AI Is Now Being Built on Borrowed Money. Here's Why That Matters to Your Bond Fund.

How big the AI buildout has gotten

To build and run AI systems, companies need enormous data centers, specialized chips, and huge amounts of electricity. That costs money on a scale that's hard to picture.

UBS estimated earlier this year that combined capital spending among the major AI companies could top $770 billion in 2026. That's spending on physical things like buildings, servers, and chips, in a single year.

For a while, the biggest tech companies could pay for that spending out of their own profits. They were among the most profitable companies in history, sitting on huge piles of cash.

That's changed. According to FactSet, new debt went from covering about 9% of these companies' capital spending in 2024 to about 32% over the twelve months ending mid-2026. In other words, a much bigger share of the AI buildout is now being paid for with borrowed money.

The borrowing, by the numbers

The size of individual bond deals has been striking.

• Oracle sold $18 billion of bonds in September 2025, one of the biggest bond sales on record, and came back in February 2026 with another $25 billion.
• Meta sold $30 billion of bonds in October 2025. That was the largest investment-grade bond sale ever that wasn't tied to an acquisition.
• Alphabet raised billions more and, in February 2026, became the first technology company in decades to sell a 100-year bond.

Some of these companies have also turned to issuing new stock and to large leasing arrangements, which work a lot like debt even if they don't show up as bonds.

A Bloomberg report this month noted that demand for credit default swaps tied to some of these companies has surged. A credit default swap is basically insurance against a company not paying its debts. When more investors buy that insurance, it tells you some of them are getting nervous about the debt piling up.

Why a strong company borrowing isn't automatically bad

Before going further, it's worth saying this clearly. Borrowing isn't a bad thing on its own.

Companies borrow all the time to invest in things they expect to pay off. A company with strong profits borrowing at reasonable rates to build something valuable can be a smart move. Many of these companies still have excellent credit ratings and very large, profitable businesses.

The question isn't whether they're borrowing. It's whether the thing they're building earns enough to justify the cost. And for AI, that answer isn't in yet. A lot of the spending is a bet on future demand.

Why some investors are uneasy

Here's the concern, laid out as fairly as I can.

Spending is growing faster than cash flow

For some of these companies, capital spending is now growing faster than the cash their businesses generate. One analysis found that for Oracle, projected 2026 capital spending is about twice its operating cash flow. When spending outruns cash flow, the gap gets filled with debt or new stock.

The technology can change fast

Chips and servers can become outdated quickly. If newer technology makes today's equipment less valuable before it's paid off, companies could be left with debt on assets worth less than expected.

Investors are charging more

Lenders are already asking for a premium. Recent Goldman Sachs and ICE data cited by Reuters showed AI-related company bonds paying around 1.15 percentage points more than comparable Treasuries, versus about 0.78 points for the broader investment-grade bond market. That extra yield is investors' way of saying they see extra risk.

The payoff is uncertain

The whole buildout assumes that demand for AI will grow enough to make all this spending worthwhile. Maybe it will. Maybe it will take longer than expected. If revenue from AI grows more slowly than the spending, some of this debt could become a problem.

Why others aren't worried

To be fair to the other side:

• These are some of the strongest companies in the world. Most have very large, profitable core businesses that don't depend on AI working out.
• Many still have strong credit. Rating agencies haven't treated this as a crisis.
• The demand is real so far. Companies are building because customers are buying AI capacity, not just on speculation.
• Big buildouts often look scary in the middle. Railroads, electricity, and the internet all involved huge upfront spending before the payoff was clear.

Like I said, I'm not going to pick a side on whether this ends well. Both sides have real points.

The part that matters most for retirees: your bond fund

Here's the connection a lot of people miss.

Many retirement investors hold bond funds, or funds that include bonds, like target-date funds and balanced funds. People usually own bonds for stability, to balance out the ups and downs of stocks.

A lot of bond funds track broad bond indexes. And just like stock index funds, many of those bond indexes are weighted by size. The more debt a company issues, the bigger its share of a bond index can become.

Vanguard pointed out in August that technology has historically been a small slice of the investment-grade corporate bond market, but its share of a major corporate bond index is now rising quickly as AI-related borrowing makes up a large share of new supply. In Vanguard's words, concentration, long a stock-market issue, is migrating into the part of the market many investors hold as a diversifier.

