Managing AI Risk in Credit Portfolios

Greg: AI is evolving at breakneck speed, and of course investors are trying to keep up in real time. Now, the challenge in credit markets isn't necessarily chasing every headline, but rather it's understanding how this emerging technology will impact business models, cash flows, and the overall credit worthiness of borrowers both in the software sector and across the broader economy.

Brad: Not all software is created equal. I think that there's gonna be winners and there's gonna be losers. There's gonna be companies. That innovate, that incorporate AI, that they become stronger as a result of it. And then there's gonna be companies that don't, right? And they're gonna fall behind.

Greg: That was Brad Lewis, technology Sector Analyst for Barings Global High Yield Team, and this is Streaming Income, a podcast from Barings.

I'm your host, Greg Campion coming up on the show. How the Barings team thinks about managing AI risk in credit portfolios today. Before we get into the conversation, remember you can follow streaming income on Apple Podcasts, Spotify, and YouTube. You can also follow Barings and subscribe to our monthly newsletter where credit is due on LinkedIn.

With that, please enjoy this conversation with Brad Lewis. All right. Brad Lewis, welcome to Streaming Income.

Brad: Hey Greg. Thanks for having me.

Greg: Psyched to have you. Psyched to have you. You're a man in demand these days. Yeah. Taking all your expertise in technology.

Brad: It's been a little busy to start the year.

Greg: Yeah. Well, I'm psyched to have this conversation with you because, obviously everything, AI is very, very topical right now, so there's a lot for us to get into.

But maybe before we do that, for our listeners and viewers who are not familiar with you, talk to me just a little bit about your background and role at the firm.

Brad: Sure. So I'm a research analyst on our global high yield and structured credit platform, which is about a 95 billion dollar platform, if you exclude structured credit, just within our global high yield platform, it's about 60 billion.

We've got over 60 investment professionals, within that I cover the technology sector, and so, that's broad. There's a lot of sub-sectors within technology, but the largest component of the market is software, which is where I spend most of my time.

Greg: Yes. And a sector that is very much in focus at the moment. How long have you been in this role or covering the technology space?

Brad: So I've been with the firm coming up on 16 years, and I've covered the technology sector the entire time.

Greg: Okay, cool. So let's, we're gonna get into everything AI related, but I want to kind of ask you, just covering the sector for 16 years, in a sector like technology where, which changes so rapidly, that's almost like covering it for 50 years right in another sector.

What's been the biggest change that you've kind of seen over that 16 year time period?

Brad: It's interesting. I would say, you know, overall in addition to just the growth of the market, I would say the structures have certainly changed over that timeframe. So I would say one big trend would be just the leverage profiles of these deals and the valuations that are in the deals as well. So if you look back, like the very first deal I did was in July of 2010, it was a software buyout. The first lien leverage profile on that business was three times, and it was a business that was being bought for 11 times ebitda. And you fast forward, certainly not, you know, this year, but before we had all this activity with the selloff and software.

You were seeing deals that were done at multiples of 20 plus times, and you would see first lien leverage profiles in the six to seven times range. So valuations have really expanded, leverage has really expanded within the system, and that's what I would say structurally. And then I would say technologically, I think that the biggest things that have changed over the course
of the last 15, 16 years. First we had cloud computing, and the growth of that when I started investing in the space you know, cloud computing was in the early days and so the SaaS business model has really, you know, taken hold and then more recently with artificial intelligence that's, you know, gonna be a big transformation in terms of technology.

Greg: Okay, cool. That I love that context, and that seems really important also to have that context of how things have developed over the last kind of decade and a half to give you that right perspective to analyze what's happening today. So in November 2022 ChatGPT comes out.

My perception is late '25, early '26, maybe we have hit another inflection point or another step change because AI goes from being kind of a story that we're all following and something that's kind of seemingly developing, like gradually.
To, okay, all of a sudden this thing is accelerating rapidly and it's gone from a story to like the story.

