Newsletter · 2026-07-28 · Ashutosh Agarwal

Japan and the Yen Now Drive the Long End More Than Washington Deficits - The Long End & Fiscal Supply - Week of July 28, 2026

Rates and fiscal-supply newsletter for the week of July 21 to 28, 2026. The 30-year Treasury has paid 5% or more for 27 straight days, but the podcast tape spent the week on Tokyo, where the yen hit a 40-year low and the 30-year JGB yields a record 3.98%.

The Long End & Fiscal Supply

Week of July 21–28, 2026: Japan and the Yen Now Drive the Long End More Than Washington Deficits

The 30-year Treasury bond has now paid 5% or more for 27 days straight, the longest run since 2007. That alone would make for a busy week. But the interesting thing about this week's podcasts is how little of the worry was about Washington. It was about Tokyo, where the yen just hit a 40-year low and the 30-year Japanese government bond is yielding more than it ever has in history. The Fed decides tomorrow. The Bank of Japan decides at the end of this week. Two people on two continents can move your bond portfolio in the next five days, and the more dangerous one may not be the American.

TL;DR

• The 30-year is stuck above 5% and the reason has changed. It is no longer only about deficits. Big technology companies are now issuing enormous amounts of 30-year debt of their own, and they are competing with the US Treasury for the same pool of long-term money. BlackRock and UBS both said so this week.

• Japan is the live wire. Odds of a Bank of Japan rate hike this week jumped from single digits to 38% in a day. The yen is at its weakest since 1986, the 30-year Japanese government bond yields a record 3.98%, and Japan is the largest foreign owner of US Treasuries, and has been selling.

• Tomorrow's Fed meeting is a genuine coin-flip, and the logic is upside-down. Roughly 36% odds of a hike are priced in. If Warsh does nothing, that is looser policy than the market expects, and the long end could go up, not down. A hike might be the thing that pulls long yields lower.

What's new

1. The 30-year's 27-day stay above 5%, and the new reason for it

Marketplace, "What's driving up the 30-year Treasury yield?" (22 July) · listen Speakers: Stephen Lipley, global co-head of bond ETFs at BlackRock; Ian Shepherdson, chairman of Pantheon Macroeconomics; Leslie Falconio, head of fixed income strategy at UBS Wealth Management; reported by Sabri Ben-Nashour.

Two reasons were given, and the second one is the new one.

The familiar reason is fiscal. Shepherdson pointed to "the intractability of the huge budget deficit that the US has been running for some time," and government debt held by the public crossing 100% of GDP in March. His verdict was blunt: "There's no plausible, credible plan to reduce that anytime soon."

The newer reason is competition. Lipley's framing of why anyone demands extra yield for a 30-year bond is worth quoting because it is the whole argument in one line: "You tie yourself up for 30 years, you're locked in. And so you're going to potentially demand a premium to take that risk." What has changed is that investors now have somewhere else to lock that money up. As Lipley put it: "All of a sudden you have this very large issuance boom in AI that's necessary to build out the infrastructure. This is all happening at the exact same time."

That is the important structural shift. For decades the US government was essentially the only issuer of very long, very safe-ish dollar debt at scale. It now has rivals. And Falconio noted the buyers are showing up for them: "Pension funds, insurance companies, asset liabilities managers, you know, they love these yields that we're seeing."

Why it matters: If the long end is being repriced partly because of private competition for duration rather than only government profligacy, then fixing the deficit, which nobody is about to do anyway, would not fix the whole problem. It also means the AI capital-spending cycle and the Treasury market are now the same trade.

2. Japan became the acute risk, and it happened this week

Mining Stock Daily, "Barry Knapp: The AI Trade is Nearing an Inflection Point" (24 July) · listen Speaker: Barry Knapp, Ironsides Macroeconomics, reporting live client flow.

Knapp laid out the chain of events he is worried about, and then gave the number that made it urgent: "the probability of them hiking next week has gone from single digits to 38% as of this morning. Yesterday, that was surging. I was talking to fixed income accounts and hedge funds about it all morning. And the equity market is just whistling past the graveyard."

His mechanism runs like this. The yen keeps falling, which forces the Bank of Japan to act. Currency intervention, buying your own currency in the market, does not work for long, so they have to raise interest rates instead. And then: "we know the Japanese are the biggest global holders of fixed income. So you put more pressure on the back end of the treasury market. And you potentially have a whole risk-off scenario unfold."

