of the Second World War.

Meanwhile, the rate the government pays has risen sharply. The 30-year Treasury yield recently touched 5.34 percent, its highest since 2007.

Here is the mechanism most people miss. The government does not borrow once. It borrows continuously, and it refinances constantly. Treasury bills issued at half a percent during the pandemic mature and must be replaced with new paper issued at today's rates.

The trillions borrowed cheaply between 2009 and 2021 are being rolled over into a market that now demands four and five percent. The old debt does not stay cheap. It becomes expensive on a schedule, whether anyone likes it or not.

That repricing is happening right now, mechanically, on a portfolio of forty trillion dollars.

There is a specific feature of this arithmetic that separates it from an ordinary budget problem.

Interest is not discretionary spending. It cannot be cut in a negotiation. There is no committee that can vote to spend less on it. The payment is contractual, and the alternative to making it is default.

So interest is paid first, and everything else is funded from what remains.

Now follow the loop. The government runs a deficit. The deficit must be financed by issuing new debt. The new debt increases the total on which interest accrues. The higher interest bill increases next year's deficit. Which requires more borrowing.

CBO projects deficits averaging 6.1 percent of GDP over the next decade, reaching 6.7 percent by 2036. And here is the detail that tells you what is actually driving it: the primary deficit, meaning the deficit excluding interest, is projected to shrink slightly over that period, from 2.6 percent of GDP to 2.1 percent.

Read that carefully. The operational budget is projected to improve modestly. The total deficit is projected to worsen substantially.

The entire deterioration comes from interest.

The government could freeze every program tomorrow, and the debt would still grow, because it is now growing from its own servicing cost. This is the point at which a debt stops being a policy choice and becomes a mathematical process.

Interest costs will exceed Medicare spending by 2028. They will exceed all discretionary spending, defense and non-defense combined, by 2038. By 2048, interest will be the single largest expenditure of the United States government.

This is the part that is almost never discussed, and it is the part that matters for a newsletter about where money goes.

A trillion dollars a year is not disappearing. It is being paid to someone. Every dollar of interest is a transfer from the taxpayer to the holder of the bond.

So who holds the bonds?

Foreign investors held $9.299 trillion of US Treasuries as of June. Japan holds about $1.117 trillion. The United Kingdom $940 billion. China $633 billion. Canada $460 billion. Those four alone account for more than $3.1 trillion.

The rest is held by American banks, insurance companies, pension funds, mutual funds, the Federal Reserve, and individuals. Ownership of financial assets in the United States is heavily concentrated at the top of the wealth distribution, which means the domestic portion of that interest flows predominantly to households that already own capital.

Interest payments are financed by general taxation, which is broad. They are received by bondholders, which is narrow.

There is no conspiracy in this. It is the ordinary operation of sovereign debt, and it works this way in every country. But it is worth stating plainly, because it almost never is: the fastest-growing expenditure in the federal budget is a transfer from people who pay taxes to people and governments who own financial assets, it requires no vote, and it grows automatically.

That is now the single largest wealth transfer mechanism operating in the American economy, and it is scheduled to double.

If you want to know what large institutions think of the arithmetic above, do not read their statements. Watch what they buy.

Central banks purchased 289 tonnes of gold in the second quarter of 2026 alone. This is not a new trend, it is an accelerating one, and it has now run for several consecutive years at record levels.

Gold trades around $4,350 an ounce. It rallied roughly ten percent in August, its best month since January, in the same weeks that federal debt crossed forty trillion dollars.

A central bank that buys gold is doing something specific. It is converting a claim that pays interest into an asset that pays nothing.

That trade only makes sense if you are more worried about the value and the safety of the claim than you are attracted by the yield. Gold has no counterparty. It cannot be defaulted on, frozen, or printed. That is its entire function in a reserve portfolio, and it is the reason institutions buy it when confidence in sovereign paper weakens.

Meanwhile the long end of nearly every developed sovereign curve has sold off to levels not seen in a generation. Japanese 10-year yields crossed 3 percent for the first time since 1996. German bunds reached their highest since 2011. The 30-year UK gilt approached 6 percent.

When every major government's long-term borrowing cost rises in the same month, the explanation is not one country's policy error. It is a global repricing of what it costs a government to borrow for thirty years in a world with more debt, more inflation risk, and fewer buyers who will accept any yield.

The bond market has started pricing developed sovereigns on creditworthiness rather than assuming it.

Three conclusions.

Every large debt is resolved in one of three ways, and only one of them is announced. A government with debt it cannot service can default, which is rare and catastrophic for a reserve issuer. It can grow its way out, which requires growth rates well above what any major developed economy is currently producing. Or it can inflate, meaning it repays in currency that buys less than the currency it borrowed. The third option is the only one that requires no vote, no announcement, and no admission. It also transfers wealth silently from holders of cash and long-dated bonds to holders of debt and real assets. When you assess the next twenty years, the useful question is not whether a crisis arrives. It is which of those three mechanisms is being used, and on whom.

Interest rates are no longer just a policy lever, they are a fiscal constraint. For forty years, central banks set rates primarily according to inflation and employment. When interest is 3.3 percent of GDP and rising, the calculus changes, because every rate decision is now simultaneously a decision about the government's own budget. Higher rates fight inflation and worsen the fiscal position at the same time. That tension is structural, it will not resolve, and it is the single most important thing to understand about monetary policy for the rest of this decade. Expect central banks to be slower to raise and faster to cut than their mandates alone would suggest, because the arithmetic gives them no other option.

Position for the mechanism, not the prediction. Nobody knows the timing, and anyone who tells you they do is selling something. But the direction is not really in question, because it is arithmetic rather than opinion. In an environment where sovereign debt compounds and the resolution is most likely gradual currency debasement, the assets that have historically held real value are the ones that cannot be printed: productive businesses with pricing power, real assets, and hard commodities. This is exactly what central banks are doing with 289 tonnes a quarter, in public, in their own published reserve statements. They are not predicting a crisis. They are reducing their exposure to a claim. You can read their filings and draw your own conclusions.

$1,000,000,000,000. Annual interest on the US national debt, crossed for the first time last month.

$885,000,000,000. What the United States will spend on its military this year.

The larger of those two numbers grows automatically, requires no legislation, cannot be cut in a negotiation, and is projected to double within ten years.

It is the fastest growing thing in the American budget, and it buys nothing.

That is not a conspiracy theory. That is the Treasury statement.

The Dark Money Letter is published every Wednesday. → thedarkmoneyletter.com

Sources