In plain terms: the same handful of tech companies that dominate your stock funds may be growing inside your bond funds too.

That doesn't mean your bond fund is risky. Broad bond funds still hold a lot of government bonds and bonds from many other companies across many industries. But it does mean the "safe" side of your portfolio may be a little more tied to the AI story than it used to be.

What this means for your accounts

A few practical things worth looking at.

Check your bond fund's holdings

Just like with a stock fund, you can look up what's inside your bond fund. The fact sheet usually shows the mix of government bonds, corporate bonds, and other types, and often the largest issuers. If you own a corporate bond fund specifically, the tech share may be meaningful.

Know the difference between government and corporate bonds

U.S. Treasury bonds are backed by the federal government. Corporate bonds are loans to companies and carry the risk that a company can't pay. That extra risk is why corporate bonds usually pay more. In a broad bond fund, you typically own a mix of both. How much of each you own changes how your bond fund behaves when markets get stressed.

Look at your total tech exposure

If you own an S&P 500 fund, a tech-heavy growth fund, and a corporate bond fund, you may have a bigger combined bet on the same few companies than you realize. It's worth adding it all up.

Remember why you own bonds

Bonds in a retirement portfolio are usually there to provide stability and income, and to cushion stock market drops. If you're counting on your bond fund for that role, it's worth making sure it's built to play it.

What past buildouts can and can't tell us

People reach for history whenever a new technology requires enormous upfront spending. It's worth looking at, as long as we're honest about its limits.

Railroads. In the 1800s, railroad companies borrowed heavily to lay track across the country. The railroads transformed the economy. But a lot of the companies that built them went bankrupt along the way, and many investors who lent to them lost money. The technology won. Many of the early lenders didn't.

Telecom and fiber optics. Around 2000, telecom companies borrowed heavily to lay fiber-optic cable, betting on explosive internet demand. The demand eventually came, but not fast enough. Several companies collapsed under their debt in the early 2000s, and bondholders took real losses. Years later, that same fiber became the backbone of the modern internet.

What these stories suggest. Big buildouts can be right about the technology and wrong about the timing or the financing. The companies doing the building, and the people lending to them, can struggle even when the technology eventually succeeds.

What they don't prove. Today's biggest AI spenders are much larger and more profitable than most of the railroads or telecom upstarts of the past. Many have core businesses that generate enormous cash regardless of AI. History rhymes. It doesn't repeat on schedule.

Questions to ask about your bond fund

If you want to understand how much the AI borrowing wave touches your own money, here are a few questions to answer from your fund's fact sheet or your plan's website:

• What share of the fund is in U.S. Treasuries versus corporate bonds? More Treasuries generally means less exposure to any one company's debt.
• What share is in technology companies? Some fact sheets break out corporate bonds by sector.
• Who are the largest issuers? Some funds list their top holdings by issuer.
• What's the fund's average maturity or duration? Longer duration means more sensitivity to interest rate changes.
• What's the credit quality? Look for the mix of high-rated and lower-rated bonds.

You don't need to become a bond expert. Knowing the answers to those five questions puts you ahead of most investors.

Why rising interest rates add to the pressure

This is happening at the same time interest rates are rising. The Federal Reserve raised rates in September for the first time since 2023, and the 10-year Treasury yield hit its highest level since 2007.

Higher rates make borrowing more expensive for everyone, including these companies. Every new bond they sell now costs more in interest than it would have a year ago. That makes the question of whether AI pays off even more important.

Higher rates also push down the prices of existing bonds. So bond funds have already been under some pressure this year, separate from anything specific to AI.

The good news, for anyone adding money to bonds, is that new bonds are paying higher yields than they have in years. That's the trade-off with bonds when rates rise: short-term price pain, higher future income.

A note for people close to retirement

If you're within a few years of retirement, this is a good moment to look at how your money is split.

The FRS Pension Plan gives many public employees a base of guaranteed income that doesn't depend on the stock or bond market at all. That's a real advantage when markets get unsettled. If you're in the Investment Plan, or you've built up a 457 or other savings, those balances depend on how your investments perform.