Not only. For all of us who just monitor technology, media and things like that, but for investors, for markets, for credit investors. So tell me if that's your perception as well in terms of like the late '25, '26 step change, and then what is behind that?

Brad: Sure. So I would agree with you in terms of your timeline, and I would say, I think in 2025, what we saw and what really started to, I think, raise eyebrows was the level of spin that was taking place from hyperscalers and the acceleration of the spin.

And then the leading model providers, Anthropic and OpenAI, they were getting bigger, they were raising more money at larger valuations, and so the market really started to pay attention to that. And then as we got into the fall, the market really started to stir about like agentic capabilities. And so this idea of these models not, not just being an interface where you put in a prompt and it gives you a response, but the ability to actually perform work. Automatically without human intervention. And so what we saw is the equity markets, at least within software, IGV, which is a, a software ETF, which is a good proxy for software equities. That actually peaked in September of 2025. So I would say post Labor Day we started to see some jitters in the market, even as it relates to credit markets.

We saw some idiosyncratic activity, so some legal service providers, certain names that started to feel some pressure in terms of trading levels and whatnot.
But it wasn't any thing that was, I would say, widespread. It was pretty orderly. And then we turned the page and we got into 2026.

And I would say there, there were really two things I think that happened to start the year. The first was the release of Cowork, which was a tool from Anthropic and the ability that tool provided. And folks started to mess around with it and play with it, experiment with it. And I think it really caught the market by surprise because not only was the market surprised by what it could do. It was also, I think, surprised by how fast things were moving and the pace of change, even from just what was out there, call it six months ago. In terms of the capabilities of these models. And then the other thing that happened on top of that, there was a research report that came out by an outfit called Citrini.

Greg: Ah, yes, that's right.

Brad: And it was a really interesting piece. It was a hypothetical thought piece. Really kind of looking at a scenario that could potentially play out.

Greg: Real, like doomsday scenario, basically mass unemployment, et cetera, et cetera.

Brad: Correct. Right, and that, and that sort of just, I think, stoked fears, and kind of added fuel to the fire, and so when this happened and the the market started to say, okay, wow, these models are really advanced and they're advancing really fast. It, the, the equity market really started to reassess software. They really started to reassess, okay, the future growth prospects, the future profitability, the competitive you know, positioning of these businesses.

And where what that really reflected too, was this reassessment of terminal values within these businesses. And so even though the outlook from a growth and an earning standpoint might have been, you know, not that unchanged in terms of as we look out to 2026 or 2027, it's really, as we look out beyond that, about what's gonna happen down the road and in the future. And so equity started to sell off and we saw multiples really compress in the public equity markets. If you look at just the IGV index that I referenced earlier, peak to trough from September when it peaked, to this year we saw a sell off of about 37%.

So a significant move lower, and a significant compression in multiples, and then that bled into credit markets. And so not only was our market you know, trying to reassess okay, the future prospect for growth and for earnings, but terminal values are just so incredibly important as it relates to credit market.
And so we started to see more widespread, broad based, repricing of risk within our market. And then, you know, the other thing that I think played out in our market were some technical factors. We had the BDCs, which have been in the headlines. As they grew, they started to participate a little bit in the broadly syndicated loan market.

And so, there was some selling as they looked to, to be able to reduce their exposure in those vehicles. We also had some technical factors within the broadly syndicated loan market just around CLO activity. Which, as most folks know, CLOs are about 70% of the broadly syndicated loan market.

They were selling lower rated credit, which a lot of these software credits tend to be lower rated in that B rating category. As they had refi activity or reset activity, or some of those deals were called, they were selling some of those kind of higher risk credits. So that, those technical factors, I would say exacerbated the sell off and the pressure that we saw. And this was really more of a phenomenon in the loan market. The loan market is really what's financed the private equity activity within software, so it's grown to be anywhere from like 13 to 15% of our market.

