Financing the AI Buildout: Opportunities and Risks Across Credit Markets

Mike Cloutier:

Hi everyone, and welcome back to a Bird’s Eye View of the Bond Market Podcast, from Merganser Capital Management, our podcast featuring insights from our investment professionals on the US investment grade bond market and major economic developments. I’m Mike Cloutier, Managing Principal, and I’ll be hosting today. We’re recording this episode on Thursday, October 15th. Today we’re discussing one of the most important themes shaping both technology and capital markets. It is the financing of AI infrastructure build out. It’s a very topical one. Today, you can’t turn on the TV or open up a newspaper without hearing something about AI. And while it’s often viewed through the lens of chips, models, and the innovation that’s going on in the space, there’s another story that’s unfolding and that’s what we’re here to talk about. It’s the infrastructure powering AI and it requires enormous amounts of capital and increasingly the credit markets are at the center of make it all possible.

Joining me today is Andy Smock, Co-CIO and Portfolio Manager at Merganser. Andy, welcome.

Andy Smock:

Thank you. It’s good to be here.

Mike Cloutier:

So Andy, why don’t you break down for the listeners what’s happening here?

Andy Smock:

Yeah, so as you said, AI adoption is definitely accelerating at a remarkable pace. It’s driving huge demand for digital infrastructure, particularly data centers, networking assets, and also power related investments. The challenge is that building all this infrastructure requires huge amounts of capital and demand is growing faster than projects can be completed. Borrowers are turning to increasingly diverse range of funding sources to finance that growth. So what we’re missing isn’t just a technology build out, it’s one of the largest financing stories in today’s fixed income markets.

Mike Cloutier:

So when investors hear about AI debt financing, I think a lot of them think about investment grade bonds from these really large technology companies that have been household names forever because that’s where a lot of this action is. But the reality on the ground is much broader than that, isn’t it?

Andy Smock:

Yeah, I can see why they think that because that’s traditionally how it’s gotten done, especially with tech M&A and other uses of capital, but it’s much broader today. Now it really spans investment grade credit, high yield bonds, levered loans, securitized products, infrastructure funding, and also private credit. We’ve seen really robust issuance across every segment. Those looking for funding are really looking for the best sources of capital and trying to find those structures that fit their needs the best. So as a result, AI exposure is really not confined to traditional technology sectors at all anymore. It’s spanning real estate, infrastructure, telecommunications, utilities, structured finance, and then also most recently CMBS and ABS.

Mike Cloutier:

So it sounds like a significant opportunity set for investors, but also a lot more complicated and many things to unpack just given things like deal sizes, structures, different liquidity profiles. How do you think about that?

Andy Smock:

Yeah, that’s true. You’re no longer just buying Meta or Microsoft. And historically, investors could just buy within these well-defined boundaries and knowing exactly what they’re getting, but they’re increasingly incorporating a range of features now that were once associated with distinct asset classes. We’re seeing structures that combine construction risk, long-term lease arrangements, amortizing cash flows, infrastructure characteristics, and guarantee structures that are increasingly complex, and that’s all within the same financing ecosystem. So in other words, it’s all becoming more interconnected. Investors need to understand a much wider range of risks and opportunities than they have in the past.

Mike Cloutier:

And the hyperscalers out there like the Meta, Googles, et cetera, they’re playing a major role in driving this?

Andy Smock:

They are. As they continue scaling their AI capabilities, they’re bringing increasingly sophisticated and unique structures to market. In some cases, they own the assets and operate them directly. That’s pretty simple. But others, they lease capacity, they enter into long-term contractual arrangements, they support specialized financing vehicles or partner with other infrastructure investors and private capital providers, and they don’t always fit neatly in traditional investment categories. So understanding that true credit of these, the real risks embedded in the structures requires looking past the legal issuer, evaluating the broader ecosystem of counterparties, contracts, assets, and cashflow relationships.

Mike Cloutier:

So as these different structures evolve and we go through time, does the traditional investment team model that different firms employ need to evolve as well? And give our listeners an idea of just how Merganser is approaching this.

Andy Smock:

Models definitely have to evolve. Teams have to work much more together. You just can’t afford to have folks looking at these deals within their specific silos. Traditionally, whatever Bloomberg Key it falls under is the team that works on it, and that just doesn’t work anymore. When similar AI-driven risk factors appeal simultaneously in investment grade credit levered finance, structured products, infrastructure, private markets, that expertise needs to flow back and forth between teams to compare and contrast risk. And the most compelling opportunities may not originate within a single asset class. Likewise, some of the risks may become only apparent when you’re looking at it across all these different structures.

Mike Cloutier:

So can you expand just a little bit more on what that collaboration amongst those different teams looks like in practice?