For money you'll need in the next few years, a lot of people prefer places that don't swing much, like short-term Treasuries, money market funds, or CDs. Money you won't need for a long time can afford to ride out more ups and downs. The AI story, whichever way it goes, is a good reason to make sure each piece of your money has the right job.

A quick glossary

A few terms come up a lot in this story. Here's what they mean in plain English.

Capital spending (capex). Money a company spends on long-lasting physical things, like buildings, servers, and equipment, rather than day-to-day costs.

Investment-grade. Bonds from companies that rating agencies consider relatively likely to pay their debts. Most large tech companies are rated investment-grade.

Spread. The extra yield a corporate bond pays compared to a Treasury bond of similar length. A wider spread means investors want more compensation for the risk.

Credit default swap. A contract that works like insurance against a borrower not paying its debt.

Duration. A measure of how sensitive a bond or bond fund is to interest rate changes. Higher duration means bigger price swings when rates move.

Cash flow. The actual cash a business brings in from its operations. When spending runs ahead of cash flow, the difference usually comes from borrowing or selling stock.

What I'm watching

• AI revenue. Whether companies start showing meaningful income from their AI spending. That's the real test.
• Credit spreads. How much extra investors demand to lend to AI-related companies. If that gap keeps widening, it's a sign of growing concern.
• New bond deals. The pace and size of new tech borrowing, and how eagerly investors buy it.
• Interest rates. Higher rates raise the cost of all this borrowing.
• Earnings season. The biggest tech companies report quarterly results in October. What they say about spending plans matters a lot.

What I'm not going to do

I'm not going to tell you AI is a bubble, and I'm not going to tell you it's the next great boom. I don't know, and neither does anyone else. Both outcomes have happened before with big new technologies.

I'm also not going to tell you to buy or sell any particular stock, bond, or fund.

What I will say is this. When it comes down to it, the biggest risk isn't that AI succeeds or fails. It's owning more of the same bet than you realize, across your stocks and your bonds, without having chosen to.

The short version

• Big tech companies are increasingly borrowing to pay for the AI buildout, not just using their own cash.
• New debt went from covering about 9% of their capital spending in 2024 to about 32% by mid-2026.
• Some record-setting bond deals have come from AI-related companies, and investors are charging them a premium.
• Tech's share of the corporate bond market is rising, so bond funds may carry more AI exposure than before.
• Rising interest rates make this borrowing more expensive.
• Check what's in your bond funds, add up your total tech exposure, and make sure your bonds are doing the job you own them for.

If you'd like a second set of eyes on what's actually in your 457, Investment Plan, or IRA, including the bond side, I'm glad to go through it with you.

Frequently asked questions

Why are tech companies borrowing so much money for AI?

Building AI requires huge spending on data centers, chips, and power. That spending has grown faster than some companies' cash flow, so they're increasingly using bonds, new stock, and leases to fund it.

Is the AI boom a bubble?

No one knows. Supporters point to real demand and very profitable companies. Critics point to spending that's outrunning cash flow and uncertain future revenue. Big technology buildouts have ended both well and badly in the past.

How does AI borrowing affect bond funds?

Many bond funds track size-weighted indexes, so companies that issue a lot of debt can become a larger share of those indexes. Tech's share of the corporate bond market has been rising as AI-related companies borrow more.

Are corporate bonds riskier than Treasury bonds?

Generally, yes. Treasury bonds are backed by the U.S. government, while corporate bonds depend on the company's ability to pay. That's why corporate bonds usually pay higher yields.

What is a credit default swap?

A credit default swap is a contract that works like insurance against a borrower failing to pay its debt. Rising demand for them can signal growing concern about a company's debt.

Educational content only. Not individualized investment, tax, or legal advice. Company names are mentioned only as examples from public reporting and are not recommendations. Past performance does not guarantee future results. Benowitz Wealth Management is a brand of Joy Financial Group LLC, a Florida state-registered investment adviser.

Neither Joy Financial Group LLC nor Benowitz Wealth Management is affiliated with, endorsed by, or employed by the Florida Retirement System, the Florida Division of Retirement, the State Board of Administration of Florida, MyFRS, or any government agency. FRS members may obtain free, unbiased guidance from the MyFRS Financial Guidance Line at 1-866-446-9377.