Newsfilter article unlocker

The private credit asset class is an institutional financing channel that's here to stay. So today we're going to go talk to Amanda Lynham. She is the head of credit research at Goldman Sachs, has been doing so much work on the AI trade when it comes to public market credit and private market credit. And so we're going to talk about all of that and what it means for your portfolios. Welcome to the latest episode of The Bridge by iCapital. I'm Shinali Basik, the Chief Investment Strategist at iCapitol. And today we have a very special topic, one of the hottest in the markets right now. We're going to bring in Amanda Lynham, who is the chief credit strategist at Goldman Sachs, who has been all over the topic of AI financing, what's happening in credit markets, what it means for private credit. I think we really have to start with AI because a lot of people are looking at what's happened in the market, looking at these companies with a traditionally very large free cash flow, now tap debt markets at a degree we've not seen before. In short, put this in a context. Yeah. Well, first of all, thank you for having me. It's hard to overstate the paradigm shift that's been happening in the corporate credit market, both in terms of the scale of the capital need, but also in the broad scope. And one thing that I think is probably not appreciated at the very start of this kind of analysis is that while the needs from the hyperscalers are significant, our work has suggested that the hypERScalers are actually only 40% of the total year-to-date AI-related debt issuance. So it's all the other stuff. So it's a lot of the other stuff, but when you're thinking, like, put yourself in the shoes of a portfolio manager that's navigating AI-related exposure, you're doing that in the equity market, the private markets, and then also in the credit markets across a broad range of sectors. To put this in context, our equity analysts are expecting this to be a multi-year, several trillion dollar investment cycle. As you noted, a large part of the universe is using operating cash flow to finance it, but no market on its own can do all of the financing here. And so where that relates to the private markets is that while there is still a runway for the traditional bond markets to service this financing need, at some point as this multi-year issuance cycle extends, we believe the private market will play an even larger role than they already have in financing this build-out. Now, there are a lot of different nuances in terms of, you know, how investors are thinking about taking this exposure, kind of picking their spots. But at a very high level, the key point from us is that we're not concerned about access to capital for this theme. But we do expect, as the multi-year issuance cycle extends for a host of reasons, for there to be a more nuanced conversation about where exactly to take this risk and at what price. And that means different markets, different structures, different currencies. Yeah, I think that is the most complicated part for any portfolio manager today. What part of the AI trade do I get into? And it's so easy for me to see why fixed income, or even public or private right now, is more attractive. People are looking to hedge their downside risk. But if only 40% of the issuance in question here is those hyperscalers, how do you best categorize what's happening with that 60%? It's interesting because a big part of that other bucket are financing structures for data centers and chip financings, but it's all still related to that broader AI theme. And I think this actually gets to the heart of one of the pieces of our research, which is the binding constraint, in our view, is really about market saturation and issuer concentration exposure in AI, as opposed to how much debt can these companies add and just still stay investment grade rated. And I think one of the interesting points that you asked at the start kind of put this in context, the credit market doesn't have an issue with funding large periods of re-leveraging. We've seen it time and time again, but the key distinction is that it usually happens with debt-funded M&A. And that's meaningful because in debt-funded M&H, our work has suggested that on average, there's a five-month lag between when an M&N transaction is announced and when the deal comes to finance it. So these are large, chunky deals, but that five-month period gives investors time to do work on a capital structure, get comfortable with the financing trajectory, make room in their portfolio. What is happening with the AI build-out is that a lot of this debt issuance is kind of not well telegraphed in advance. We all know the need is there, but it's not following kind of that same playbook. And really when the credit market works best in funding periods of debt issurance, It's when there's a quantifiable need, when there is an end in sight, and then when there is a period of debt reduction on the other side. And so there's not really a quantifiably firm need for CapEx because the estimates keep moving higher. There's not an end-in-sight because this is a multi-year period. And we don't have the same focus on debt reduction like we do, for example, in M&A when a corporate levers up temporarily to do a transaction and then commits to deleveraging on the side. So that's what's – we put out a piece kind of like why this time is different. That's the difference, and that's new for credit investors because typically we don't have kind of large jumbo deals that aren't well telegraphed in advance. And so that has implications. Yeah, to put it in real-world terms, it's kind of, like, well, okay, you order a pizza, and the pizza comes back to you, and you don't know if it's any good, but it's the only thing you can eat, and then you have five minutes to eat it. I mean, or said another way, like the way maybe I would approach it is I'm going out to a really fancy, great meal. But if I plan in advance, I can actually have room to digest that meal. Whereas if it's a surprise and I'm like and I've just eaten lunch and then I'm sitting down to a fancy meal, then maybe I don't have as much room to digests like that. I feel a little better about having spent the two hundred dollars. You actually don't know how far those two hundred hours will go in the case of that. And I think the key point is that let's just focus on the US IG bond market, which is kind of the largest, deepest liquid market. It's