Wall Street is helping finance a multitrillion-dollar AI buildout that is reshaping more than just the technology sector.
getty
Wall Street is finding out just how deep its pockets really are.
Goldman Sachs estimates hyperscalers such as Microsoft, Amazon, Alphabet and Meta will spend about $5.3 trillion on AI infrastructure through 2030. Barclays analysts expect those companies to issue more than $200 billion in debt this year alone, with borrowing expected to climb further in 2027.
No one disputes the spending. The debate is what happens next.
Apollo Chief Economist Torsten Slok argued this week that borrowing on that scale is already crowding out demand for U.S. Treasurys and other fixed-income investments.
Todd Czachor, global head of fixed income research at Columbia Threadneedle, said he isn’t seeing evidence of that in the corporate bond market. “At least not yet.”
“Appetite for corporate bonds in this yield environment has been strong with non-AI related issuance being well received,” he says.
The data backs him up. The ICE BofA U.S. Corporate Index Option-Adjusted Spread, which measures how much more investors demand to lend to companies than to the U.S. government, finished June at about 0.76 percentage points, essentially unchanged from the start of the year. If investors were becoming more nervous about lending to companies, that gap would widen.
Instead, Czachor said, recent moves in corporate bond spreads have been driven more by supply chain disruptions tied to the conflict with Iran than by a flood of AI debt issuance.
That doesn’t necessarily mean Slok’s concern is misplaced. The yield on five-year Treasury Inflation-Protected Securities, a measure of the inflation-adjusted return investors earn on Treasury debt, has climbed about 0.4 percentage points since the start of the year. But Treasury yields move for many reasons, including changes in inflation expectations, Federal Reserve policy and the outlook for economic growth, making it difficult to isolate AI’s impact. Further complicating the analysis, real yields began rising more sharply around the time Kevin Warsh was confirmed as Federal Reserve chair, after he struck a more hawkish tone than some investors had expected.
Brij Khurana, a fixed-income portfolio manager at Wellington Management, also doesn’t believe AI borrowing is crowding out other parts of the credit market today.
Khurana agrees AI’s financing needs are reshaping markets. He just thinks investors may be looking in the wrong place for evidence. Rather than showing up first in corporate bond spreads or Treasury yields, he believes the effects are becoming more visible in the stock market.
The Invesco KBW Bank ETF ($5.3 billion in assets) has returned 13.7% this year, outperforming the S&P 500’s 10.2% gain. The Roundhill Magnificent Seven ETF ($3.6 billion in assets), which tracks seven of the largest U.S. technology companies, is flat.
Khurana said investors have long viewed companies such as Microsoft, Amazon, Alphabet and Meta as unusually safe because they generated enormous amounts of cash and carried relatively little debt. As those companies borrow more to finance AI investments, he believes investors may be placing less faith in that financial strength than they once did. Combined debt at the four companies climbed 60% over the 12 months ended in March, rising from $333 billion to $533 billion.
He also said many investors expected private credit firms to finance much of the AI buildout. Instead, banks and public debt markets have played a larger role than many anticipated, creating more opportunities for banks to earn underwriting fees and interest income from the borrowing boom. Not everyone thinks the rally has further to run. This week, analysts at investment bank Oppenheimer downgraded several large banks, including Goldman Sachs and Morgan Stanley, and said they now prefer alternative asset managers, while maintaining a favorable outlook for banks overall.
It’s still too early to say whether AI borrowing will make it harder for governments and other companies to raise money. So far, there’s little evidence that it has. But that doesn’t mean AI isn’t changing markets. Investors just have to look beyond the bond market to see it.
Leave a comment