The Debt MachineCornered governments must inflate and grow at the same time — and the collision quietly moves wealth from wages and savings to real assets and borrowers. Who wins, who loses, and why — in the US and Canada.

A government in this position has three obvious escape routes, and modern politics has quietly closed all three.

It can't easily cut spending. Most of the budget is promises already made — pensions, health care, and interest on past debt. In the US, those plus defence leave only a sliver that is even legally discretionary. Cutting enough to close the gap would mean breaking promises to voters, which no government survives.

It won't raise taxes. This one is now written into law. In July 2025 the US made the 2017 tax cuts permanent (the "One Big Beautiful Bill Act"), at a cost the official scorers put near $3.4 trillion of extra deficits over ten years. So "taxes don't rise to close the gap" isn't a hypothesis — it is the current plan, baked into the official forecast.

So it borrows — but borrowing has a ceiling. A government can borrow for a very long time, but not forever, and two things pull the ceiling closer. The first is trust: lenders keep buying the bonds only while they believe they'll be repaid in money that still has value — lose that and they demand much higher interest, or stop buying. The second is the interest bill itself, which we'll see (§7) has grown big enough to start eating the budget from the inside.

This is the part to slow down for, because everything else hangs off it. Call it the squeeze: a government pressed from two sides at once, with the saver caught in the middle. A cornered government that won't default outright is left with exactly two ways to deal with a debt it can't repay — and, crucially, it does not pick one. It runs both at once.

Now watch what happens when you pull both levers at once. Lever 2 wants interest rates low. Lever 1 makes inflation high. Put them together and you get money that earns you less than prices are rising — a negative real return on cash and bonds. That isn't an accident or a policy failure. It is the whole point, and it has a name:

The honest caveat — the presses aren't running at full tilt yet. Two things say the squeeze hasn't fully closed: the real return on a 10-year bond is still positive (~+2.3%), so savers aren't yet being openly repressed, and the aggressive, deficit-funding kind of printing isn't running. But the direction has already turned. The Fed ended QT — its bond-shrinking programme — in December 2025, and has been quietly expanding its balance sheet again since, up roughly $90 billion, buying Treasuries to keep bank reserves from running short. It calls this "reserve management," not stimulus; sceptics call it stealth QE. Either way, the printer has gone from off to a low setting. And live money-market stress — strained overnight funding, plus the unwinding of the yen "carry trade" (cheap yen borrowed to buy other assets) draining global liquidity — is exactly the pressure that can turn a technical trickle into a flood. So the squeeze is the direction of travel, and it has begun to close; how firmly is the forecast, and it comes in three shapes:

Odds are an informed judgement to force honesty about uncertainty — not a measurement. The value is in the shape: most likely a slow grind, a real chance growth bails it out, a smaller chance of a scare.

There are two ways to measure the size of an economy, and the gap between them is the whole game. Nominal GDP counts output in today's dollars — it rises when we make more stuff and when the same stuff just costs more. Real GDP strips the price rises out. Roughly: nominal growth ≈ real growth + inflation.

Put the latest US year side by side and the illusion is plain.

Of the 6.1% the US economy "grew," roughly 2.7 points were real and 3.3 points were price. A little over half of the headline was inflation, not output. That is the difference that matters for the squeeze: inflated GDP shrinks the debt ratio only by taxing everyone through higher prices (Lever 1 in disguise); real GDP shrinks it by genuinely producing more (Lever 2, the clean exit). When a government boasts of "growth," the first question is always: real, or just the yardstick shrinking?

It helps to remember what "the economy" is even made of. GDP is consumption + business investment + government spending + net exports. Government purchases are only about 17% of it (and most of that is state & local — schools, roads, police — not federal). Which is why a government can't simply spend its way to real growth: its direct slice is small, and over the past year US government purchases added almost nothing to real growth. The private sector does the real growing — which is exactly why the "grow" bets in §6 are aimed at private investment, not government payrolls.

First, untangle "printing," because the word hides three different things with very different consequences.

Here is the fact that makes Lever 1 real regardless of what the Fed does next: the money created in the pandemic never left. US M2 — a standard tally of money in the system — has grown from about $15.5 trillion at the start of 2020 to about $23 trillion now: roughly 50% larger in six years, most of it created in the 2020–21 burst. And the taps are not fully off: after shrinking its balance sheet through 2024–25, the Fed ended that programme in December 2025 and has been slowly expanding again (to keep bank reserves ample). So the stock of money is permanently far higher, and the flow has quietly turned positive again — modestly for now. That stock is still working its way into prices.

And money reaches prices slowly. This is the crucial, under-appreciated part. When the money supply surges, inflation follows with a lag of roughly a year to two. You can see it in the last cycle: US money-supply growth peaked at an extraordinary ~27% in early 2021, and consumer-price inflation didn't peak until 9.1% in mid-2022 — about 16 months later. So a burst of printing is not "over" when the printing stops; its price effects keep arriving for a year or two afterwards. That lag is exactly what lets a government run Lever 1 while official inflation still looks like it's cooling — the subject of §5.