an $11.5 trillion market when you include kind of all of the potential debt that you can capture on the corporate side. It's significant. There's a lot of financing capacity. But the point is, just another stat, of the jumbo deals that we've seen in that US IG bond market today, which we captured $10 billion and higher, more than half of them have been tech related. So that's just, it's a heavy amount of concentration in large deals that are coming through the market kind of in quick succession. And as you know, in the credit market, when there are new issues, they tend to come with what we call new issue concessions. So they price at a bit of a discount relative to the existing bonds. And so you can see how kind of repeated issuance in a given theme can really weigh on existing holdings. So you just have to be careful about how frequently you're tapping the market and when you're committing to staying out and deleveraging. And I think that's truly why this time is different in terms of the ability of the credit market to finance such a large theme. I've heard you talk about how that when you look at really what's going on, you're now really only looking at this theme account for what, 12% of the total outstanding. So AI related exposures, as a total of the entire IG market index that is outstanding, it's 12%. And that is behind where banks are, right, which is more than 20%. That's right. Right. So but it's, it doesn't seem to me that it's just the volume. It's the pace at which it rose. That's very rapid rise. Do you ever, you know, do you get concerned on how quickly a lot of these companies are raising money in debt markets? Well, I think they're starting from a very low base. It actually reminds me a lot of in a prior cycle, I actually covered the pharmaceutical and insurance sectors when I was first starting at Goldman Sachs. And the pharmaceutical sector is very different from AI, but there are a lotof parallels into kind of how those balance sheets evolved over time. And I'll just give you one kind of snapshot is that when I first started covering pharma, many of those companies were very highly rated. Some of them were actually AAA. There was a wave of debt-funded M&A in the sector that caused those companies to add debt to their balance sheets kind of strategically in a pretty widespread way, and the entire sector moved down four to five rating notches over the span of a couple of years. That may sound surprising, especially to maybe someone who's not in the equity market, but the starting point was so high in ratings and so low in leverage that actually that sector adding a significant amount of debt and encountering multiple notches of ratings downgrades actually left them comfortably still within investment grade territory, easily accessing IG capital markets. And actually, their cost of debt was a little bit higher, but it was well worth it. So exactly. So the point is, is that from where the IG tech sector is starting, it's actually a very similar position. You have a handful of companies that are very highly rated, very low levered, plenty of room to add debt capacity. What actually matters for corporate credit investors is why are you re-leveraging? If you are re-leveraging to do a debt-funded share buyback, that's a lot more difficult to internalize because as a bondholder, you may not be benefiting from growth, from additional expansion opportunities. You're just adding debt to your balance sheet to return it to shareholders. And this is so competitive that there is a degree of if you're not raising money when you can at attractive enough rates, then how are you going to be competitive against the rest? That's right. I mean, I think the risk to the build out from what I can see from what these companies are posturing is that there's a risk to under-investing, not over-investING. The other key lesson from the corporate issuance side over the past few cycles, especially the financial crisis, but also the pandemic is issue when you can, not when you have to, right? Because capital markets can be fickle and kind of close. So I think that's what we're seeing. And actually, it's a great point because when we were thinking about our forecast for debt issuance for the hyperscalers, the initial approach that we were going to take was to look at the cash flow from operations, less the CapEx, and kindof come up with the difference. What we quickly realized is that this sector started issuing in 2025, when cash flow from operations was still in excess of CapEx. Meaning, if you just looked at the financial statements and looked at The Cash Need, you wouldn't have expected them to become such heavy issuers already. So what to me that suggests is that these companies are appropriately getting ahead of a planned investment spend and are issuing proactively. And actually, as you well know, they haven't even focused solely on the US IG bond market to do that. they've issued equity, they've issued across other currencies and regions, they've already been involving the private markets, which also suggests that not only are they proactively getting financing in place, but they're diversifying those financing sources while they do that. And I think that speaks to what we discussed at the start, which is the binding constraint is kind of market saturation. And, I almost think of it like a waterfall of capital markets availability, kind of accessing each capital markets in turn. And that's how they will build up the firepower to do this investment. Well, talk about what this means for private markets. But I feel like people really need to understand what's happening with issuance because there's the hyperscalers. And from everything I could tell, it looks like those hypersalers have kind of been able to not only raise a lot of capital, raise it at fairly attractive rates until this point, raise it with longer duration. That's right. It seems like a lot of the other types of companies, neoclouds and otherwise, have a lot shorter of a duration than those large hyperscalers, how big of a problem is that? Because when they go to refinance, it could be at a much higher rate and with a much weaker balance sheet. Well, you're right. So to frame it, so the average, so the hypersalers globally in the IG market have issued 229 billion of debt so far this year. Yeah. Pretty meaningful acceleration from 108 billion in all of 2025. And you just raised your estimates. We raised our estimates. We were expecting 400 billion