Greg: Okay. Not so much in bonds.

Brad: Yeah. So in high yield technology, exposure is probably half of that. And then software is a subset of that. So it's probably software is less than 5%, so there are certainly some software names within high yield that felt that, but it was not as big of a part of that market.

Greg: Okay. Okay, cool. Alright, that's great context. So really helpful to understand how it kinda started in the equity market, bleed over to the credit markets, and then it's kind of a combination of both fundamentals and technicals, I think, that are weighing on, I guess especially the loan market.

Given that sort of backdrop you just mentioned, you know, how are you thinking about the overall, I guess risk profile or how are you thinking about software just generally speaking?

Brad: Sure. So, you know, no shortage of headlines.

Greg: Yes.

Brad: And certainly, you know, I think that those can be attention grabbing and, you know, really sell newspapers or sell clicks, or what have you.
And we think that it's over-exaggerated that concept of just the death of software, or SaaS apocalypse is hyperbolic, and that it is overstated. However, we don't think that concerns about the space are baseless. I do think, and we do think that AI will be disruptive.

But I think it's important that, all software is not created equal. And so you've got very different types of software companies, depending on the use case of the software, depending on the types of verticals that the software sits in, the types of customers that they serve, et cetera. So there's very, you know, many different types of software companies which I think is important to point out.

And I think that what you're gonna see is that simple applications are gonna be potentially, you know, displaced, right? Or disrupted. But I think very complex enterprise grade enterprise applications aren't going anywhere, so I think you're gonna see real dispersion in the space.

There are gonna be winners and losers. You can't just paint the space with a broad brush. And I think ultimately, the winning strategy, you know, for lack of a better term, I think is really complex mission critical software coupled with AI capabilities. And the software companies that innovate and are able to embed AI into their offerings, I think are gonna actually have an opportunity to potentially increase their value proposition. And the folks that don't innovate and aren't able to embed AI into their capabilities, they're gonna get disrupted in one way, shape, or form.

Greg: Yeah. That makes sense. And I think you and I have talked about before that this is not the first technology disruption that we've seen before. There's been other examples throughout history where a new technology has come out. Everybody's immediately jumped to the conclusion that this is the end for the old technology.

Brad: Sure. Yeah. So historical context is important in the nineties, mainframe computing was really big, and along came the client server model and everybody thought that the mainframe was dead. You know, it was similar type of fears and headlines and not only was the mainframe not dead, client server did grow, but the mainframe, it's actually still growing to this day.

So that's one example, you know, where again, a lot of fear. Things evolved, but it wasn't exactly what everybody thought at first.

Greg: Yeah. That's a really interesting point. Alright, I want to get into a little bit more detail about, so you were kind of, you know, mentioning the idea that
there's, or alluding to the fact that there's gonna be winners and there's gonna be losers in the software space. So, let's talk in a little bit more detail about that. So we hear a lot about so-called moats, right? So whose business model is defensible? Who's is less defensible? Et cetera. So talk me through, I know that you and the team have a kind of really robust framework for how you are, you know, doing the real fundamental analysis and credit underwriting on all these businesses.

We don't need to get like, super granular, but like, talk me through like just how you are thinking about it. You know, what are some of the vectors you're evaluating companies on to try to distinguish who are those winners and who are those losers?

Brad: Yeah, so we do have an investing framework that we use as we evaluate software companies and their credit worthiness.

And it's important that every software credit that we look at before it goes into portfolio, it has to check all those boxes. It can't just check some of the boxes. There are real criteria that we look for in these businesses to say, hey, this is a defensible business, we think that this is a good

Credit investment, as it gets to moats, I think it actually helps to kind of start a level higher. If you think about enterprise architecture as it relates to software, I think, and again, I'm making a generalization, but I think any enterprise, ourselves included, think about it in kind of two different buckets.

There's core systems, and core software. And then there's non-core software. So core software is really mission critical software that you run your business on. It's absolutely critical to the business, so many people within the organization touch it. And then you've got non-core software systems, which may be very functional.