Recall the bind. Lever 2 (grow) wants rates low; the debt's interest bill (§7) also screams for rates low. But a central bank can only justify low rates if inflation looks contained. So there is enormous pressure on that one number. Three honest reasons it can read cooler than the inflation people actually feel:

1 · The lag (from §4). Because money hits prices a year or two late, official inflation can be falling today even as the full effect of money already created is still arriving. "Inflation is coming down" and "there's more inflation in the pipeline" can both be true at once.

2 · The measurement itself is contested. Official inflation (the CPI) is built on genuine but debatable choices — how housing costs are estimated (via "owners' equivalent rent" rather than actual house prices), how quality improvements are subtracted out ("hedonics"), how the basket shifts when people trade down to cheaper goods ("substitution"). Each is defensible; each also tends to lower the measured number relative to the rising cost of an unchanged life. This is a long-running, legitimate debate among economists — not a fringe theory — about whether official CPI understates true cost-of-living.

3 · The incentive all points one way. This is the part worth naming plainly. A low inflation print lets the central bank keep rates low, which keeps the government's debt serviceable and feeds the growth bet. A government drowning in debt has every incentive to tolerate inflation and to lean toward the lower reading — the fiscal dominance pressure from §7. So when the official number and your grocery bill disagree, the institutional thumb is on the scale in one direction.

4 · And the referee is being reviewed. This isn't only theoretical. Since taking over the Fed in 2026, Chair Kevin Warsh has launched formal task forces to re-examine how the Fed reads inflation and the data it relies on. That can be read two ways: a genuine attempt to measure better, or the groundwork for a softer official reading — and the lower rates a softer reading would justify. Donatien's Stagflation Reaction Function report read it as the latter — Warsh "building the data machinery to justify easing on a softer inflation read." Whichever it is, the direction of the effort points the same way as the incentive above: toward a number that lets rates fall.

If a government is going to try to out-grow its debt, it has to bet on something that lifts real output. Two bets are live right now, one on each side of the border.

So where does the money flow? Toward the growth being bet on, and toward the real things that growth needs. Descriptively — not as any kind of tip — that means capital gravitating to the technologies governments are backing for real growth (AI and its adjacencies — data centres, power, and the frontier areas investors group with them), to the real-asset and commodity industries that a build-out consumes (energy, metals, materials), and to companies with genuine pricing power that can pass inflation on rather than absorb it. This is the "winners" column from the top of the report, seen from the capital-flow side.

Two facts make the corner concrete. The first is the interest bill. In the last fiscal year, US net interest on the debt crossed $1 trillion for the first time — now the government's second-largest expense after Social Security, bigger than the entire military. With a deficit of ~$1.8 trillion and interest of ~$1.0 trillion, more than half of this year's borrowing is just interest on past borrowing. That is the loop that feeds itself: debt → interest → bigger deficit → more debt.

The second is the trajectory. On the CBO's own current-law baseline — which already assumes taxes stay where the 2025 law put them — US federal debt held by the public climbs from about 100% of GDP now to ~120% within a decade and ~175% by mid-century, higher than at any point in American history, including just after World War II.

The importance of the interest bill isn't just its size — it's what it does to the choices. Every extra point of interest rate now costs the government enormously, which is the mechanical reason the whole system leans toward keeping rates low (Lever 2, and §5's incentive). Past a point, a central bank can't raise rates to fight inflation without bankrupting the government — the doorway from "borrowing" to "printing," and the reason the tail scenario in §2 exists at all.

The debt is, like-for-like, close to America's

Canada's federal debt looks small — about 41% of GDP — and its deficit is smaller than America's too: the federal shortfall was ~C$36 billion (~1% of GDP) in 2024-25, and even on the broader all-government basis Canada runs about 2% of GDP against the US's ~6%. But in Canada the provinces do much of the spending, and on the like-for-like measure — general-government gross debt, all levels combined — Canada sits around 113% of GDP, versus the US at ~123%. Broadly similar. That combination — a much smaller annual deficit but a similar accumulated debt — is the tell: Canada borrowed heavily before and is now consolidating faster (the IMF sees its deficit drifting toward ~1% by 2030), while the US keeps borrowing ~5.5%+ a year. Similar debt today; steadily diverging from here. (Ignore the ~10% "net debt" figure sometimes quoted; it subtracts pension-plan assets that can't be used to pay down bonds.) And Canada printed too: the Bank of Canada's balance sheet ballooned from ~C$120 billion to ~C$575 billion in 2020–21, inflation peaked at 8.1%, and it has since wound that emergency programme back down. The mechanism chapters apply directly.

