Last November, we made the case that, as bond investors, we weren't totally buying the AI hype, flagging power availability and a forthcoming supply glut as potential key risks. Five months on, those concerns have only intensified. We have no desire to be professional party poopers, but unlike equity investors, our job is to focus on the downside risks, not to make bets on AI's future or Big Tech's success.
The AI infrastructure boom has continued at pace even as attention has shifted to the Gulf. Debt issued by the hyperscalers (Amazon, Microsoft, Meta, Alphabet and Oracle) now amounts to almost 4% of the US Investment Grade (IG) Corporate Bond Index. This may not seem like much, given they were less than 1% at the start of 2015, but it represents a marked increase in the quantum of their borrowing. For the record, we're not expecting to reach the 20% levels of concentration seen in the S&P 500.
Figure 1. Hyperscalers' share of IG versus the S&P 500
Source: Morgan Stanley calculations, as of 1 April 2026.
Problems loading this infographic? - Please click here
Wider spreads still not compensation enough
But the market is already feeling the weight of that supply. Since September, spreads on A-rated and above hyperscalers have widened by between 14 and 29 basis points (bps), according to ICE Data Indices, compared to just 4 bps for the broader market. In our view, this is driven by the glut of issuance and growing concerns around buildout delays fuelled by growing labour shortages, increased public backlash and regulatory intervention.
Despite that widening, those spreads still don't look compelling to us. These tend to be the safer and steadier AI names in the IG space, but spreads are not yet at levels that adequately compensate for the infrastructure risks ahead.
AI’s energy problem
The latest International Energy Agency (IEA) report on energy and AI makes fascinating reading. Around half of all data centre electricity still comes from fossil fuels, according to IEA projections. With the Strait of Hormuz still not fully open, that raises the cost of powering the AI buildout at precisely the moment demand is surging.
The IEA expects the mix to shift toward clean energy over the coming decade, but that transition might be years away. In the meantime, capital expenditure by the five largest technology companies surged to more than US$400 billion in 2025 and is set to increase by a further 75% in 2026. Electricity demand from data centres soared 17% last year, far outpacing 3% growth in global electricity demand, and the IEA expects it to double by 2030.
Coupled with physical bottlenecks for chips and other IT components, this may lead to further pressure on AI firms. As IG bonds form approximately 80% of AI-related financing according to calculations from a Morgan Stanley report, published on 10 April, we see most of this risk developing in the high yield and private credit arenas.
AIn’t no stopping us now: AI issuance is not just a US phenomenon
The debt issuance party isn't just limited to the US. In February, Alphabet went from having no sterling-denominated debt to being a top five issuer in the GBP IG market. Its unique 100-year bond generated significant interest from pension funds.
And there are upsides (we’re not total AI grinches) for fixed income investors. Despite hundreds of billions of dollars of newly issued debt, leverage remains low compared to other sectors. This is a consequence of the very strong profitability and cashflow generation of most of these firms. In fact, hyperscaler borrowing is only a fraction of what has been seen in most other non-financial sectors.
Figure 2. Leverage in the sector is still lower than other sectors
Source: Morgan Stanley, as at 31 December 2025.
Problems loading this infographic? - Please click here
What remains important is that investors recognise that not all hyperscalers are created equal, just as not all software firms are likely to be disrupted to the same extent. Those with diversified cashflows and revenue streams have significantly more flexibility to weather any storms should the AI revolution take longer than expected.
All data sources Bloomberg unless otherwise indicated.
Author: Jon Lahraoui, Director, Client Portfolio Manager, Discretionary Credit at Man Group.
You are now leaving Man Group’s website
You are leaving Man Group’s website and entering a third-party website that is not controlled, maintained, or monitored by Man Group. Man Group is not responsible for the content or availability of the third-party website. By leaving Man Group’s website, you will be subject to the third-party website’s terms, policies and/or notices, including those related to privacy and security, as applicable.