They may solve very specific, you know, use cases, and be very good, but they're more niche in nature, right? They're more sort of targeted. We call those oftentimes point solutions within software. So what we want to do from a credit standpoint is we want to be invested in these core systems because we think that ultimately core systems have what, you know alluded to, is moats. And the moats that we think about with software, and there can be many, but I think that there's really three moats that we care most about. Number one, which I alluded to, is the software needs to be mission critical. So this is really the use case of the software.
And when I say mission critical, that, you know, that's a term that can sound cliche, but it really means that if this software does not work, the business doesn't run. There is a really high cost of failure for this software going down. So you think about a general ledger for, you know, a business or you think about within our business, you know, we use tools that go across the enterprise for our investment teams in order to be able to manage our portfolios. On a day in, day out basis, we can't make trades, we can't, you know, actually do client reporting, and that sort of thing if those systems are not running.

So mission critical, high cost of failure, that's the first one. The second one that I would mention is proprietary data and really deep domain expertise. And I think that domain expertise is, it's something that I'm really high on because it is very difficult to replicate. First of all, if the data's proprietary that speaks for itself.

But domain expertise that's been built up over 20, 30 plus years across enterprises is very, very difficult to replicate. And so, when that's embedded in a system, we think that's a very defensible structural moat that's in place.

Greg: That's a really good point that data is not just something you could pick up off the internet and replicating that software.

Okay, so mission critical, proprietary data and deep domain expertise. You mentioned there were three.

Brad: Yeah. So, the third one and the last one that I would say I think is really important is a network effect. So the idea that as more enterprises use this software, the software itself becomes more valuable. And it also becomes more valuable to the customers that use it. So in our world, I think an easy example of this would be Bloomberg.

Everybody in our market is on Bloomberg, and so Bloomberg becomes more valuable and more entrenched, the more people use it, the software itself becomes better.

Right? Because it's getting more and more data from more and more users, at more and more enterprises, that's allowing it to refine and make their product even higher value.

Greg: Yeah. I feel like every other week I see one of these posts on LinkedIn and it's like, this company just came out with this new finance tool, RIP Bloomberg.
And you're like, whoa, wait a minute.

Brad: Once something has become the industry standard, and it's become adopted, and everybody uses it, there's a real sort of centrifugal force that exists. And so, when software, you know, gets like that in certain industries, it's really, really hard to switch off of that, even if you wanted to, right? It's really difficult to switch.

Greg: Yeah. That makes sense. Alright, so you talked a little bit about, you know, maybe it being over exaggerated, the SaaS apocalypse and all this kind of stuff, and it's much more nuanced than that. Just a question for you in terms of how the market has reacted to this, and I know, you know, market prices are changing every day, but like if you were to make a broad, general statement, do you think the market has kind of priced in this AI risk efficiently in your markets? Is it overdone? Is it not factoring it in enough? Where are we on that spectrum?

Brad: Yeah. So, I would say that I think it's a little bit of both. I think that there are examples for sure where risk needs to be reassessed, right? And risk needs to be repriced, and so I think the market has done an effective job in repricing that risk in light of the recent developments and what's going on. And then, I think that there are other cases where certainly the market is oversold. So, I think it just depends. There's definitely examples of both.

Greg: Okay. And that brings up the idea of, you know, potentially being able to capitalize on opportunity when the market is overly aggressively negative on certain credits. Which is not just an AI phenomena, but I think it's something, you know, as I've had conversations with members of the high yield team for years, that's always, you know, there's always been opportunity in volatility and this seems no different.

