Pimco Chief Warns AI Financing Could Reach $10 Trillion and Eventually Be Too Much

Roman was speaking at the Global Financial Leaders' Investment Summit, hosted by the Hong Kong Monetary Authority. Bloomberg, as cited by Al Bayan, reported that he saw the AI sector needing about $2 trillion of financing a year over the period to pay for capital spending. He said the volume would eventually become excessive and that the sector would face a downturn at some point that could cause some investments to fail.

"At some point in time, there's going to be a downturn, and then things are going to break," Roman said, according to Semafor and CNBC TV18. Possible triggers he named included rising US interest rates, a wider war in the Middle East and a crisis in the euro zone.

Reports differ on how Roman framed the $10 trillion figure. Bloomberg described it as AI-related financing. CNBC reported it as an estimate of AI-related capital spending, which Roman attributed to General Atlantic chief executive Bill Ford. CNBC also reported that Roman said about $550 billion of AI-related debt had been issued in various forms so far this year. He said the total required would be "gigantic" even with a mix of debt and equity, compared with a global corporate bond market of roughly $20 trillion. "At some point in time, it's going to be too much," he said.

Wall Street Leaders Divided

Other speakers at the summit were less cautious. Blackstone president Jon Gray said demand for computing power was running well ahead of supply and that "some of this scepticism today is probably healthy", while arguing that more of the build-out should be financed with equity. Goldman Sachs chief executive David Solomon said demand for compute still exceeds supply, according to The Next Web, and Morgan Stanley chief executive Ted Pick estimated that building out AI could require $10 trillion to $15 trillion over the next decade.

A Goldman Sachs report published on Monday estimated that global data centre demand will require about $8.5 trillion of capital spending over the next five years, CNBC said.

Investors Buying Protection

Roman's comments come as investors grow more uneasy about how the AI build-out is being paid for. Semafor reported that demand has risen for options and credit-default swaps that protect against a fall in what it called the biggest market bet in history. Alphabet, Meta and Oracle have all issued new bonds or are reported to be planning to, at a time when interest rates are rising.

Semafor also cited a Wall Street Journal report that institutional investors have urged Anthropic to lower its target valuation for its planned listing. The AI company expects to spend nearly $900 billion on AI infrastructure in the coming years, according to the report.

The summit followed a sharp fall in technology shares in October after Alphabet's quarterly results raised questions about whether heavy AI spending will pay off, The Next Web noted. European technology stocks led markets lower again on Tuesday.

For Gulf investors, many of whom are funding AI infrastructure directly through sovereign funds and joint ventures, the debate matters because the scale of capital involved now rivals the size of major bond markets. Roman's view is that the bigger danger is taking on too much risk. Gray's is that being too cautious about the coming productivity gains would be a mistake.