Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Big Tech companies are on track to dominate borrowing in the US bond market, in a shift that could expose some of the world’s safest securities to greater risk from artificial intelligence.
By 2030, half of the 10 largest borrowers in the US investment-grade corporate bond market will be so-called hyperscalers — companies such as Alphabet, Amazon, Meta, Microsoft and Oracle that are building colossal data centres — according to Apollo Global Management.
Until now, the major borrowers in the league table have mostly been big banks and telecom companies, meaning that credit investors are largely insulated from shocks in the tech-dominated stock market.
Investors have grown increasingly concerned that the gap between runaway capital expenditure on AI and the returns it generates could amount to a bubble that ultimately hits both equities and the credit market.
Market participants also worry that AI exposure is greater than it might at first appear because the investment boom has also boosted demand in associated sectors such as utilities and industrials.
“What appears diversified across issuers and sectors increasingly represents a single macro trade on AI,” Apollo analysts wrote in the firm’s 2026 credit outlook report.
Morgan Stanley estimates that hyperscalers and their adjacent companies will raise $400bn from the US high-grade market in 2026, a dramatic increase from $170bn last year and just $44bn in 2024.
By contrast, banks are likely to borrow less because of regulatory changes that allow them to hold less capital on their balance sheets.
JP Morgan says the AI and data centre-related sector now makes up 14.5 per cent of the JPMorgan US Liquid Index — a benchmark for the nearly $10tn US investment-grade bond market — bigger than the share accounted for by banks.
While the bank has only tracked the sector since last year, it is growing quickly and JPMorgan forecasts it could account for more than a fifth of the index by 2030.
Lauren Wagandt, a portfolio manager at T Rowe Price, added that hyperscalers’ rapid AI expansion could increase volatility in the previously sleepy high-grade bond market.
“It’s probably a bad thing if we’re more correlated to equity and becoming less of a diversifier than in the past,” she said.
The surge in AI-related bond issuance has already pushed up borrowing costs for the most indebted companies.
Oracle’s credit spread — the premium investors demand to hold its bonds relative to US Treasuries — jumped more than 0.75 percentage points after it borrowed $18bn from the bond market in September, according to S&P Global data.
“If [hyperscalers] have to borrow $10bn every quarter for the rest of the year, how is the market going to react?” said Dominique Toublan, head of US credit strategy at Barclays, which has an underweight stance on the US tech sector.
But Nathaniel Rosenbaum, strategist on the US high-grade credit team at JPMorgan, said that the hyperscalers’ strong credit ratings made the surge in issuance “a net positive for ratings in the investment-grade universe”.
Cash-rich companies such as Alphabet and Meta still have plenty of room to increase their borrowing without harming their credit ratings, added John Lloyd, global head of multi-sector credit at Janus Henderson Investors.
“If AI blows up, it will be bad for their equity, but their credit would likely still be very solid,” Lloyd said.