What's going on with the bond market?

Long-term interest rates jumped this fall. The 30-year Treasury yield closed at 5.67% on October 7, its highest level since 2002, and the 10-year reached 5.31% two days earlier, also a 2002 high¹. A common question a lot of folks are asking: what's going on, and should we be worried about bonds? The short answer is that three forces are pushing long-term rates up, and you can still generate good yield with low volatility by keeping bond exposure short-term.

The short version: Long-term rates are up because governments around the world are borrowing heavily, big tech companies are borrowing hundreds of billions to build AI, and foreign buyers have become a smaller part of the market. Experts disagree about how much each one matters, and I'll walk through these forces.

Monthly average yield, January 2020 to October 2026. Dashed line: the 5.67% daily close on Oct. 7, 2026, the highest since July 2002. Source: U.S. Treasury constant-maturity yields via FRED (DGS30).

Short-term rates vs. long-term rates

There are two kinds of interest rates worth separating: short-term rates and long-term rates. The Federal Reserve controls ultra short-term rates. Long-term rates, like the 10- and 30-year Treasury yields, are set by investors buying and selling bonds every day². When long-term rates go up, bonds that were previously issued and pay lower rates become less valuable, so their prices fall.

1. Governments everywhere are spending more than they take in

This isn't just a U.S. story. Long-term rates are climbing across the developed world at the same time:

• UK: the 30-year gilt yield hit about 5.9% in early September, a 28-year high, and the 10-year reached its highest level since 2008³.
• Japan: the 10-year yield touched 3.0% on September 1, its highest since 1996, on worries about government spending⁴.
• France: the 10-year yield neared 5% in early October after its worst quarter since 1987⁵.

10-year government bond yield, monthly average, January vs. September 2026. Source: OECD long-term interest rates via FRED (IRLTLT01).

Here at home, the U.S. government just closed its books on fiscal 2026, which ended September 30, with a $1.99 trillion deficit. That's up from $1.78 trillion last year, the largest in five years and the third-largest ever, at about 6.2% of the economy. Even record tax collections couldn't keep up: revenue rose 3% while spending rose 6%⁶. Net interest on the debt topped $1 trillion for the year, more than we spend on defense⁶, and the national debt passed $40 trillion in August. CBO projects annual interest costs reaching $2.1 trillion by 2036⁷.

Federal spending, $ billions per year (seasonally adjusted annual rate), quarterly through Q2 2026. Source: BEA via FRED (A091RC1Q027SBEA, FDEFX).

If you are a banker and a client keeps spending more than they earn with no plan to change, the bank charges a higher rate. Closing that gap is hard for any government, so bond investors are asking to be paid more to lend long-term.

Debt levels and deficits aren't the only reasons why rates are up. A stronger-than-expected economy and stubborn inflation, partly from higher energy prices, push rates higher too⁸. Most analysts see government debt as one important piece of the story.

You can see this in a number the New York Fed tracks called the term premium: the extra yield investors demand to lock up money in a 10-year Treasury instead of rolling over short-term bills. In 2020 it was negative; investors were effectively paying for the safety of long-term Treasuries. Today it is about 1.1 percentage points, the highest since 2010⁹.

10-year Treasury term premium, percentage points, monthly average through Oct. 9, 2026. Source: Federal Reserve Bank of New York ACM model via FRED (THREEFYTP10).

2. Big tech's AI borrowing is competing for the same dollars

This one is more uniquely American. Building data centers, and the power to run them, is enormously expensive, and the biggest tech companies are increasingly borrowing to pay for it. From 2020 to 2024, Alphabet, Amazon, Meta, Microsoft and Oracle borrowed about $35 billion a year combined in the bond market¹⁰. This year they had borrowed about $220 billion by mid-August¹¹. According to the IMF¹², these five companies went from less than 1% of high-quality corporate bond sales worldwide in 2024 to about 10% today.

What stands out to me is how long that debt is. The U.S. Treasury has long been the biggest borrower in the long-term bond market. Companies have always issued some long-term bonds, but never on this scale or this fast. Now Meta has sold 30- and 40-year bonds, and Alphabet has sold a 100-year bond¹³. The Dallas Fed estimates that this year's AI borrowing adds roughly 12.5% as much long-term debt to the market as the U.S. Treasury does¹⁴.

Why does that matter? Pension funds and insurance companies that have traditionally bought long-term Treasuries now have another option. When more long-term bonds compete for the same pool of buyers, borrowers, including the government, have to offer higher rates to attract them.

In the minutes of its September meeting, the Fed noted that market observers pointed to "competition for capital" from AI borrowing as one reason long-term rates rose, and a few Fed officials themselves cited expectations for more AI borrowing⁸. Fed Chair Kevin Warsh and Treasury Secretary Scott Bessent¹⁵ have both described big tech's borrowing as competing with Treasuries for investors' money.

It is worth pointing out that not everyone thinks this is a big factor. Morgan Stanley's credit strategists point out that the Treasury now borrows mostly through shorter-term debt while tech companies borrow long, and that the two tend to attract different buyers¹⁶. Researchers have also found little direct evidence so far that investors are choosing tech bonds instead of Treasuries. I think any increase of supply can create real pressure, but one among several, not the main driver on its own. An increase in supply with constant demand leads to lower bond prices, which means higher bond yields.