Carney is betting on Lever 2, out loud

Where the US leans on both levers, Carney is trying to make the grow leg do the heavy lifting. His first budget (Budget 2025, November 2025) balances day-to-day spending by 2028–29 but keeps borrowing — deliberately — to invest, via a capital budget and a Major Projects Office fast-tracking ~C$116 billion of building. It is the report's optimistic ~25% scenario (§2) pursued as national policy: grow the real economy faster than the debt, rather than inflate the debt away.

For Canadian households, the whole machine runs through the house

The report's winners-and-losers split has an unusually literal Canadian form. Canadians hold roughly 55% of their wealth in real estate, carry the highest household debt in the G7 (about C$1.80 of debt per C$1 of income, roughly double the US), and ~75% of that debt is mortgages. So "real assets held against cheap debt" — the report's biggest winner — isn't a metaphor here; it is the family home with a mortgage on it. Those who bought and locked low rates in 2020–21 rode exactly the transfer this report describes.

And Canada can't escape the American machine

About three-quarters of Canada's exports go to the United States (~76% in 2024; even after 2025's tariffs cut it to ~72%, still the overwhelming majority). The loonie is weak — around C$1.42 per US dollar — and a weak currency imports inflation, because so much of what Canadians buy is priced in US dollars. So America's deficits, inflation and rates set much of Canada's weather regardless of what Ottawa does. One quiet footnote: as the world's central banks bought gold at a record pace after 2022, Canada holds none — it sold the last of its official reserves by 2016, the only G7 country at essentially zero.

Everything so far points one way. Under financial repression — cheap money, hotter inflation, negative real returns on the safe stuff — capital doesn't sit still and accept the erosion. It drains from what loses to inflation and pools in what resists it:

The distributional punchline. The assets that resist debasement are disproportionately owned by people who are already wealthy — in the US the top 10% of households own about 93% of all stocks, and the top 1% hold roughly as much wealth as the entire bottom 90% combined. So the very process that quietly shrinks the government's debt tends to widen the gap between those who own assets and those who live on wages and savings. Not through anyone's villainy — through the plumbing.

The same logic sorts economies

Scale the machine up to whole countries and it separates the likely winners from the likely losers on three tests:

Globally, the same slow tide: central banks buying gold at a record pace, the US dollar's share of reserves drifting from ~71% (2000) to ~57% — though the IMF notes most of that recent slippage was exchange-rate maths, not a stampede for the exits. The likelier picture over 5–10 years isn't a currency collapse; it is a slow, shared erosion of paper money against real things, with the dollar keeping its crown while the crown loses some shine.

1 · A government that borrows in its own currency can't be forced to default. The US and Canada issue debt in a currency they alone create, so they can always meet an obligation in their own money. The risk isn't non-payment; it is payment in devalued money — a slow cost, not a cliff.

2 · "r versus g" can quietly do the work. Debt as a share of GDP falls whenever growth (g) outruns the interest rate on the debt (r), even while running a deficit. For much of the last 20 years r sat below g and the debt was, in that sense, self-managing. If growth holds up or rates come down, the arithmetic works in the government's favour again — which is precisely the bet in §6.

3 · The dollar's "exorbitant privilege." The world needs US dollars for trade, reserves and safety, so demand for US debt is enormous and durable — letting the US borrow more, more cheaply, for longer than the raw numbers suggest. (Canada has a milder version: a stable, resource-backed economy investors are willing to fund.)

4 · It has been done before, without catastrophe. This is the strongest card. After World War II US debt peaked near 106% of GDP — about today's level — and fell to ~23% by the mid-1970s without any default. But be precise about how: research finds growth alone did only about a quarter of the work. The majority came from surprise inflation plus financial repression — years of rates held below inflation. In other words, the happy precedent is the base case of §2. It worked. It also quietly taxed a generation of savers to do it.

The machine, in one breath. Can't cut, won't tax, can't borrow forever → so the government pulls both levers at once: let inflation shrink the debt, and keep money cheap to chase real growth. Cheap money plus hot inflation is financial repression — negative real returns on the safe stuff — and it moves wealth, quietly and without a vote, in one consistent direction.

Who wins: owners of real assets and real estate; companies with pricing power; borrowers on cheap fixed-rate debt against real assets; the industries the growth bet funds (AI, energy, commodities); and resource-rich, real-growth-investing economies — Canada's resources among them.

Who loses: cash savers; wage-earners without bargaining power; bond-heavy pensions and fixed incomes; and economies that inflate without investing, or must import the very real things that are rising.

The one clean exit. There is a way out that isn't a hidden tax on savers: real growth, fast enough to outrun the debt. It is the hardest path and the rarest — and it is the one Canada is now openly attempting, and the one Washington is betting AI will deliver. Whether those bets pay off is the single most important open question in this entire report. If they do, most of this becomes a footnote. If they don't, the quiet tax does the work instead.

A closing note true to what this project is: this is written by an AI-assisted analysis project, not a professional economist, and it is analysis for understanding, not a recommendation to do anything. If it helped the machine click into place, it did its job.