The Bond Market Is Warning of a Major Collapse

There is a recurring moment before financial collapses when the public becomes least capable of recognizing danger because the visible surface of the system still appears triumphant. Markets rise, headlines celebrate record highs, and investors convince themselves that soaring asset prices are proof the underlying economy remains healthy. The final stage of speculative cycles rarely feels like panic. More often, it feels like vindication. Skeptics are mocked, risk is dismissed as pessimism, and the crowd mistakes momentum for stability precisely as the foundations underneath it begin to weaken.

That is the atmosphere surrounding the American economy now.

The stock market continues levitating on the back of a small cluster of massively overvalued technology companies whose extraordinary valuations increasingly conceal the deterioration spreading beneath the surface of the broader economy. Donald Trump, whose understanding of economic health rarely extends beyond the level of a stock ticker, points to all-time highs as proof of national prosperity restored. Yet the deeper signals inside the financial system are telling a radically different story, one centered not on speculative enthusiasm but on the rising cost of money itself.

In the third week of May, the yield on the 30-year US Treasury climbed to 5.19 percent, reaching levels not seen in nearly 19 years. That number matters far more than another record close in the S&P 500. Long-duration Treasury yields underpin the pricing of mortgages, corporate borrowing, government debt, commercial real estate, and future equity valuations throughout the global financial system. When those yields rise sharply while investors continue celebrating speculative rallies, a dangerous divergence begins to emerge between the optimism of equities and the arithmetic of the real economy.

History is filled with moments when markets reached euphoric highs shortly before reality reasserted itself with extraordinary force. The crowd sees all-time highs and interprets them as evidence the system is strong. Bond markets, far more concerned with debt, deficits, refinancing risk, and long-term credibility, often reach a very different conclusion first.

What’s really happening, in simple terms

A bond is a loan. When the US government spends more than it collects in taxes, which is nearly always, it borrows the difference by selling bonds, IOUs that pay a fixed amount of interest each year and return the principal after a set term. The longest and most watched is the 30-year Treasury.

The yield is the return you earn for holding it, and the trap that confuses nearly everyone is that the price and the yield move in opposite directions, like a seesaw. The interest a bond pays is fixed forever. A bond paying 5 dollars a year that originally cost 100 dollars yields 5 percent. If demand is strong and buyers bid its price up to 125 dollars, that same 5 dollars is now a smaller return on a larger price, and the yield falls to 4 percent. If demand collapses and the price drops to 80 dollars, the same 5 dollars becomes a larger return on a smaller price, and the yield jumps to 6.25 percent.

So the rule, the key to the entire story, is this: when a bond’s price falls, its yield rises. Which means a spiking yield, the kind that makes a reassuring-sounding headline, is the opposite of reassuring. It is the sound of investors backing away and demanding to be paid more to stay. A 30-year yield at a nineteen-year high is not a statistic about interest rates. It is a measurement of refusal, a record of how little the world will now pay to hold the debt of the United States. The most trusted borrower on the planet is being repriced as a worse credit, not by an enemy, but by its own creditors, who have studied the trajectory of American debt and inflation and concluded, one by one and then all at once, that the old price was no longer safe. That repricing is the warning beneath every other warning in this piece, and it is the one almost no one is willing to follow to its conclusion.

The Fed lever is failing in the Trump economy

To understand why the quiet professionals are frightened, begin with the single fact that orders all the others. For more than forty years, across the last seven times the Federal Reserve began cutting rates, long-term Treasury yields obediently fell in the months that followed. Every cycle. Without a single exception. It was as close to a law as finance permits.

This time the law broke. The Fed cut 175 basis points between late 2024 and late 2025, and the 30-year yield rose by nearly a full percentage point while it did. The Fed pushed the gas pedal that had worked for two generations, and the car did not move. Bank of America’s strategists marked the breach of the 5 percent level as the threshold past which the market, and no longer the central bank, sets the price of long money. The implication is graver than the phrasing suggests. The institution everyone counts on to be the adult in the room tried to pull yields down and discovered, in public, that its authority over the long end of the curve was gone.