Andy Smock:

Yeah, so consider a large data center financing, traditionally building that big box in the desert with all the servers in it. An investment grade analyst might look at that and understand the hyperscaler’s balance sheet, the income statements, strategic priorities, et cetera, whereas a structured product person comes at it with a deep understanding of the lease terms, the real estate and collateral protections. Individually, they all have something to contribute to analyzing these deals really regardless of the structure and collectively they can develop a much more complete assessment of the risk and also the relative value of that. And so as hyperscalers continue introducing more innovative financing arrangements, that collaboration becomes increasingly important. The ability to connect insights across sectors and capital structures can create meaningful investment advantage.

Mike Cloutier:

So for active bond managers out there, does this really reinforce the importance of the collaboration amongst these teams and the relative value investing framework?

Andy Smock:

It does. I mean, relative value has always been part of particularly what we have done in looking at assessing risks and ratings and spread levels and using similar terms across sectors, but this is just a really magnified example of that. As issuance continues to expand, investors need to evaluate each opportunity across the full credit spectrum. The key question isn’t whether you like AI or not. Now that’s the question we get a lot is are you in favor of AI or not? But simply saying yes or no isn’t complete. The more important question to me is whether we’re being compensated for the risks of each of these structures. That might mean comparing an investment grade bond with a securitized structure, evaluating a public versus a private, or assessing different forms of data center exposure that may look similar on the surface but have very different underlying risk profiles.

 

And I think actually here it makes sense to chat a little bit about why they’re issuing in these different markets, what are the motivations? When public corporate market’s kind of the most traditional, you can get where we started this conversation with senior and secured debt of Meta or Amazon where they just raise the capital, it’s on their balance sheet and they build out these data centers. And that’s the most simple. It’s just as real as any other, but it’s a little bit more diversified with other business lines. It can be a secured deal, which is generally secured by lease payments. And so in that case, it’s off their corporate balance sheet, but you have a very tight company involvement. These public market bonds have been coming really cheap, so they are attractive on the surface. The reason that issuers like to do this is they can do size.

 

They can do $25 billion per deal pretty easily. So it’s a very easy market to issue into. In ABS, there are some pros and cons to this one. They’re typically backed by leases of established buildings. Some of the sponsors are REITs or pension funds, others are startups. Some have big corporate tenants and oftentimes only one tenant, so Microsoft might have a long-term lease. The downside to ABS is that they’re much smaller deals, less than five or less than 10, certainly maybe even less than 5% of all deals have been within the ABS market. So you’re talking kind of a billion-ish round numbers to get it done there. But the positive is that they’re much tighter spreads because you have much better protections and controls and line of sight. In the CMBS market, interestingly, CMBS probably has the most history with data centers because they’ve been doing these co-located deals and towers, these server farms, et cetera, for a very long time.

So there’s a little bit more comfort on the side of investors there. That’s something we continue to see and that market will continue to play a role. Lastly, I would just say that the private credit market is oftentimes these companies are coming in the private market and then they’re converting to public. So it’s increasingly complex, especially when you think about the size of these deals and whether they’re going to be included in the index. And certainly there’s a big role for active managers here, but also for passive. They care very much if something’s in the index and they’re going to participate. So ultimately, given all that flexible relative value framework is really, really important.

Mike Cloutier:

So a lot of ground covered here in a couple of minutes, and this situation is evolving really fast and investors are digesting new information pretty much every day on this topic. So as we wrap up our conversation, what do you think is the key takeaway for investors right now?

Andy Smock:

I think for me and for markets, it’s that the AI revolution is reshaping far more than just the technology side. It’s driving one of the largest infrastructure investments that we’ve seen in a very long time. It’s creating significant financing needs across the credit landscape. And at the same time, it’s introducing new structures, new interdependencies, and new sources of risk. Success here is going to depend on the ability to look across markets, understand evolving financing models, and foster deeper collaboration among investment teams. Credit markets aren’t simply supporting the AI revolution, they’re helping make it possible. And that’s through the lens of an investment grade investor here, but tremendous amount of dollars are getting raised and finance and risk is getting introduced to the system and we have to understand that risk.

Mike Cloutier:

It’s a great perspective, a fitting conclusion. Andy, thanks again for taking the time today.

Andy Smock:

Yeah, my pleasure. Thanks for having me.

 

Disclosures:

This presentation is for informational purposes only and should not be considered investment advice or a recommendation of any particular issuer, security, strategy or investment product. Opinions and estimates offered constitute our judgment and are subject to change without notice as our statements of financial market trends, which are based on current market conditions. We believe the information provided is reliable, but do not guarantee its accuracy or completeness. This podcast contains or incorporates by reference certain forward-looking statements which are based on various assumptions, some of which are beyond our control. Actual results may differ materially. Past performance is no guarantee of future results.