Washington is reacting either way. Since September, the Treasury has been buying back some of its older long-term bonds¹⁷, and Wall Street expects its next borrowing plan, due November 4, may reduce sales of 20- and 30-year bonds¹⁸. Other borrowers are adapting too: some companies are issuing shorter-term bonds and timing their sales to avoid going head-to-head with big tech deals¹⁵.

3. Foreign buyers are a smaller part of the picture

For decades, foreign governments, especially central banks in China and Japan, were among the most dependable buyers of U.S. Treasuries. They tended to buy regardless of price. That role is shrinking. China's holdings fell this summer to their lowest level since 2008¹⁹, and total foreign holdings slipped from a record $9.49 trillion in February to $9.25 trillion in July.

The bigger story is their share. Foreign investors still own close to a record dollar amount of Treasuries, but U.S. debt has grown much faster than their appetite²⁰. They owned about 34% of federal debt in 2012; today it's under 24%. The Fed flagged the same shift in June²¹: Treasuries are increasingly owned by private investors who are more price-sensitive, meaning they want a higher rate before they'll buy.

Federal debt held by foreign and international investors as a % of total federal debt, quarterly through Q1 2026. Source: U.S. Treasury via FRED (FDHBFIN, GFDEBTN).

While it is tempting to blame tariffs for the lower share of U.S. Treasuries held by foreigners, the evidence is mixed. Part of the recent drop in foreign holdings reflects bond prices falling rather than actual selling, and recent Treasury auctions haven't shown a sudden collapse in demand²². Trade tensions and geopolitics likely play a role, but this is a slow shift over a decade, not a sudden exit.

Why we prefer ultra-short and short-term bonds

This is why we prefer investing in ultra-short and short-term bonds in client portfolios, using funds like the PGIM Ultra Short Bond ETF (PULS), the PGIM AAA CLO ETF (PAAA) and the Dimensional Short-Duration Fixed Income ETF (DFSD). PULS and DFSD hold high-quality bonds that come due within a few years. PAAA holds the highest-rated (AAA) portions of pooled business loans, which pay interest that adjusts as rates change, so their prices move very little when rates rise. I want bonds to be the steady part of a portfolio, paying income and cushioning stock-market swings, not another source of risk.

Long-term bonds can swing a lot when rates move. In 2022, when the Fed raised rates from near zero, the broad US bond market fell 13%, its worst year since the index began in 1976²³. Before that, it had lost money in only four calendar years, and never more than 2.9%; that was 1994, when the Fed raised rates six times by a total of 2.5 percentage points. Losses grew with maturity²⁴: two-year Treasuries lost about 4%, while ten-year Treasuries lost about 16%.

Bloomberg U.S. Aggregate Bond Index, calendar-year total return. Dashed line: previous worst year, 1994 (-2.9%). Source: Bloomberg U.S. Aggregate Total Return Index (^BBUSATR) via YCharts; 1994 return from Bloomberg data via MUFG Americas.

Short-term bonds also adjust quickly. They come due within a year or two, and we reinvest at whatever rates are available then. So when rates rise, short-term bonds start earning more fairly soon instead of being stuck at lower rates. Six-month Treasury bills²⁵ currently pay around 4.3%.

Here's how those three funds have held up against the broad U.S. bond market, represented by Vanguard's Total Bond Market ETF (BND)²⁶. In 2022, the worst year for bonds in decades, PULS gained 1.5% and DFSD lost 5.9%, while the broad market lost 13.1%. This year through October 19, PULS is up 3.1%, PAAA 3.8% and DFSD 0.3%, while the broad market is down 1.8%.

Calendar-year total return with dividends reinvested, net of fund expenses; 2026 through Oct. 19. PAAA launched July 2023. Past performance does not guarantee future results. Source: YCharts total return data.

Two honest caveats. Short-term doesn't mean risk-free: DFSD, which holds somewhat longer bonds than the other two, fell as much as about 8% at its low point in 2022. And when rates fall, longer bonds win. In 2025 the broad market returned 7.1%, ahead of PULS (5.0%) and PAAA (5.4%). I'm comfortable giving up some of that upside for a much steadier ride.

Fund returns are total returns based on market price with dividends reinvested, net of fund expenses but not of advisory fees. Past performance does not guarantee future results. This is not a recommendation to buy or sell any security.

What we're watching

This is written as of mid-October, and a few things could change the picture. Here's what I'm keeping an eye on:

• Treasury's borrowing plan on November 4. If the Treasury cuts back on 20- and 30-year bond sales, that would confirm Washington sees the competition for long-term buyers as real.
• The deficit, month by month. Interest costs are now one of the biggest drivers of the deficit. If they keep climbing, the pressure on long-term rates likely stays.
• Inflation and the Fed. A cooling in inflation, especially energy prices, would take some pressure off rates.
• Big tech's spending plans. This quarter's earnings reports will show whether AI spending, and the borrowing behind it, keeps growing or starts to slow.
• Foreign demand. Monthly Treasury data shows whether foreign buyers keep stepping back or steady out.