The historical company is not reassuring. Bank of America has traced this exact configuration, long-term yields ripping higher in open defiance of monetary policy, to the doorstep of the three greatest equity collapses of the modern age: Japan in 1989, before the Nikkei fell into a grave it has still not fully climbed out of; the United States in 1999, at the precise summit before the dot-com world came down; China in 2007, weeks before its market broke apart. In each of those endings, mercy took one form. A single great economy fractured while the others held, so the fleeing capital had somewhere solid to run. That mercy is precisely what is missing now. The long end is breaking in Tokyo and London and Frankfurt and Washington in the same season. The bunker that capital has always retreated to when the equity world burns has, for the first time in living memory, hung out a sign that reads no vacancy.

How a bond yield actually breaks a stock market: the four links

The phrase “rising yields put pressure on stocks” is repeated so often it has gone inert. It describes a real and forceful chain of causation, and that chain has four distinct links, each operating independently, all of them now operating at once.

The first link is the discount rate. The value of a stock is, in the end, the present value of all the cash a company will generate in the years ahead, and converting those future dollars into today’s dollars requires discounting them by a rate anchored to the risk-free Treasury yield. When the 30-year sits at 2 percent, distant future earnings are discounted gently, and a company that will not turn serious profit for a decade can still justify a towering price. When the risk-free rate moves to 5 percent, the discounting turns severe, and the punishment falls hardest on exactly the companies whose value lives furthest in the future, which is to say the high-growth technology names that dominate this market. A rising long yield does not nick these valuations evenly. It guillotines the most expensive ones first.

The second link is the competition for capital. A 30-year Treasury yielding a guaranteed 5 percent, backed by the taxing power of the United States, is not a passive benchmark. It is a live, competing product. Every pension fund, insurer, and endowment on earth must now weigh the certainty of 5 percent against the strained promise of stocks priced, by the most credible long-run measures, to return somewhere between zero and slightly negative over the coming decade. Capital is not sentimental. As it migrates toward the guaranteed return, it withdraws the marginal buying that a record-high market requires merely to stay aloft, and a market that stops receiving new money does not plateau gently. It rolls over.

The third link is the corporate cost of capital, and it is the true detonator, because it runs not on sentiment but on contractual deadlines that cannot be wished away. We will return to it in full, because it deserves its own section.

The fourth link is the one that turns a decline into a cascade: concentration. Because seven companies now constitute roughly 35 percent of the entire S&P 500, the index is no longer a broad democracy of 500 firms. When the giants fall, the index funds that hold the savings of tens of millions of ordinary people are forced to sell in proportion, mechanically, regardless of how the other 493 companies are doing, and that forced selling drives prices down further, which forces more selling still.

Hold these four links in mind. Every scenario that follows is simply a different way of pulling on them.

The fire no one can put out

The force driving the yields higher is an inflation the world had been assured was dead and buried. As recently as January it had cooled to 2.4 percent, close enough to the Federal Reserve’s target that the comforting story of a soft landing could still be told around the table. By April it had risen to 3.8 percent, the highest in nearly three years, and the readings beneath the surface were worse than the headline. Producer prices, the costs businesses swallow before passing them to the public and therefore the truest leading indicator there is, reached 6 percent, the fastest in over three years, meaning the inflation already loaded into the pipeline has not yet finished arriving. The Philadelphia Fed’s panel of professional economists, which had projected 2.7 percent for the second quarter just three months earlier, now expects 6 percent. No one more than doubles a forecast in a single quarter unless the earth has moved beneath them.

It has already reached the place where ordinary people live. For the first time in three years, the average American worker is moving backward in real terms, take-home pay shrinking against prices, down 0.3 percent across the year and falling 0.5 percent in a single month. The engine of it is energy, and the energy shock was not an accident of weather or fate. It was chosen. Prices are up 17.9 percent over the year and gasoline up 28.4 percent, the direct consequence of the war the United States and Israel opened against Iran at the end of February, which has shut the Strait of Hormuz, the narrow water through which roughly a fifth of the world’s seaborne oil must pass. Brent crude has surged more than 45 percent since the first strikes, and now Tehran lets it be known that any settlement may carry a permanent toll on every tanker that runs the strait. This is a supply shock, the one kind of inflation that interest rates cannot reach, the kind that has historically forced central banks to choose between two catastrophes.