in 2027. We tweaked it a little bit to 420 billion. The weighted average maturity of that debt this year is 16 and a half years in the IG bond market. So it's a little longer than what you typically see in IG. The other part of the market that you're referencing, the high yield rated, most of those structures are kind of like five years. But that's very consistent with the market convention because, as you know, high yield is much shorter duration. So there is a focus on, I would say, more deal level nuances in such that if you do have refinancing risk, If you are not fully amortizing, kind of what will the environment look like in terms of data centers and compute when you go to refinance? So I would say that the key takeaway is that credit investors are kind of doing that bottom-up work in those more nuanced structures that are technically classified as project finance that are coming through our market, which has also been a new asset class for us. That's actually not something that we've seen historically until this year. in the acceleration in high yield has been really swift, more than 15% of the supply in the high yield market has been project finance. Amazing. So like kind of almost a new asset class overnight. So when we think about project finance, I think a lot of people are looking what's happening in that project finance space. And there's a lot confusion about what's going on. So it's creating a lot nervousness. What does it mean? Are we financing real estate? Are We financing something that is closer to the hyperscalers credit ratings or other firms around it? And how much do we have to worry about kind of those derivative effects? And, you know, I'm wondering, you look at what credit spreads are doing, you look At what CDS is doing, what is that telling you about fear in the market? And does it match the reality of what's going on? So I view it's a great question. I view the widening in credit spreads or the widenning in CDS spreads as more of a technical indigestion as opposed to a broad based concern about fundamental deterioration. The CDS market is interesting because it can be a directional barometer on sentiment in the credit market, but it has its limitations in truly reflecting fundamentals. I think from the perspective of the project finance issuance, we've emphasized what we call the three Cs, which are construction risk, claim to tenant, and coverage. Those are kind of the three areas that we really recommend focusing when you're doing the bottom-up work. As it relates to the relationship with the hyperscalers, there was a perception earlier this year that investors would likely think about those in terms of different credit risks. So they would look at the issuer of the project finance deal differently than they would Look at the issue of the hypERScaler, even though there was A relationship where the hypER scaler would be kind of backing the lease. I think what has shifted, especially since the summer, is that I sense that investors are thinking about their AI exposure in total. That's really the challenging point, I think, from the market indigestion perspective is that this is all kind of exposure to one theme. And so it's less about capital being raised in this bucket or this hyperscaler bucket or this other area of tech, but it's more about as an investor, how am I managing my overall exposure to this, especially if I already have exposure in the equity market? So I think that the periods of episodic widening, as we've called them. They're less reflective of true concern about fundamentals, but more about, okay, I can only have so much AI exposure in my portfolio. And so where am I consciously choosing to take it? So let's talk more about the role of private markets, because you had said that that need is going to most likely rise in the coming years for private market exposure and financing of a lot of this AI build out. What's the biggest risk to that. You had mentioned that people feel a little overweight in artificial intelligence already. There are a lot of questions about how this project finance plays out when it's pretty much a new asset class, as you've been saying. So what's the crystal ball there? I mean, what does this amount to? I sense from our conversations that there's still a runway to put money to work. So I don't sense that investors are overweight yet. And actually, in many instances, I feel like they're strategically underweight, but it's not clear to them at what point they should start to get involved. Because again, this is unique for the credit market in that you usually have like one wave of issuance in an issuer or a sector, and then that's it. In a multi-year issuant cycle, I think it's a new learning curve for investors to think about, okay, at what time do I step in? Do I step into the 2026 vintage? Do step in in the 2028 vintage of issuancy? How do I think about that? For the private markets, we expect private markets to do even more of the heavy lifting in the financing, call it in kind of the 2028 and beyond timeframe, for the simple reason that we still see runway for issuance in the traditional credit markets, so the syndicated markets. But after the hyperscalers reach the level of some of the largest issuers today, which happen to be some of the U.S. banks, we think that there will need to be more diversification across investor types. Now, what is notable to us is that the lines appear to be blurring a bit in the private markets about what areas are participating. So private infrastructure and private real estate are kind of the real asset categories. They have some inflation protection. They have especially, I would say, relevant expertise in digital infrastructure. But we've also seen financings classified as venture debt. We've seen private equity participate. So what is most relevant, in my view, is what is the financing firepower in total in the private markets that can help participate in this theme. And as you know, it's substantial. It's $4.5 trillion globally across all strategies in terms of dry powder on the sidelines. The biggest risk would be if the investor diversification in the privates is not considerably different than in the syndicated markets. Because then you go back to that binding constraint that we talked about at the start, which is investors probably only want so much AI-related exposure to the market. And so if you have a lot of overlap between the same investors in the private markets and the same investments