The bond giant buys bonds in preparation for the bursting of the AI bubble.

Credit bond giants DoubleLine Capital and Oaktree Capital are already positioning themselves for the possible burst of the AI bubble - buying bonds that can survive in a deep credit cycle. "What are the chances of an AI bubble? I would say about 100%," DoubleLine Capital fund manager Robert Cohen said at the Bloomberg Global Credit Forum on Wednesday. The influx of massive AI debt The scale of borrowing triggered by AI infrastructure construction is rapidly expanding. According to a Barclays report on May 21, US mega tech companies have issued over $155 billion in unsecured bonds globally this year, a 45% increase from the total issued last year. Bloomberg Intelligence predicts that over the next five years, companies will spend around $5 trillion on AI, with a large portion coming from debt financing. Financing outside of the super tech companies is also hot. Just this week, data center company Hut 8 issued about $4 billion in investment-grade bonds for financing projects in Texas, and a $36 billion bond issuance for AI model developer Anthropic's chip procurement is also close to completion. "What kind of credit bonds can survive a deep cycle" Under the deluge of debt, no fund manager can ignore the situation. But Cohen believes that while bond prices and valuations are not overheated at the moment, with technology companies continuing to pour massive amounts of money into AI, it will undoubtedly reach bubble levels in the coming months or years. The core problem of investing in AI-related debt is that many bonds being sold won't mature for decades, by which time the current technology may be outdated. Data centers, in particular, face the risk of overbuilding - construction cycles are lengthy, and many projects are being launched simultaneously. "You have to think about what kind of credit bonds can survive a deep cycle," Cohen said, "What you need are assets that can survive through structural arrangements or strong balance sheets." Cohen defines a credit bubble as investors providing financing to companies that need actual growth to pay off their debts. Historically, technological booms often lead to bubbles in this way. Buy, but be selective Oaktree Capital's strategy is similar - even though it is not yet known when the bubble will arrive, the company has already positioned itself for potential speculation. Christina Lee, co-portfolio manager of private credit at Oaktree Capital, said that data center financing is a huge and growing opportunity, "but we have to be selective because it is still unclear who will come out on top in this competition." Oaktree Capital co-founder Howard Marks warned in a December 2020 memo: "No one should bet the farm without acknowledging the risk of destruction. However, no one should stay completely on the sidelines and miss out on a great leap in technology." PIMCO Group's Chief Investment Officer Dan Ivascyn is more cautious. He believes that AI is not an overweight sector, but due to the massive financing demand, "you can maintain a defensive stance on overall exposure while unlocking significant value." He also warned that in the event of a default, the loss could exceed investors' past experiences. With a large amount of debt continuing to pour in, PIMCO hopes to capture truly attractive opportunities through its own size. Ray Dalio, founder of Bridgewater Associates, pointed out this week on Bloomberg TV that large-scale technological revolutions have always been accompanied by speculation. The dilemma facing companies is: "Either spend a lot of money to grab market share, regardless of whether you are spending too much or too little, hand over market share."