That is the room the Federal Reserve is now locked inside, and the walls were built at home. It cannot raise rates hard enough to break the oil inflation, because the interest owed on more than 34 trillion dollars of debt would become a number the Treasury cannot pay. It cannot cut rates to rescue the bond market or flatter the stock market, because to ease into a rising inflation would collapse the dollar and drive food and fuel higher still. Both doors are welded shut, and the same hands that demanded the easy money helped weld them. For a year the administration pressed for lower rates, savaged the former chairman Jerome Powell by name as suffering from “Trump derangement syndrome,” and installed Kevin Warsh, sworn in only last week, for the express purpose of delivering cuts, while the same administration’s war lit the oil market that has now made those cuts impossible to justify. The hand reaching for cheap money and the hand that struck the match belong to the same body, and the body is beginning, already, to look around for someone else to blame.

The market has read the trap faster than the men who set it. Futures now lean toward the Fed’s next move being a hike rather than a cut. The minutes of its last meeting record officials openly discussing the need to raise rates if the war keeps inflation high. And in the most telling moment of all, Christopher Waller, a man this president put on the Board, admitted in the week of Warsh’s swearing-in that he could no longer rule out hikes, and that the Fed’s next words should make a cut no likelier than a rise. When the president’s own appointees begin, in public, to prepare the ground for higher rates, the rescue that the entire stock market quietly assumes is coming should be understood for what it is: the most fragile hope in the building.

The Japan scenario: what 1.2 trillion dollars can do, and why it has already begun

The most catastrophic of the plausible scenarios does not originate in Washington. It originates in Tokyo, and it is not hypothetical. It is underway.

Japan is the largest foreign holder of US government debt, sitting on roughly 1.2 trillion dollars of it, a position accumulated over decades for one specific reason. Japanese government bonds paid almost nothing, so Japanese life insurers, pension funds, and banks were forced abroad in search of any yield at all, and American Treasuries were the natural home. That entire structure rested on a single condition: that Japanese rates stayed on the floor. The condition has now failed. The Bank of Japan has been raising rates, and a hotter domestic inflation has pushed Japanese government bond yields up to levels that, for the first time in a generation, make them a genuine alternative to US Treasuries. A Japanese institution can increasingly earn an acceptable return at home, in its own currency, without the exchange-rate risk that holding dollars entails. The reporting out of Tokyo in May was explicit that this repatriation is no longer a theory but a beginning.

Follow the mechanics of what happens when that institution decides to come home, because this is where the doom loop reveals itself. To sell a US Treasury, the institution must find a buyer, and it sells into a market where the Federal Reserve is no longer the buyer of last resort it was during the years of quantitative easing. More sellers and no marginal new buyer means the price of the bond must fall to clear, and a falling price, by the iron seesaw, means a rising yield. A sustained wave of Japanese selling therefore pushes US long-term yields higher mechanically, independent of anything the Fed or the inflation data does. The selling institution then converts its dollars back into yen to redeploy at home, which means selling dollars and buying yen, which pushes the dollar down and the yen up, and a weaker dollar makes every imported good, above all oil priced in dollars, more expensive for Americans, feeding the very inflation that started the cycle.

The danger is that each step worsens the next. Higher US yields and a falling dollar reduce the value of the Treasuries that every other foreign holder is sitting on, sharpening their incentive to sell before the losses deepen, which pushes yields higher still. The United Kingdom holds close to 900 billion dollars, China roughly 700 billion. The question that should keep policymakers awake is not whether Japan dumps its entire position in a single afternoon, which it will not, but whether a steady, rational, undramatic repatriation by the largest creditor on earth establishes a floor under US yields that the Federal Reserve is powerless to break, precisely because it cannot cut rates into an inflation without accelerating the dollar’s decline and handing every foreign holder a fresh reason to sell. In that scenario, the Fed’s traditional tool is not merely ineffective. It is counterproductive, and pulling it makes the fire worse.

What makes the Japan scenario the most important of all is that it removes the comforting assumption buried inside every optimistic forecast: that American interest rates are ultimately set by American conditions and controllable by American institutions. They are not. They are set, at the margin, by a global pool of creditors, and the largest member of that pool has begun, quietly and for entirely sound reasons of its own, to head for the door.

The refinancing wall: the detonator hiding on the calendar

If the Japan scenario is the most catastrophic, the refinancing wall is the most certain, because it does not depend on any actor choosing to do anything at all. It depends only on the calendar, and the calendar does not negotiate.

