A number of leading credit investors, including DoubleLine Capital and Oaktree Capital Management, are taking a cautious stance toward the rapid expansion of AI–linked borrowing, according to a report by Bloomberg.
The move comes as concerns grow that today’s infrastructure buildout could evolve into a future credit correction. Acco
The firms are selectively deploying capital into debt instruments they believe could remain resilient even if the AI investment cycle ultimately turns down.
At a recent industry forum, DoubleLine portfolio managers noted that while credit spreads are not yet stretched, valuations could tighten meaningfully as AI-related issuance accelerates. The firm warned that the long-dated nature of much of this debt creates additional uncertainty, particularly given the rapid pace of technological change.
The broader concern among investors is that AI infrastructure—especially data centres—faces the risk of overcapacity, as multiple large-scale projects are being developed simultaneously across markets.
According to DoubleLine, credit selection should prioritise balance sheet strength and structural protections, particularly in scenarios where borrowers may need to withstand prolonged downturns in cash flows.
On the supply side, large technology firms, often referred to as hyperscalers, have already issued more than $155bn in unsecured bonds globally, marking a sharp increase on last year, according to research cited by Barclays.
The issuance wave is being reinforced by a growing pipeline of AI-linked financing. Recent transactions include multi-billion-dollar bond deals tied to data centre expansion and chip procurement, alongside large-scale financing structures linked to leading AI developers.
Market estimates from Bloomberg Economics suggest AI-related capital expenditure could reach around $5tn over the next five years, much of it expected to be debt-funded, reinforcing expectations of sustained primary market activity.
Within private credit, managers are acknowledging both opportunity and risk. Oaktree Capital Management has described the current phase of data centre financing as early-stage, emphasising selectivity given uncertainty over long-term winners in the sector.
The firm’s investment approach assumes the potential for speculative excess, even if the timing and scale remain uncertain. Separately, long-standing commentary from Oaktree leadership has warned that investors should be mindful of downside scenarios while still participating in structurally driven opportunities.
Other large asset managers are similarly positioning defensively. Pacific Investment Management Company (PIMCO) has highlighted the difficulty of forecasting profitability in AI-related sectors, while noting that elevated issuance could still create attractive entry points for disciplined investors.
Industry voices have drawn parallels with previous technology cycles, where rapid innovation has often been accompanied by periods of credit excess and misallocation of capital.
More broadly, investors such as Bridgewater Associates have pointed to the historical tendency for major technological shifts to generate both outsized investment and speculative imbalances.
At the centre of the current cycle are mega-cap technology groups and infrastructure developers. Recent activity includes large-scale financing packages involving firms such as Apollo Global Management and Blackstone Inc., supporting AI infrastructure expansion including data centre buildouts.
Other notable transactions in the ecosystem include debt financing linked to AI infrastructure developers such as CoreWeave and crypto-linked data centre operator Hut 8 Corp., reflecting the breadth of capital formation across the sector.
Despite strong demand for AI-related credit, several investors warn that rising leverage, long-dated asset risk, and uncertain end-market profitability could increase downside volatility if expectations for AI monetisation fail to materialise at scale.
