Some of the world’s largest companies are issuing huge amounts of debt, where are the opportunities and risks?
Each month at WP we take a deep dive into a topic highly relevant to Canadian financial advisors. This April we’re focusing on fixed income.
AI infrastructure spending estimates grow by the year, month, week, and day. In Q4 of 2025 Amazon spent $40.5 billion (USD), Microsoft spent $36.2 billion, Alphabet spent $28.5 billion, and Meta spent $22.5 billion. Sell side estimates put the total capital expenditure (capex) on AI infrastructure at somewhere between $2.9 trillion and $5 trillion through 2028. That number could keep growing as the insatiable appetite for compute power and data centres continues to grip the tech sector.
Many of the so-called ‘hyperscalers’ making these huge investments in AI infrastructure are doing so with debt issuances. Even companies like Amazon and Alphabet, with huge cash flows and deep pockets, are using debt to finance this buildout. All that new debt has had a significant impact on credit markets, both public and private, leaving advisors with the task of assessing how this macro theme that has been such a driver for equity markets might play a role in fixed income allocations.
“From 2025 to early this year, we started seeing much more debt issuances by all types of AI companies as well as the AI adjacent space, the so called picks and shovels of AI. Everyone seems to have increased their budgeted spend and that results in increase in debt issuances as well,” says Queenie Mak, credit research analyst at CIBC Asset Management. “Three years ago, no one was really talking about the securitized debt market for AI because it was non-existent. But over the past three years this has actually grown quite a bit. I would say that for AI in general right now the build out is so large and the debt issuance requirements are so large that there is a pretty much a suitable debt investment for every risk appetite out there.”
While one of the hallmarks of the AI theme in equity markets has been a tendency for investors to chase hype and narrative, as perhaps best demonstrated by the Allbirds stock spike, Mak argues that credit investors tend to be more thorough in the diligence around AI-related debt. She notes that in spread levels, we can see investors placing a premium on relatively new debt issuers in this space, understanding that there may be some new risks here borne out of these companies’ relative inexperience issuing and servicing large-scale debt.
While credit markets had placed some premiums on AI-related debt issuance, Mak sees a broad and often attractive market for securitized credit capturing the AI buildout. She sees opportunities across investment grade, high yield, and even private credit solutions, noting that US dollar denominated securities would offer the easiest means of access. Canadian dollar denominated credit is also available, she says, but the shelf is more limited.
For investors and advisors who prefer to access credit through index strategies, Mak notes that while AI-related credit is growing, it doesn’t yet represent the per centage of credit indices that AI-related equity names represent on major equity indices.
Jason Lemire, Chief Investment Officer and Portfolio Manager at Bold Wealth, sees some risk in the growth of AI-related credit on major corporate debt indices.
“[AI hyperscalers] are going to start amassing a crazy amount of debt to finance all of these capital expenditures and then they end up at the top of the credit indices as well,” Lemire says. “So, basically people have bonds to diversify away from equity but in the end if it’s the same issuers, if it’s starting to not go well on the equity side, that’s going to have an impact on the credit side and vice versa.”
Lemire notes that he doesn’t see this as a short-term risk for credit investors. AI debt will have to grow to comprise a far larger proportion of credit index investments in order to result in that unintentional dual exposure. Nevertheless, he notes that the current growth of AI capex debt issuances could make concentration a real risk in the longer-term.
In assessing the long-term prospects for the AI theme in credit markets, both Mak and Lemire highlight the test of monetization. If these companies can begin to show they’re monetizing their data centres and other AI infrastructure, especially at a faster rate than the depreciation of their GPUs, then that could be a sign that this debt is sustainable, and the theme has room to run. Lemire says that he has seen and is continuing to expect positive news on monetization from the hyperscalers. Mak adds that she and her team are broadly bullish on this side of the credit market.
In assessing and communicating opportunity in credit markets now, Lemire believes that a focus on active management can help advisors differentiate themselves, finding ways to maximize returns on credit allocations while offering the flexibility to avoid any emerging issues like concentration risk.
“If I look at the data over the long term, I think that passive is quite hard to go against on the equity side. But when it comes to the fixed income side, I’ve managed those index bond portfolios, the methodologies are horrible,” Lemire says. “Basically, the more debt you have, the higher weight you’re going to have in the index. When it comes to fixed income, passive is a pretty terrible way to structure a bond portfolio. I think that’s the angle that’s extremely important for financial advisors, segment the debate and then encourage, or at least look seriously at, going active for at least part if not the entire fixed income component for their clients.”