Here is the setup. During 2020 and 2021, when the Federal Reserve had pinned rates near zero, corporations borrowed with abandon at rates that will not be seen again in this generation. Investment-grade companies locked in coupons below 2 percent. High-yield issuers, the riskier firms that would normally pay 6 or 7 percent, secured money at 4 percent or less. It was, as one credit manager put it, a gift, and the gift came with an expiration date written into every contract. That debt does not vanish. It matures, and on the day it matures the company must either repay it in cash, which most cannot, or refinance it at whatever rate the market demands that day.

The scale is not abstract. Roughly 1.35 trillion dollars in non-financial corporate debt matures in 2026 alone, and S&P Global data shows about 7.3 trillion dollars of rated corporate debt coming due over a 36-month window, with companies typically forced to refinance twelve to eighteen months ahead of the actual maturity, which drags the pressure forward into the present. Consider what the refinancing does to a single ordinary balance sheet. An investment-grade company rolling a 2 percent coupon into a 5 percent coupon does not suffer a modest rise in interest expense. It watches that expense more than double. For a firm carrying several billion dollars of debt, that is hundreds of millions of dollars a year that used to flow to shareholders as profit and now flows to bondholders as interest, vanishing from the earnings line that is the sole justification for the stock price.

The strain is already visible in the credit markets for anyone who looks. Credit spreads, the extra premium investors demand to hold corporate debt over risk-free Treasuries, have been widening, with investment-grade spreads around 120 basis points and high-yield spreads pushing toward 470, the market’s way of pricing in the rising probability that some of these borrowers do not survive the refinancing at all. The weakest issuers face the starkest arithmetic, because they confront both a higher base rate and a wider spread stacked on top of it, and for some of them the new all-in cost of debt will exceed what their operations can possibly service. Those firms do not refinance. They default, and their failures widen spreads further, raising the cost for the next firm in line.

This is why the refinancing wall is the detonator rather than merely a symptom. The discount-rate effect and the capital-competition effect work on investor psychology, and psychology can in principle reverse on a change of mood. The refinancing wall works on contracts with fixed dates, and it converts the abstract problem of high rates into the concrete problem of specific companies missing specific earnings and, in the weakest cases, failing outright. It is the mechanism by which a signal in the bond market becomes an event in the stock market, and it cannot be talked out of happening, because no amount of optimism moves a maturity date.

What the “Magnificent 7” companies would actually have to earn

Now to the question at the center of the whole mania, posed with the specificity it deserves. The market is at record highs because it is making one enormous bet: that an artificial intelligence boom will lift company profits so far and so fast that it outruns every force described above. As long as that bet pays off quickly enough, the optimists win. So the question is not rhetorical. What would the companies carrying this bet actually have to deliver, in hard profit, to justify what investors are paying, and is that delivery physically plausible?

Start with the broad measure. The cyclically adjusted price-to-earnings ratio, created by the Nobel laureate Robert Shiller to compare prices against a full decade of inflation-adjusted earnings, sits near 42 against a historical average of roughly 17 over almost 150 years. It has crossed 40 only once before in American history, right before the dot-com crash of 2000, after which the market lost roughly half its value. Translated, investors are paying about 42 dollars for every dollar of durable annual earnings when history says the going rate is 17. For the market to return to its normal valuation without prices falling at all, corporate earnings would have to roughly two-and-a-half times themselves while prices stood perfectly still, simply to make today’s prices look ordinary in hindsight. Shiller cautions, fairly, that the measure cannot predict the timing of a fall. It has a far better record predicting the decade of weak returns that follows, and it is now at the second-highest reading in its recorded history.

Apply that lens to the giants, where the numbers turn vivid. The seven largest technology companies carry a combined value near 22.5 trillion dollars and make up roughly 35 percent of the entire S&P 500. This is not 500 companies thriving together. It is a handful of AI bets holding up the whole room, and beneath the record headlines the number of stocks actually rising has been quietly shrinking, the pattern that has historically come before a turn rather than during a healthy climb. Nvidia alone commands a value near 5.07 trillion dollars, larger than the entire economy of Japan. Microsoft sits above 4 trillion, Apple near 4 trillion, Alphabet above 3.2 trillion, Amazon near 2.5 trillion, Meta near 1.9 trillion. To justify a 5 trillion dollar valuation on any conventional basis, a company must eventually generate annual profits in the range of several hundred billion dollars, because at maturity a company is worth some reasonable multiple of its earnings, and no firm in history has sustained the kind of multiple that would let 5 trillion dollars rest on today’s profit base indefinitely. The valuation is not pricing the company’s current earnings. It is pricing a future in which those earnings multiply several times over and then stay there, fending off every competitor, every customer that tries to build its own chips, and every cyclical downturn in demand, for a decade or more.

