Why Bonds Deserve Another Look in an AI Driven Market
By Paul Rubillo ·
The S&P 500 can give an investor the appearance of broad diversification while a surprisingly small group of companies drives much of the result.
As of September 4, 2026, the ten largest companies represented 37.8% of the index, according to S&P Dow Jones Indices.
The concentration extends further than the weight of those ten stocks.
Artificial intelligence spending has become an important source of demand for utilities, industrial equipment, data center construction, financing and other businesses that sit outside the technology sector.
Several sectors can rise together while still depending on the same capital spending cycle.
However, this could create some issues with your portfolio from a diversification standpoint.
You may own several funds and dozens of stocks, yet have far more exposure to one economic theme than the number of holdings would lead you to believe.
The Economy Has Become More Dependent on AI Spending
The construction of data centers is producing real economic activity.
Companies are buying equipment, expanding power capacity and committing large sums to computing infrastructure. Those projects do absolutely support revenue, employment and business investment today.
However, we do need to point out that the amount of growth coming from this area also exposes the economy to a slowdown in that spending.
I saw an estimate from economist David Rosenberg recently that AI data center construction is producing about half of current economic growth...Woah!
In his analysis, much of the remaining support comes from consumer spending tied to the wealth effect of higher stock prices, particularly among higher-income households.
He also estimates that technology-related business investment is growing about 16% in real terms, while the rest of business investment is roughly flat from a year ago.
To be fair, those are estimates rather than official divisions of GDP, but they do identify the question investors need to answer in my opinion:
How much economic growth would remain if AI capital spending slowed and rising stock prices stopped encouraging consumers to spend from accumulated wealth?
Now, let's take a look about the financing behind all of this.
Some of the latest AI investment is moving beyond cash already held on corporate balance sheets and into the bond market.
Heavy corporate issuance competes with Treasury borrowing for investor capital.
That places upward pressure on real interest rates, which then reaches other parts of the economy through higher mortgage rates and more expensive business financing.
The housing market is particularly sensitive to this type of pressure.
Residential investment is probably a smaller share of GDP than consumer spending, but home construction, transactions and financing affect employment and purchases throughout the economy.
A capital boom in one area will draw resources away from other areas that depend more heavily on affordable credit.
The technology will undoubtedly be valuable, but that doesn't automatically mean great investment returns.
Rosenberg recently estimated that AI-related revenue may need to grow roughly 50% annually for five years to justify the amount of capital now being committed.
The current valuations leave very little room for ordinary execution, slower adoption or lower returns on all that spending.
When I was trading from 1996 through 2007, I sometimes took a position much larger than I normally would.
My judgment became clouded sooner, and the results were mediocre at best. The same issue can develop across an entire portfolio when one investment theme becomes too large.
Rosenberg recently put the concern plainly: "Diversification has become a dirty 15-letter word."
Why Bonds Offer a Useful Counterweight
Bonds have spent several years giving investors reasons to avoid them.
Prices fell as yields rose, and investors who believed bonds would always protect them from stock losses had a painful experience in 2022.
But it's important to point out that the starting yields available today are VERY different.
On September 8, 2026, the 10-year Treasury's nominal yield was 4.80%. The 10-year real Treasury yield, based on the Treasury's TIPS curve, was 2.43%.
Note: That real yield represents the return above measured inflation implied by the market if the security is held within the assumptions of the TIPS structure.
Those yields certainly improve the case for holding bonds even if rates stay elevated.
But on the bright side, they also create the possibility of price appreciation if economic growth weakens and yields fall.
It's also worth pointing out that the recent bond selloff has been influenced heavily by a change in Federal Reserve expectations.
Earlier in the year, markets anticipated rate cuts (and at the very beginning of the year, the market expected 3 rate cuts!).
By late summer, investors were considering the possibility of rate increases.
The 10-year Treasury yield moved from below 4% in February to 4.80% on September 8.
So keep in mind, that same shift can work in reverse.
If inflation moderates or the economy weakens enough to remove the prospect of rate increases, longer-term yields could give back part of that move.
A 50-basis-point decline would lower a 4.80% yield to 4.30%.
Duration provides a rough estimate of the price effect.
A bond or fund with a duration of seven years could gain approximately 3.5% if the yields affecting its holdings fell 0.50 percentage points in parallel.
That estimate comes before interest income and ignores convexity.
A comparable rise in yields would produce an approximate price decline of the same size.
The Treasury's November 4 quarterly refunding announcement is another event to watch.
Treasury can change the amount of borrowing assigned to bills, notes and longer-term bonds. A decision to rely more heavily on short-term bills could reduce some supply pressure at the long end.
Treasury has only confirmed the announcement date, so any change in issuance remains a forecast.
Match the Bond to the Job
Fortunately, an investor does not need to make an all-or-nothing decision between stocks and bonds.
The question that we do need to be asking ourselves is how each part of the portfolio will behave when you need the money.
Short Treasury bills can provide stability and a known maturity date, but the income must be reinvested at whatever rate is available when they mature.
Intermediate and long-term Treasuries lock in yields for longer and carry more duration risk.
They can produce stronger gains if rates fall and larger price declines if rates rise.
Time horizon changes the answer. A younger investor making regular contributions may have the ability to wait through a large equity decline. Someone drawing retirement income may be forced to sell assets while prices are down. Bonds matched to expected spending can reduce that pressure, even when they do not produce the highest long-term return.
Bond Compass Reading
As of September 9, 2026, the Bond Compass scored the fixed income market at 63.8 out of 100, placing the reading in the Medium Duration range.
The strongest support came from the 2.4% real yield on 10-year TIPS, tight investment-grade credit spreads and relatively subdued bond-market volatility.
The reading favors CD ladders and Treasuries in the three to seven year range.
Investment-grade corporate bonds around five years also look attractive, while municipal bonds remain competitive on an after-tax basis for investors in higher brackets.
I would treat the reading as support for adding duration gradually, with an emphasis on intermediate maturities instead of making a large bet on an immediate decline in yields.