Why the AI Borrowing Spree, Not a Credit Scare, Is Pushing Long-Term Yields Higher

When long-term Treasury yields climb to levels not seen in decades, anyone funding a retirement feels it in both directions at once: the bond fund statement shows red, even as the income on new money edges higher. The usual suspects get the blame — the Fed, inflation, Washington's deficits. Mohamed El-Erian is pointing at an additional, less-celebrated driver: the AI giants have turned into the bond market's biggest borrowers, and their flood of long-dated debt is competing with Treasuries for the same finite pool of investor dollars. That competition changes what a higher yield actually means, and whether it is something to fear or something to reinvest into.

The same companies, borrowing instead of paying cash

For most of the past decade, the big technology platforms funded data centers and AI infrastructure out of operating cash flow — they simply did not need to borrow. As the buildout ballooned, that self-funding model stopped being enough. In 2025 the hyperscalers — Amazon, Alphabet, Meta, Microsoft, Oracle — issued roughly $121 billion of U.S. corporate bonds, more than four times their recent annual average. This year they are pushing well beyond that, and the money is the point.

Investment-grade corporate issuance reached about $1.7 trillion through July, up 27% from a year earlier and on pace to exceed $2 trillion for the first time. AI-funded borrowers now account for roughly 18% of year-to-date investment-grade supply, up from just 1% in 2024. And they are issuing where it matters most for yields: forty percent of all investment-grade debt longer than fifteen years has come from AI-linked companies this year. That is long-dated paper landing on top of long-dated Treasuries in the same slice of the curve.

How corporate borrowing becomes higher Treasury yields

Normally, a flood of corporate debt is absorbed by paying a fatter premium — wider spreads — over risk-free Treasuries. This time the striking part is that investors wanted the AI paper badly enough that spreads stayed unusually tight. When demand is that strong, the market does not adjust by pricing the corporate bonds cheaply; it adjusts by demanding more from the safe benchmark competing for the same scarce capital. Economists call it "reverse crowding out": the crowding happens on the demand side, and the price signal lands on Treasury yields rather than on corporate credit premiums.

El-Erian frames it as a mismatch between supply and reliable buyers. Government, hyperscaler, and corporate issuance now, in his words, "far exceeds what you can count on in terms of reliable buyers", at a moment when some of the traditional large Treasury holders — China, Japan, Gulf funds — are pulling back. The result is persistent upward pressure on yields that has little to do with a panic over credit quality.

This is supply, not a broken income engine

Step back, and the distinction is the whole story for an income investor. Rising yields are habitually read as a credit scare — proof that lending is turning dangerous. This situation is nearly the opposite. The issuers are investment-grade, the cash flows behind the debt are strong, and for much of the year investors were oversubscribed on the deals. The pressure is technical and supply-driven, not a sign that the income engine is cracking. That is exactly the difference that decides whether a yield move threatens a payout or simply clears a heavier calendar.

The one wrinkle is fatigue. Oversubscription on hyperscaler deals fell from as high as 4.5x in the first half of the year to about 1.4x on a later deal, new issues now carry yield concessions of 10 to 15 basis points, and Amazon priced a $25 billion long-dated sale at roughly double last year's spread. This matters because the mechanism cuts both ways: if the AI borrowing wave eases, the marginal pressure it put on long Treasuries would ease with it. The bearish reading — that rising yields are the start of a structural bond crash — leans on a one-way extrapolation of issuance bankers themselves are warning could stall.

What it means for the income portfolio

Translate all of this into the actual income question. Higher long-term yields are a gift if you are buying and holding: each new rung of a bond ladder, each new bond fund dollar, buys more coupon income. That is the "income now beats liquidation later" arithmetic working in your favor. The pain is mark-to-market — long bond funds and long Treasuries lose value as yields rise, which is uncomfortable but not, by itself, a loss of income.

The remedy is the discipline that protects any income portfolio rather than a forecast. You do not need to call the top in rates; you need a ladder, or a fund with a duration you can actually hold, whose maturities line up with when you expect to need the money. The portfolio remains the yield machine, diversified across maturities and instruments so no single rung — or single macro headline — breaks the plan.

And do not let a headline trigger a portfolio decision. The cash flows financing these bonds are real, which is why this yield move reads as a reinvestment opportunity rather than an alarm. If your income stream is still sound, the higher yield may simply mean you can buy more future income on better terms. The condition that would change that answer is credit, not price: watch whether hyperscaler balance sheets deteriorate or whether issuance stalls. It is the supply, not the scare, that is carrying long-term rates higher.