Crucially, he thinks the damage happens even without a hike: "even if we don't get a rate hike next week, and I don't think we will and we shouldn't, we're likely to get a hawkish press conference." His historical parallel is May 2004, when the Fed started talking about removing support "in a measured way," nobody understood what Greenspan meant, and then they hiked: "boom, 10-year treasuries went up 120 basis points or 110 basis points in six weeks. And real rates shocked higher and vol shocked higher."

He also explained why a weak yen no longer rescues Japan the way it used to, because Japan buys all its oil from abroad, and oil is rising: "Last night's trade surplus, trade deficit, excuse me, was supposed to be, I don't know, 120 billion yen or something and it was 400 and change. And that was because crude imports were up 58% or something. So Japan doesn't benefit from a weak yen the way they once did because there's short crude."

Why it matters: This is the first week in over a month that Japan got serious airtime, and it arrived with a date attached. Knapp's conclusion, "having an August risk-off is looking like an increasingly probable event," is a specific, falsifiable call from someone who spent the week on the phone with bond desks.

The numbers behind it came from The Peter Schiff Show Podcast, "Japan Is About to Pop the Biggest Bubble in History... And It Takes Us With It" (26 July) · listen. The yen fell to 163.8 per dollar, the weakest in 40 years, since 1986. The 10-year Japanese government bond yields 2.8%, the highest since 1996. The 30-year yields 3.98%, and as the show noted, that is not a 30-year high but an all-time high, because Japan did not start issuing 30-year bonds until 1999. Japan's debt is above 200% of GDP. And its policy interest rate is still just 1%.

Schiff's read on the trap: "either they're gonna act aggressively and we're gonna have a crisis in Japan that's gonna spill over into the US, or they're gonna be too timid and we're gonna have a different sort of crisis that is also gonna spill over in the US."

3. Tomorrow's Fed decision is a real coin-flip, and not hiking might raise long yields

The Financial Exchange Show, "Warsh Faces a Real Fed Test as Oil Risks Build" (24 July) · listen

The pricing: "there is a 64.2% chance priced in that there is no hike. There's a 35.8% chance of a hike priced in. That's pretty good. We haven't seen like a meeting be this, you know, up in the air in a while." Meanwhile economists surveyed by Bloomberg see no move at all until a cut in the third quarter of 2027, a spread between market pricing and forecaster consensus that is itself remarkable.

Then the part worth thinking about carefully. Because a hike is partly priced, doing nothing is an easing: "If the Fed does nothing at this meeting... It is implicitly the Fed creating easier monetary policy because the market right now is priced for the chance that they actually hike." Which flips the usual intuition: "if they don't hike next week, watch the long end of the yield curve because I think it goes higher."

And the reverse: a hike would be a credibility signal that could pull long yields down. "A single hike is both incredibly meaningful and incredibly not... it's not going to actually change the course of inflation... But the signal alone might be powerful enough that it says, hey, this guy's not messing around on this. And that could actually bring the long end of the yield curve in."

Why it matters: If you own long bonds, the naive trade, hoping the Fed stays on hold, may be exactly backwards. The long end wants proof of resolve more than it wants cheap money.

4. Britain got a new government and a 5% gilt in the same week

Wake Up to Money, "Cabinet compass" (22 July) · listen Speakers: BBC presenters with a markets commentator (Russ) and Lord Boateng, former Treasury minister.

A new UK cabinet under Prime Minister Andy Burnham, with John Healey, previously the defence secretary and a "slightly unexpected choice," installed as Chancellor in place of Rachel Reeves. Day one brought an immediate tax cut and about £45 a year off electricity bills.

The bond market's response was not panic, but it was not welcoming either. As the programme's markets commentator put it, the 10-year gilt "has just crept over 5% for only the fourth time since 2007." Sterling barely moved. But the comparison he drew is the one that should sting: "it does cost the UK more to borrow for 10 years than it does Portugal, Ireland, Italy, Greece or Spain, countries whose finances were in absolute shambles 10 or 15 years ago and were a bit of a laughing stock." And if Britain were still in the EU, "only three members of that zone would actually currently pay more to borrow for 10 years than the UK."