in the syndicated markets, then you may just bump up against that constraint. Now, this is part of the reason why we think private infrastructure and real estate will play a role, because my sense is that those are actually more different investor types. And where I see the most overlap is, say, between syndicated market and the private credit market. So that would be kind of point one. The other, I would say, notable point that we've sensed is that in my conversations with clients that have patient capital, oftentimes the private markets are a pretty meaningful draw for where they want to pick their exposure within credit in AI. Because if they feel like they have the expertise, the patient capital to participate in private markets transactions, they can earn a higher spread. And that's been attractive. Is there a trade-off there, a higher spread, you're saying, in private markets? Is there more downside protection in privatemarkets for the end investor relative to public markets? Or is it the other way around? No, there can be. And I think actually, in particular, this is true across private markets, not even just with AI. But if you take the broadly syndicated leveraged loan market, for example, so the public credit market, and you look at direct lending in private credit, what you see is that there are more covenants in private credit and there are More opportunities to kind of tighten terms and conditions relative to broadly syndicated loans, which are effectively covenant light. So there are more investor protections. Now, of course, that's a bit of a generalization. It will matter in terms of what specific structures you are participating in. But that was always kind of the appeal of the private market in total is that you have an opportunity as a lender to really drive terms, pricing, and you can do that because you're providing certainty of financing and certainty of execution to the borrower. And that was the method. So we're sitting here speaking just after we saw the Federal Reserve execute on one rate hike in 2026. There's more expected. What happens as the cost of borrowing rises. Do some parts of this trade come under fire simply because rates are rising? I think in general for the leveraged finance universe, and so this would include private credit and broadly syndicated loans, what we are expecting is that some of the fundamental improvement that we've seen over the last few quarters will stall. And I think it's a function of floating rate borrowers will automatically encounter higher debt service costs because of the rate hikes. It doesn't mean that fixed rate credit is immune. It's just that fixed rate credit will encounter higher borrowing costs only if and when they refinance lower coupon debt. So in general, that's what we're bracing for. Now, I don't think it will go as far to cause a new default cycle because of two reasons. One, while the Fed has hiked and our economists expect one more hike in October, the overall trajectory for rates has actually been materially lower than the peak. So we've had 175 basis points of rate cuts since the Fed was at peak rates in 2023. And then, of course, we hiked in September. But there's still a material amount of interest rate relief for the floating rate borrower that was delivered since that time. Now, further out the curve, yields are actually at multi-decade highs as of the time that we're recording this. So that's a bit of a different backdrop. But for the leveraged finance universe, it's really that floating rate debt that matters. The second key point is that growth is pretty good. And so the worst cocktail for kind of leveraged finance is an environment of sluggish growth and higher debt service costs. That's not actually what we have now. We actually have pretty resilient growth and modestly higher debt surface costs, but again, still well below the recent peaks of 2023. So that is the framework that we're operating under. I think the most challenging part of the market as it relates to both syndicated credit and private credit is the capital structures that just never really grew in to a higher interest rate world. They never aged into their capital structures. Maybe they were formed in 2021, ultra low rates, peak valuations for software. And then from the perspective of the higher interest rate environment that we have, we'll just struggle to achieve kind of deleveraging. That's that left tail that we keep talking about, which just like hasn't participated. Right. And it's been a lot of can kicking. Exactly. Yeah. And so, you know, while I still have you, one last question for you, because this higher for longer rate environment that we are now experiencing for private credit largely, on one hand, it's great for a floating rate asset class. On the other hand, It can cause some strain on some borrowers net net. How do you think that this is going to play out? I think what is most surprising to me is that private credit fundraising reached a record in the second quarter of 2026. And actually, the third quarter is on pace already to achieve last year's third quarter, which suggests to me that it probably skews more towards bringing in demand for the asset class as opposed to a renewed kind of outflow because of concerns about fundamental credit risk. I think the key distinction is that in 2022 and 2023, we had significantly higher rates, but we had a lot more concern about recession risk and downside risk to growth. In this environment, we have pretty supportive yield backdrop, but we don't have as much focus on the downside risk of growth, which is good. So I think that will actually spark some demand for credit risk broadly. If that balance turns such that we become more concerned about growth, then I would expect appetite for kind of speculative grade credit to wane. But that's not what we're seeing. So on the margin, I think that it could actually be supportive for demand. Not that dissimilar to the yield based demand that we see in the credit market for IG and high yield bonds where spreads are tight, but yields are attractive. And there's been a lot of demand for the asset class. Just an incredible backdrop. Amanda, thank you so much for joining us here today. It's definitely been a year that I don't think a lot people would have thought would have played out this way. That is Amanda Lynham, who has been doing a tremendous amount of work on this AI CapEx build-out at Goldman Sachs, and you've been watching The Bridge by iCapital.