Here the spending data delivers the decisive blow. The major technology companies are projected to spend roughly 725 billion dollars on capital expenditure in 2026, the overwhelming majority of it on the data centers, chips, and power infrastructure that artificial intelligence demands. Set the two figures side by side and the gamble comes into focus. The market has assigned 22.5 trillion dollars of value to seven companies that are, together, spending three-quarters of a trillion dollars every year on the bet that the revenue will eventually arrive to justify it. For that spending to earn even a pedestrian return, these companies must produce hundreds of billions of dollars in genuinely new, incremental, durable AI profit, not someday but on a schedule fast enough to service the depreciation the spending itself creates, because the chips being bought today grow obsolete in a handful of years. The revenue must arrive quickly, or the assets lose their value before they have paid for themselves.

That is the precise hinge on which the entire market now balances, and it can be stated without melodrama. Either AI revenue scales into the hundreds of billions of dollars of incremental annual profit within a few years, justifying both the valuations and the 725 billion dollars of annual spending, or it does not, in which case the spending becomes a sunk cost, the valuations fall toward the earnings that actually materialized, and seven companies representing 35 percent of the index lead the whole market down in proportion to how far their prices ran ahead of their profits. There is no version of the arithmetic in which 22.5 trillion dollars of value is justified by hope alone.

Follow the smart money, not the White House

If you remember only one thing from this account, make it this. The single most revealing signal is not in any chart or yield or valuation. It is in what the best-informed investors are doing with their own money, right now.

The great institutions, the funds with research floors and a professional terror of being the last one holding, have been cutting their technology exposure at one of the fastest paces in a decade, feeding their shares into the very records the headlines celebrate. Michael Burry, the investor who saw the 2008 housing collapse coming when almost no one else did, has placed heavy bets directly against the semiconductor companies. These are the people who understand precisely what they own, and they are walking, without hurry and without announcement, toward the exit while the band still plays.

The ordinary investor at home is doing the exact opposite, pouring savings into technology funds at a record pace, buying with both hands the very thing the professionals are unloading. This divergence, the informed money quietly handing its inventory to the uninformed near the top, is among the oldest and most reliable signs that an ending is near, and it was unmistakable in early 2000, when the institutions sold the dream to the public at the summit and the public spent the next ten years discovering what it had actually bought. The pattern is repeating now, only louder, amplified from the highest office in the country, where the same records the smart money is selling into are waved like a banner and the only sin is staying out. The cruelty of it is plain. The least protected people in the market are being urged to commit everything at the exact moment the best protected are stepping quietly back. When the people with the most information and the most to lose are leaving, the prudent response is not to rush in and take their seats.

The benign (albeit unlikely) scenario

Intellectual honesty requires building the optimistic case at full strength, because it is genuinely possible and an enormous amount of capital is wagered on it.

In this scenario, the geopolitical fever breaks. A settlement or de-escalation with Iran reopens the Strait of Hormuz, and the roughly one-fifth of the world’s seaborne oil that runs through it flows freely again. Oil prices, which surged more than 45 percent after the late-February strikes, eventually fall back toward their prewar range. The energy-driven component of inflation, which is the largest single driver of the present spike, drains out of the data over two or three quarters. With inflation cooling on its own, without the Fed having to force it, long-term yields ease, the discount-rate pressure on stocks relaxes, the refinancing wall is scaled at merely uncomfortable rather than ruinous rates, and the technology companies are granted the one thing they most need, which is time for AI revenue to grow into the lofty, mind boggling valuations already paid.