What the market wants is specific and unglamorous: "they're looking for the government to stick to those rules and not start fudging them or changing the way in which they measure debt or finding wheezes to keep spending off balance sheet through private finance initiatives as they used to do." Investors are not against borrowing, "They're not frightened of governments investing," provided "there is a clear path to getting a payback." The cautionary example was named directly: "which is where, in the end, Liz Truss went wrong, what, three, four years ago?"

The tension is already visible. Healey wants a clear route to defence spending of 3% of GDP. Lord Boateng defended it as achievable "within the existing fiscal rules which do allow room for borrowing in order to invest, including investment in our national security," which is, stripped of the diplomacy, an argument for more borrowing.

Why it matters: Britain is the cleanest live experiment in whether bond markets still discipline governments. A brand-new Chancellor with a defence spending ambition and a 5% 10-year yield is a setup, not a resolution. Watch it as the leading indicator for everyone else.

5. The supply problem is not the size. It's the shape and the buyer.

Marketplace All-in-One, "Unemployment filings hit a 55-year low" (23 July) · listen Speaker: Greg Ip, Wall Street Journal, discussing his piece "How Sky-High Deficits Threaten the Bond Market"

Ip made three points that fit together into the most complete account of the supply problem anyone gave this week.

First, the scale: "every year, the Treasury has to come to Wall Street and say, 'We need to borrow $2 trillion by selling you Treasury bills and Treasury bonds.'" These are, he said, "the largest deficits relative to GDP that we've ever run in peacetime on an ongoing basis."

Second, the buyer has changed character. "There are fewer patient investors, like foreign central banks, and more sort of like impatient investors like hedge funds. And these are the kinds of people that will trade in and out of markets a lot faster. And they may flee if something goes wrong."

Third, and this is the mechanical heart of it, the Treasury has been managing the optics by borrowing short. There are two ways to borrow: Treasury bills, which mature in under a year at lower rates, or longer-term bonds, which cost more and, if you issue a lot of them, push up market interest rates and therefore mortgage rates. As Ip explained: "what we've seen for the last few years, and in fairness, this happened under President Biden also, is that even though the deficits are very large, the Treasury has kept down the size of the auctions of long-term bonds because it doesn't want to put upward pressure on long-term interest rates. But that causes a problem. It means by relying so much on Treasury bills, more and more of the debt comes due and must be refinanced every month and increases the risk that something goes wrong."

Put the second and third points together and you get the real exposure: "not only are we coming to market much more often to refinance this debt, but we're asking people who have no long-term commitment as patient holders of that debt to step up and lend us the money."

Does Washington understand? Ip says yes, and that this is precisely the problem: "Everybody has known for a long time... nobody wants to say that probability is high enough to say it's going to happen soon... crises are like that. We can see the contributing factors. Nobody can call the moment." Both parties are "politically incapable and unwilling to deal with the underlying problem, which is that our taxes are too low and our spending is too high."

Why it matters: Short-dated funding is cheap right up until the month you cannot roll it. This is the single most under-appreciated risk in the whole complex, and it is entirely self-inflicted.

One footnote worth flagging, from the same episode: the jobless claims number that helped drive rate-hike odds higher this week may be a broken gauge. Claims fell 22,000 to the lowest level since 1969. But Michele Evermore of the National Employment Law Project said flatly, "Initial claims data is no longer a very reliable economic indicator," fewer people bother applying, benefits are thin, and eligibility has been tightened. University of Michigan economist Betsy Stevenson noted fewer than one in three unemployed people are even eligible. Entry-level job postings are down 6.3% year over year while senior-level postings are up 15%. So the strong-labour-market print that pushed hike odds up rests on a statistic that two labour economists say no longer measures what it used to.

The debate

Both sides showed up this week. The bears were louder, more numerous and more specific. But the bull case was more careful, and it came from the single most credentialed voice in the whole set.

The bear case, in three distinct flavours

These get conflated constantly, and they shouldn't be, because they imply different trades and different time horizons.

Flavour one: the arithmetic bears. No plan, degraded buyer base. This is Shepherdson ("no plausible, credible plan") and Ip ("taxes are too low and spending is too high"). It is not a forecast of a crash; it is a statement that the distribution of outcomes has fattened on the bad side. Its weakness is that it has been true for years.

