Betting on boom – and bust

ACCORDING to a Bloomberg report, some of the biggest names in global credit investing are quietly positioning themselves for a very specific outcome in the artificial intelligence (AI) boom: not whether it succeeds, but what happens when the financing cycle eventually gets messy.

Firms such as DoubleLine Capital LP and Oaktree Capital Management are already buying debt that is designed to hold up even if today’s AI-fuelled credit expansion turns into tomorrow’s stress test.

The thinking, according to Bloomberg, is that while bond valuations are not yet stretched into obvious bubble territory, that may not last long.

As technology companies continue pouring trillions into AI infrastructure, credit markets are likely to drift into more expensive and riskier territory over time.

At the centre of that debate is Robert Cohen, portfolio manager at DoubleLine.

Speaking at the Bloomberg Global Credit Forum, he described a market that is still relatively orderly but heading towards a phase where discipline will be tested.

“You have to think about what credit will survive a deep cycle,” Cohen said, speaking at the forum on Wednesday.

“You want credits that either through structure or just a very strong balance sheet will survive.”

His point, as Bloomberg reports, is that AI-linked debt is structurally complicated.

Many of the securities being issued today will not mature for decades, meaning investors are effectively underwriting companies and assets that may be tied to technologies which look very different – or even obsolete – by the time repayment comes due.

That long horizon collides with another feature of the AI buildout: physical infrastructure that is both capital intensive and prone to cyclical misjudgement.

Data centres, in particular, are emerging as one of the most exposed parts of the ecosystem.

They take years to plan and construct, and multiple projects are now being launched simultaneously across different regions.

That raises the risk, as Cohen pointed out, of overbuilding.

If too much capacity comes online at once, the economics of those assets could deteriorate quickly, leaving debt holders exposed in a downturn.

Still, stepping away from the sector entirely is not really an option for large credit managers.

The scale of issuance is already too big to ignore.

According to a May 21 report from Barclays, so-called hyperscalers – the large US technology firms building out cloud and AI infrastructure – have issued more than US$155bil in unsecured bonds globally.

That is already more than 45% higher than their total issuance for the whole of last year.

And that figure does not capture the broader wave of AI-related borrowing now moving through the system.

This week alone, Hut 8 Corp sold around US$4bil of high-grade bonds to help finance a Texas data centre project.

At the same time, a much larger transaction – a US$36bil bond deal tied to chip purchases for Anthropic – is reported to be moving closer to completion.

According to Bloomberg Intelligence, the funding demands behind all this are only just beginning to scale.

Companies are expected to spend roughly US$5 trillion on AI-related capital expenditure (capex) over the next five years, much of which is likely to be financed through debt markets rather than internal cash flow.

That backdrop is shaping how institutional investors are approaching the sector.

At Oaktree Capital Management, the strategy is to assume that speculative excess could emerge, without trying to predict exactly when or how it will show up.

Christina Lee, co-portfolio manager in private credit, said the data centre financing market is still in its early stages.

“We need to be selective because we really don’t know yet who the winners and losers of this competitive set will be.

“Data centre financing is a really large and growing opportunity set. We’re just in the early innings of it,” Lee said.

That cautionary tone is echoed in earlier remarks from Oaktree co-founder Howard Marks.

In a December note, he warned investors against both extremes – going all in, or staying completely out – given the uncertainty around long-term outcomes in fast-moving technological cycles.

“I’d advise that no one should go all-in without acknowledging that they face the risk of ruin if things go badly,” Marks wrote.

“But by the same token, no one should stay all-out and risk missing out on one of the great technological steps forward.”

The balance between risk and opportunity is also shaping views at other large asset managers.

At Pacific Investment Management Co, group chief investment officer Dan Ivascyn has warned that losses in some parts of AI-linked credit could end up being larger than many investors are currently used to.

But he also sees opportunity emerging from the scale of issuance itself.

“It’s not a sector where we want to be overweight just given the uncertainty, the volatility, the need to predict how companies are going to make money in this space,” Ivascyn said in a video in late May.

“But because of the massive funding needs, you can be defensive in terms of overall exposure and unlock tremendous value.”

That approach – cautious exposure combined with selective positioning – is becoming increasingly common as credit investors try to balance participation in the AI boom with protection against its potential downturn.

Outside traditional fixed income circles, the same concerns are being echoed in broader macro discussions.

Bridgewater Associates founder Ray Dalio said in an interview on Bloomberg Television this week that major technological shifts tend to bring periods of excess alongside genuine progress.

“Nobody can get it exactly right.

“You have to either spend a ton of money to capture your market share and don’t worry about whether it’s too much or not, or you don’t spend enough money and you lose your market share.”

For Cohen at DoubleLine, the risk is not theoretical.

He defines a credit bubble as a situation where lenders end up financing companies that rely on continuous growth just to meet their debt obligations.

Historically, he said, technological booms tend to produce exactly that kind of dynamic.

When asked about where the current cycle might be heading, he was blunt in his assessment.

“What’s the probability that we will be in an AI bubble? I’ll put maybe 100% on that,” Cohen said.