While this scenario remains unlikely, it is nevertheless possible and cannot be entirely disregarded. It carries, however, a complication its proponents rarely follow through to the end. The single greatest threat to the benign path is the attempt to force it. Should the administration succeed in pressuring the Federal Reserve into cutting rates aggressively while inflation is still elevated, in pursuit of the cheap money it has demanded all year, the cut would not rescue the situation. It would invert it. Easing into live inflation would weaken the dollar, and a weaker dollar raises the price of dollar-denominated oil and imported goods, reigniting the very inflation the cooling was meant to resolve, and handing every foreign Treasury holder, Japan first among them, a fresh and urgent reason to sell. The benign scenario survives only if it is allowed to happen naturally and is actively sabotaged if anyone reaches for the obvious lever to accelerate it. The thing most likely to kill the soft landing is the attempt to manufacture one, and the attempt is exactly what the present administration has spent a year promising to make.

The more-likely adverse scenario

The adverse scenario is not a single event but a sequence, and tracing the sequence reveals why a correction from this altitude does not resemble an ordinary pullback.

It begins with yields staying sticky near 5 percent or climbing, whether driven by persistent inflation, by Japanese and other foreign repatriation, or by the sheer volume of new Treasury issuance required to fund deficits on a debt past 39 trillion dollars. The discount-rate effect compresses the richest valuations first, which means the technology leaders that constitute 35 percent of the index. As their prices soften, the passive index funds that hold the savings of tens of millions of ordinary investors are forced to sell in proportion, because those funds must mirror the index, and a market this concentrated means the giants’ decline mechanically drags the whole index down regardless of how the other 493 companies perform. At the same time, the refinancing wall does its work on corporate earnings, the weakest issuers begin to default, credit spreads widen, and the cost of capital rises across the board, compounding the pressure on the very profits that valuations depend on.

Each of these effects accelerates the others. Falling prices trigger forced selling, which lowers prices further. Defaults widen spreads, which raises borrowing costs, which produces more defaults. Foreign selling raises yields, which deepens the discount-rate pressure, which lowers prices, which prompts more foreign holders to cut their dollar exposure. This is the anatomy of a correction that overshoots, and the historical record at this valuation is unambiguous about how far an overshoot can run. The only two prior occasions on which the broad market reached a cyclically adjusted valuation near today’s resolved in declines of 89 percent and 78 percent from their peaks. Those figures are not predictions of identical outcomes. They are the historical range of what becomes possible once a market priced for perfection meets a reality that is merely ordinary, and they establish that the downside, should the adverse sequence run its course, is measured not in the 10 or 20 percent of a routine correction but in fractions of the market’s total value.

The asymmetry that should govern the only decision you control

Stand back from the scenarios and a single structural feature dominates the whole landscape: the positioning of those who know most against those who know least, and the asymmetry of what each side stands to gain and lose.

On one side stands a benign scenario that is possible but fragile, dependent on a geopolitical de-escalation outside anyone’s control and actively endangered by the very policy most likely to be attempted. On the other stand multiple adverse scenarios, the Japanese repatriation already in motion, the refinancing wall already stamped onto the calendar, the valuation arithmetic already demanding a near-impossible delivery of future profit, each of them self-reinforcing, several of them requiring no one to make any decision at all in order to unfold. The probabilities are not symmetric. The magnitudes are not symmetric, and the people best equipped to weigh both have already chosen their side and acted on it with their own capital.

When the two markets finally reconcile, the logic is brutally simple. If you can earn a safe 5 percent lending to the government, why hold expensive, risky stocks the best long-run measures say will return nothing? As capital makes that switch, money drains from stocks and prices fall, while the companies that borrowed cheaply must refinance at far higher rates that eat the profits which justified their prices in the first place. The cost of money and the price of stocks are moving in opposite directions, and only one of them is tethered to anything real.

You do not need to predict which scenario unfolds, or when. You need only an honest reckoning with what you are looking at: two markets forecasting two futures, the smartest money in the world quietly betting on the darker one, ordinary savers urged from every direction to bet everything on the brighter one, and a set of mechanisms, running through Tokyo’s portfolio decisions, through the maturity dates stamped on more than a trillion dollars of corporate debt, and through the gap between 22.5 trillion dollars of valuation and the profits seven companies can plausibly produce, none of which requires prophecy to see. It requires only the willingness to follow each mechanism to the place it actually leads, and to make certain that when the reckoning arrives, and it is coming, you are not standing among those left holding what the informed money has already, quietly, sold.

The bond market has named the price of its own salvation with precision. The price is the stock market. The only thing left undecided is who will be standing there to pay it.