Flavour two: the regime bears. These are the people who think the level of real interest rates is going structurally higher and is not coming back.

Russell Clark, a London hedge fund manager, gave the most extreme version on Monetary Matters with Jack Farley (22 July) · listen and again on Other People's Money with Max Wiethe (22 July) · listen. His target is a 10% US Treasury yield, and the reasoning runs through housing politics rather than bond math: "If I look at people 40 and under, those in their 20s and 30s, their number one problem is they can't afford housing. If you want to get housing back to some more reasonable levels, you need to have wages rising at about 7% a year, so doubling in 10 years. And then you need to have the housing market be flat in nominal terms, so falling in real terms. So that requires you to have a real rate of about 3%. So people keep their money on deposit rather than sticking to real assets. So that gives you an interest rate around 10%."

He also gave the sharpest single statistic of the week on the fiscal position: government revenue "now sort of barely covers its sort of mandated expenses of like Social Security interest payments and these sort of things. I think we're about 90%." That is before defence, education or infrastructure.

Clark has been bearish since 2022, and the trigger was geopolitical rather than fiscal: when Russian foreign reserves were frozen after the invasion of Ukraine, he asked "why would any country that could theoretically disagree with the Trump administration, which is basically everybody... why would anyone hold treasuries as foreign reserves?" He also made a historical point that is easy to forget: "Until 1980, the idea of holding another country's fixed income as a foreign reserve was unknown. All foreign reserves were basically gold." And he thinks the evidence is already visible at the edges: "if you look at markets with more peripheral sovereign bond markets, Japan is a big one... the yields there have risen tremendously. But the UK as well, the gilt market remains very unstable. The long end keeps selling off."

Chris Whalen of Whalen Global Advisors gave the domestic version on The Julia La Roche Show (25 July) · listen. Asked how the Fed can credibly target 2% inflation while the Treasury runs a deficit of 6% of GDP, he said: "I think it's hard... at some point the Fed is going to have to reconsider that 2% inflation target because it's laughable. We're not going to get there. The Treasury's deficit spending is making this economy run hot and everybody kind of likes that."

His point about what the shock actually is deserves attention, because he is explicit that he ignores consumer price indices: "I focus on big picture commodity price inflation, energy, key inputs, if you will, at the top of the food chain because the measures that we have at the bottom... are so heavily manipulated and qualified that to me they're almost meaningless." His example is unnervingly concrete: "The price of sulfur and sulfuric acid since the start of the Iran war has gone up 150%... There is nothing in this economy that doesn't depend on sulfur. Because it's a key manufacturing component for things like fertilizer. You can't grow food without sulfuric acid." He expects consumer inflation to get "very close to double-digit," and notes "we're already there in the wholesale world. We've been there for a couple of months."

On what higher-for-longer actually means for households: mortgage rates of "somewhere around 6.5% to 7%," with a rule of thumb, "If you look at the yield on the 10-year Treasury bond... add 1.5 to 2 points and that'll kind of tell you where it is." He also flagged that the Bureau of Labor Statistics has been redefining how it measures several inflation components, which he read uncharitably: "When you have a problem that you can't fix... you move the goalposts."

Flavour three: the plumbing bears. Knapp's Japan-triggered August risk-off, above. This is the only flavour with a date.

The bull case

The strongest version came from Richard Clarida, former Vice Chair of the Federal Reserve, now at PIMCO, on Macro Hive Conversations With Bilal Hafeez, "Ep. 368" (24 July) · listen. This is the only voice in the entire week who has actually voted on US interest rates, and his argument is a direct rebuttal to the regime bears.

His central claim: this is not a wage-price spiral, and the 1970s comparisons are wrong. "I get tired of all the analogies to the 1970s... a very important difference between this cycle and the 70s is that in the 70s, you had not only high price inflation, you had very high wage inflation. You had what economists call a cost push cycle in the labor market, which was driving up price inflation. We don't see that at all now. In fact, if you look at wage inflation adjusted for productivity, it's running right where it was back in 2018-2019."

Which matters enormously, because it means "the labor market is really not a driver of inflation right now." Even the immigration-crackdown fear did not materialise: "there was a lot of hand-wringing about if there is a crackdown... it will create a problem for the Fed because it'll push up wage inflation because you won't have enough workers. We didn't get that."

