AI bond sales lose pace as investors push back on Big Tech debt, FT says

What’s going on here?

According to the Financial Times (FT), the pace of AI-related borrowing has eased as investors grow wary of the debt piling up behind the data center boom, testing demand after months of record bond issuance. Other reporting backs up the shift. Orders for hyperscaler bond deals covered about five times the amount on offer in February but slipped to around two times by July, Startup Fortune noted. Spreads, the extra yield over Treasuries, on AI-related issuers sat around 115 basis points, against 78 for the wider investment-grade market, Reuters reported in September, citing Goldman Sachs and ICE BofA data.

What does this mean?

Hyperscalers are the giant cloud companies, such as Amazon, Alphabet, Meta, Microsoft and Oracle, that are building the data centers AI runs on. They have increasingly paid for that buildout with bonds rather than cash alone. Morgan Stanley estimates AI-related global debt issuance could total roughly $570 billion this year, over twice last year's rate, Startup Fortune reported. Tech now makes up roughly 18% of US investment-grade bond supply, a record share.

Bond buyers are not predicting defaults. Portfolio managers told Reuters that the issue is volume and unpredictability: when the same borrowers keep coming back for more, investors demand a bigger discount and start hitting limits on how much of one company they can hold. Some deals have needed large concessions. Alphabet had to offer one to complete its August bond sale, BNY said, while an Aon acquisition financing in September drew $65 billion of orders for $13.5 billion of bonds. In short, money is still available, but buyers are choosing other borrowers first.

Oracle shows where the strain is sharpest. Its five-year credit default swap, a kind of insurance against default, climbed to around 203 basis points, its highest in 18 years, and S&P Global Ratings cut the company to BBB-, one step above junk, citing its AI spending pace. Some financing has also moved into separate vehicles funded by outside investors, which keeps it off the tech giants' own balance sheets but adds leverage to the system.

What is new is that the warning is coming from lenders rather than stock investors. Credit markets tend to be more focused on cash flow and repayment, and they are signaling that the cost of financing AI is rising.

The bull case

The biggest borrowers, such as Microsoft, Alphabet and Meta, earn huge amounts of cash and carry high credit ratings. Wider spreads largely reflect supply and demand, BlackRock's Russell Brownback told Reuters, not worries about credit quality. A slower pace of borrowing could help the market digest past deals, and higher yields give bond investors better pay for lending to strong companies.

The bear case

Free cash flow across the largest AI spenders is expected to approach zero or turn negative this year as capital spending surges, Startup Fortune reported. If investors keep demanding more, every new data center costs more to finance. Weaker borrowers, such as Oracle and AI firms funded with high-yield debt, are most exposed, and tighter credit could eventually slow the spending that has lifted AI stocks.