Japan's bond yields just hit a 30-year high. Berkshire's CEO says that's not a problem for the trading houses — yet.

When the yield on a 10-year Japanese government bond climbs to its highest in about three decades, the reflex reaction from a company's lenders, its analysts, and its shareholders is the same: that's a problem. Debt just got more expensive. The man who now runs Berkshire Hathaway, Greg Abel, looked at that number and said the opposite — that rising Japanese rates are not a challenge for the five trading houses Berkshire owns. A claim that sounds like a mistake. It isn't. Here's the machine it's actually describing.

Start with what you'd be buying. The "trading houses" (the Japanese call them sogo shosha) are not traders in the way a stock-trader is a trader. They are sprawling, diversified conglomerates — energy, food, metals, logistics, a bit of everything — that make money off the whole pipeline of global commerce rather than a single product. Berkshire built its positions into the biggest five: Mitsubishi, Mitsui, Marubeni, Itochu, and Sumitomo. By May it had crossed the 10% ownership line in all of them, and the package is worth around $20 billion and pays a dividend in the 5% range. At the May annual meeting, Abel and Buffett were blunt: no plans to sell, "not now, not ever," a plan to hold them "for 50 years or forever". So the real question isn't whether Berkshire is exiting. It's whether the rising yield breaks the thing he's holding.

Here's the mechanism most people miss, and it's the whole ballgame. The trading houses are running a financing loop. They borrow yen, and they spend a big chunk of it buying back their own shares and funding investments. Mitsubishi, for instance, has been leaning on record yen borrowing this year to fund investments and lift its capital efficiency. The reason that works is leverage, not revenue. As long as what the house pays to borrow is less than what its shareholders could earn by keeping their money elsewhere — the borrowing rate below the "cost of equity" — every dollar of that debt is net positive, and using it to retire your own shares is one of the clearest ways to show it. These houses once traded at a discount to the value of their assets, which made buying back a share especially cheap. A lot of that discount has since closed (Berkshire's own visible ownership did some of that work), but the condition that matters is the gap between the two rates. That gap is the engine.

Now the yield. "Highest in 30 years" is doing a little sleight of hand, because the high is measured against a baseline of zero. For more than a decade Japanese rates sat at essentially nothing, so borrowing was almost free; the 10-year is now around 2.9%. That is a lot higher in relative terms and still low in absolute terms. The buyback loop only keeps giving back value while the borrowing rate stays under the cost of equity, and at ~2.9% that gap is still open. That's why "not now" is a fair sentence and not a shrug.

There's a second offset the yield is hiding. The houses are global, and a big slice of their profit is earned in dollars and other currencies before it's translated back into yen. When the yen is weak — as it has been through the past several months — that translation mechanically pads their yen-reported earnings. So the two things that look bad (higher yen borrowing costs, a weaker yen) tend to arrive together, and for a company that earns in dollars, the weak-yen piece quietly pays for a good part of the extra interest. "Right now" is doing real work in Abel's sentence. It's a statement that the offsets currently outweigh the drag.

So when does it stop being fine? There's a line, and it's worth knowing where it sits even if you can't name the exact number. The loop flips from value-creating to value-destroying the day the cost of yen debt starts to rival the cost of equity. Borrow at 5% to buy back a share your shareholders could have earned 6% on, and you're no longer creating wealth — you're paying interest to shrink your own company for nothing. The same yield is quietly taxing Berkshire's own leverage, too. It raised ¥272 billion in yen bonds this spring to fund part of its buying, and it originally borrowed at "ridiculously low" rates near 0.5%. But Berkshire sits on a roughly $380 billion cash cushion, which is a very different animal from the trading houses it's levering up.

The condition to watch, then, is not the headline "rates rose." It's the gap. As the Bank of Japan keeps hiking toward a normal rate, that gap narrows. Close it, and the trading houses either slow the buybacks — which were a big part of why the shares re-rated in the first place — or the economics of the whole position change. That's the moment "not now" quietly becomes "not yet."

The headline isn't a blunder. It's a claim about a spread, and the spread is still in your favor. The question over the next couple of years isn't whether Japan's rates rise — they're already doing that — it's how fast they cross the cost of equity the trading houses' own shareholders set. Until they do, the highest yield in 30 years is fuel, not a fire.