He supported it with a striking structural fact: "labor share of national income is now the lowest it's ever been in the data going back 80 years." Workers are not the ones capturing the gains, which is bad news socially and good news for inflation.

So where is the pressure coming from? The same place as the bond supply story: "the upper pressure on inflation, while it's not coming from the labor market, is... coming from the AI CapEx boom. And in particular, on the headline basis, is pushing up electricity prices; on a core basis, pushing up the price of memory chips." He put a number on it: "we've done some calculations at PIMCO... that this year alone, the direct effect on the PCE price index of memory, higher memory and software prices will probably add about half a percentage point to core inflation."

And the counterfactual: core inflation "got down to 2.4 in September of 2024," and "if you make a calculation for the direct effect of tariffs and memory chips, would probably be around two right now." Other central banks made it: "before the hostilities in the Middle East, inflation in the Eurozone was running at two. Inflation in Canada was running at two." On the recent data he was constructive: the soft June CPI was "welcome," core came in softer than expected with downward revisions to prior months, producer prices were soft too, and PIMCO colleagues expect the monthly PCE reading "somewhere below 0.2."

He is not naive about the politics of it. He cited Governor Chris Waller's point that "after five going on six years of missing the inflation target, even if you can convincingly argue a non-monetary reason why inflation has been above target, at some point the Fed has to take ownership for that miss."

The second bull argument is simpler: 5% is a level where buyers appear. Falconio's pension-fund and insurer demand, above. And there is a precedent, laid out on The Banker Next Door, "Strategy Room Bonus" (26 July) · listen: since the 2008–09 financial crisis the 10-year has reached 5% exactly once, in October 2023, and "the yield stayed above 5 percent only for part of a morning with that round number leading to a surge of demand from investors. It was back below 4.9 percent by that afternoon and under 4 percent by the end of the year."

Third, the oil shock cuts both ways. Marc Chandler, chief market strategist at Bannockburn Capital Markets, on The KE Report (24 July) · listen, leaned against a hike tomorrow: "the Federal Reserve recognizes, at least traditionally, that higher oil prices cut both ways. The first punch is obviously higher inflation or higher inflation expectations. But the higher prices for oil act like a tax on American consumers. This would slow down consumption... I would lean against a rate hike this week. I think that there's, it would come as a big surprise. I don't think Warsh wants to do that."

Chandler also supplied a small but real dent in the "foreigners are done buying" story. The Fed acts as custodian for foreign central banks, so their holdings are observable: those holdings "fell for the past four weeks before rising in this latest week through Wednesday of this week." One week is not a trend. But the bears should notice it.

Danny Moses, of The Big Short fame, was similarly unconvinced on RiskReversal Pod (27 July) · listen: "I don't think the Fed is hiking this year." He expects Warsh to argue oil is "temporarily high," and noted Warsh "is a big proponent that AI is deflationary, that there'll be a massive deflationary effect from all this CapEx spend going forward." Moses does expect something new tomorrow, though: "I think there will be dissenters this time. The last meeting was his first meeting, so they all will all just agree."

The evidence that adjudicates between the bear stories

This was the sharpest analytical moment of the week, from The Wall Street Skinny, "Why Jamie Dimon Won't Buy Stocks OR Bonds Right Now" (24 July) · listen.

There is a popular story that AI and technology bond issuance is "crowding out" the Treasury market, that investors are choosing Meta's debt over the government's. The hosts pushed back with a test. A swap spread is, roughly, the gap between the yield on a Treasury and the cost of an equivalent interest-rate swap; it widens or narrows depending on whether Treasuries specifically are in trouble versus everything being in trouble. As one host put it: "If Treasuries were selling off in a more idiosyncratic way, we would expect to see swap spreads tighten. They're not, okay? It's that nobody wants any of this stuff."

And the corporate bonds are doing worse than Treasuries, not better: "if you look at the 30-year bonds that they've issued recently, Meta's are 40 basis points wider, and Oracle's and SpaceX's are 60 basis points wider since they were issued in the last month. That is a massive move... They want them a heck of a lot less than they want 30-year Treasuries."

The conclusion: "You are seeing VIX sell-off and Treasury bonds sell-off and now equities selling off. Nobody wants anything. The risk premium baked into pretty much every asset class seems insufficient right now."

So the crowding-out story is probably wrong as stated. Investors are not preferring tech debt to Treasuries. But the broader competition-for-duration point from BlackRock survives, and the generalised repricing of risk is arguably worse news than a clean rotation would be.

Trades in play

• Long-end Treasuries, short. Clark's 10% target is the most aggressive expression of it, and he runs money against the view.

• Bonds vs equities, for the first time in years. The Wall Street Skinny's framing: with real (inflation-adjusted) yields in the "low to mid 2%" range, "at a certain point, you become incentivized to take your money out of super risky stuff like the stock market and put it in safer stuff like the bond market, where you can earn a relatively high risk-free return as long as you hold these things to maturity." The caveat they flagged themselves: that only holds if you hold to maturity, because you still carry price risk if you sell.

• Gold, held not chased. Moses: "I'm a believer in it. I'm not selling it here, nibbling a little bit here. I'd actually be more prone to buy it on the way up." His trigger is the Fed: "if there's any blinking by the Fed and/or we get through this Fed meeting that they're going to be dovish... I think gold will have a leg higher." Gold has touched "3,900 and change... a couple times in the last few weeks," well off the 5,500 it reached in January.

• Nothing at all. Jamie Dimon said on Monday he would not buy US equities or Treasuries at current levels, reported secondhand on The Wall Street Skinny, not from Dimon directly, so treat accordingly. The positioning data behind that mood: short interest in S&P 500 index stocks is near its highest since 2010, Russell 3000 shorts are at a record 6.3%, and SpaceX was the ninth-most-shorted stock with $25bn against it, nearly a third of its free float.

• Mortgage-sector equities, into earnings. Whalen: "The mortgage earnings this quarter are going to be very interesting because all these firms have had to manage their way through a rising rate environment. And that is not necessarily good if you don't know what you're doing." Firms positioned for falling rates and higher volumes got the opposite: "I think we are definitely going to see more M&A. And we may see some surprises too... it's hard to hedge interest rates when you have no visibility. It's like driving on a foggy morning."

Read-throughs

Mortgages and housing. Whalen's 10-year-plus-1.5-to-2-points rule of thumb puts mortgages at 6.5–7% and staying there. Marketplace made the more subtle point about why the 30-year matters even though it doesn't set consumer loan rates directly: "30-year yields don't influence mortgages or car loans the way 10-year yields do. But the thing about the long term is that after a while, it becomes the now."

Long-duration and AI-capex equities. This is now the same story as the bond market. Alphabet raised its 2026 capital spending guidance from roughly $175–185bn to $195–205bn, and, the detail that spooked people, "their free cash flow for the quarter turned negative for the first time since going public" (The Wall Street Skinny; corroborated on The Banker Next Door). Marketplace clocked quarterly capital expenditure at "almost $45 billion" with a dry aside: "definitely not a bubble." The Wall Street Skinny's estimate for the top five hyperscalers' 2026 capital spending was "basically like at $800 billion... which is like in spitting distance of how much we are spending on the US military," the host's own framing, so treat as an approximation.

Knapp added the cycle-timing metric worth writing down. Capital spending as a percentage of cash flow marked the top in previous booms: "when we got to 80% for the telecom sector in 2000, that was pretty much the peak... The same thing happened in the energy sector in 2014 and 2015... we're above that level for the big four spenders, meaning Google, Amazon, Microsoft and Meta." His read: "we've reached a real inflection point in the rate of change of spending on AI."

Corporate credit and private credit. Meta 30-year paper 40bp wider, Oracle and SpaceX 60bp wider, within a month of issue. Knapp noted the stress showed up somewhere unexpected: "while I thought we'd get some credit spread widening to signal the market was starting to struggle with that, instead what we got was redemptions from private credit. And so investors were pulling money out of there, presumably because they didn't think it was marked properly." The Wall Street Skinny also flagged that some of this debt sits off balance sheet through project financings and lease structures, "perfectly legal," they stressed, not fraud, but "Meta's credit rating that was kind of backing this debt," and "stuff like that too is starting to widen out."

Swap spreads. Not tightening. Read that as: this is a broad risk repricing, not a Treasury-specific buyers' strike. See above.

The dollar. Chandler's view is that it is tracking the US 2-year yield above all else, with the dollar index having peaked for the year around 101.80 in late June. His explanation of how America is funding itself is important: with a large current account deficit, foreigners must buy something American, and "it seems clear they're not acquiring the bonds... Foreign investors have been buying a record amount of US equities." So the rotation is "out of global bonds and into the AI tech sector."

Schiff takes the opposite side, arguing today's rate-driven dollar strength is temporary and misread: "I look at rising interest rates as a repudiation of the dollar, of lack of willingness to loan dollars to the United States at rates that we can afford to pay." He expects the dollar to eventually "decouple and start to fall." Note this is a directional call that has been wrong so far this year.

Gold. Knapp explicitly rejects the simple real-rates framework: "I don't view gold as being particularly related to U.S. financial conditions to real rates." His driver is trade rebalancing, countries like China accumulating non-dollar surpluses and parking them in metal rather than Treasuries or their own currency. Which, if right, means gold can keep working even with rising real yields and a firm dollar. The correlation break was noticed on Mining Stock Daily: gold rose this week alongside higher yields and a stronger dollar, which is "typically antithetical."

Cross-sovereign correlation. Chandler's year-to-date scorecard is the single most useful table nobody drew: 10-year yields up about 50bp in the US, 35bp in Europe, 55bp in the UK, and 72bp in Japan. On the week: US +7.5bp, Europe +2bp, Japan +10bp, Australia +13bp. His conclusion: "it does look like people globally, investors have reduced the duration. They've sold the long-term bonds." This is not an American problem with global spillover. It is a global repricing in which America is roughly in the middle of the pack, and Japan is the worst.

Japan as the transmission channel. The most important read-through of the week. Moses on the plumbing: "there is a relationship with the largest foreign holder of US Treasuries. They have been selling. Oil for them is really the stress point. They import it all." And on the policy bind, which he credits to Elizabeth Thomas: "a deteriorating bond market and a deteriorating currency, and they're going to have to sort of pick one. And in picking one, it's going to screw up the other. So I don't know if there's an elegant way out of this thing."

He also thinks Washington knows: "Scott Bessent is all over this in the sense of what he needs to do, talk to Japan and figure it out. He can't afford to lose a buyer of our Treasuries and he can't afford to have this carry-trade potentially unwind on him." The carry trade here means borrowing cheaply in yen to buy higher-yielding assets elsewhere, and an unwind means forced selling of those assets to repay yen loans.

Knapp's version extends the damage beyond Treasuries: if Japanese investors are selling, "they wouldn't just be selling treasuries, right? They'd be selling French OATs and Bunds and everything else."

What changed vs last week

Quite a lot, and mostly in one direction.

• Yields broke out rather than stalling. Last week the 30-year sat around 5.1% and the 10-year around 4.59–4.62%. This week the 30-year closed at 5.16% with an intra-week high of 5.19%, a 20-year high and the highest since 2006, and the 10-year brushed 4.71% intraday on Thursday, its highest since January 2025, closing the week near 4.68%. It also cleared its previous 2026 high of 4.687%.

• Oil went from a worry to a shock. Last week WTI was around $80 and up 20% for July. This week Chandler had the September contract up "almost 25%" in three weeks and "almost 8%" in the week alone, with Moses putting WTI near $90 and Brent near $100. This is the single biggest driver of the change in rate-hike odds.

• The hike moved forward. Last week September was the consensus month and July was not seriously in play. This week tomorrow's meeting is priced at roughly 36%, up from about a sixth a week earlier by Chandler's reckoning, and the market now prices at least one hike this year with "about an 80 percent chance of two."

• Japan and the UK stopped being blind spots. For weeks the biggest gap in this coverage was that nobody discussed Japanese government bonds, the yen or gilts. This week five separate podcasts did, with numbers. That is the most important change, because it was the tail risk nobody was pricing.

• The AI-issuance channel went mainstream. Last week's version of this story was Schiff arguing hyperscalers had become borrowers instead of lenders. This week BlackRock, UBS, a former Fed Vice Chair and a rates strategist all independently connected AI capital spending to the long end, through bond supply, through electricity and memory prices feeding core inflation, and through capex-to-cash-flow ratios at cycle-peak levels.

• The crowding-out narrative got tested and partly failed. Swap spreads are not tightening, and AI corporate bonds are underperforming Treasuries, not outperforming them. Useful discipline on a story that was getting sloppy.