Okay, so we've talked about software, but obviously AI is not limited to impacting software businesses, right? It's, and this is both from a risk perspective, and from an opportunity perspective, right? It's not, you know, naturally we're talking about credit markets, so we're gonna be looking at the potential downsides, right? But there's opportunities as well. Businesses are gonna be getting more efficient, new businesses are gonna be created, et cetera, et cetera. But, I'm just curious from your perspective and given your domain expertise, where are you, kind of, getting pulled into conversations on other sectors and are there other sectors that jump out to you that are either, opportunities are being created or risks are being created.
Brad: Sure. So I would say that there's certainly both, and I'm getting pulled into a lot of conversations, whether it's across other sectors in our coverage, on our team, or just across other investment teams.

So, from the opportunity side, there's a huge build out that's going on in data centers. We've seen a lot of issuance within the high yield market to help finance that. In fact, just this week, an over $5 billion deal got priced in our market, which is a huge financing.

And there's a whole ecosystem, or supply chain if you will, of businesses that are gonna benefit from that build out. And I think that there are a lot of opportunities in those credits. Some of those are more industrial type credits, but they're supplying, you know, different things that go into the data center that are gonna be important.

In fact, I was just sitting in our investment committee this morning and we were looking at a third-party logistics provider that is really benefiting from the build out and data center. So, there's a lot of opportunities, not just investing in the AI providers themselves, but a lot of the folks that are tied to that, that could create compelling opportunities. And then, I think away from that, on the risk side, it's, you're right, it's not just software. I think where we've seen the other space that's been hit the hardest is more of the service type businesses. And services is a broad term, but think legal services, or staffing services, or business process outsourced.

Outsourcing services, anything that the markets starting to think about. Okay, could this be automated? There's a human that's in this process, you know, they're charging a fee for it. Could we either do this ourselves, or could, you know, new competition come in with AI capabilities and really, you know, change the business model across sectors.

Greg: liquidity spectrum? So you mentioned, kind of what we saw earlier this year in the BDC space, a lot of questions around who owns, you know, what percentage of software, loans, all that sort of stuff.

Obviously, we've got a big private credit team here. I talked on our last podcast with Brian High a little bit about this topic, but I know that you're interacting with our private credit team all the time. Are there any differences that investors should be thinking about or keep in mind? Like with regards to, kind of, like private market or private credit exposures to software, versus public credit exposures? Are they like different size companies? Like any factors to think about there?
Brad: Yeah, so I do think that there's several factors to think about there. If you start out just at the size of the exposure within technology, or within software in general, that it's certainly higher in private credit. You know, anywhere between 20 to 25% depending on what data source that you're looking at, maybe even higher.

Greg: You're talking about like market wide?

Brad: Right. In terms of just like how big it is within the market. And I think that also varies depending on where you are in the private credit market. Are you lower middle market? Are you in the upper middle market? I think it's larger exposure as you get into the upper middle market, whereas on the liquid side and the broadly syndicated loan market, as I mentioned earlier, it's about 13 to 15% of our market. And then within high yield it's even lower.

So, and some of that is private credit is not invested in every sector, so they might avoid consumer businesses or they might avoid energy businesses. So I think that plays a role. So, I would say just the size of the exposure in the different markets. You know, the other thing that I think is interesting in terms of the structures that we see in our market, leverage and ratings really play a factor and differentiates between the two.

So within the broadly syndicated loan market, as I mentioned, CLOs are really the driver of demand in that market. CLOs have limitations on triple C investments that they can make, and so for a new issue deal to come through, it really can't clear the market if it's triple C rated.

So it's gotta get a single B rating, which really puts a governor on the leverage profile that that business can have. So, I mentioned kind of getting up to maybe six, six and a half times on a first lien leverage basis. That's really kinda where we tap out and where the ratings agencies tap out.

Whereas in private credit, ratings are not a constraint. So you'll see software businesses that would probably be triple C rated if they came to our market. And so, they've gotta get financed in the private market because CLOs can't in invest in those types of businesses in our market.

So those are a couple of differences that I would point out that I think are important. In terms of the businesses, I think it depends, like, at the lower end of the private credit markets, certainly the business profiles are smaller, but as you've seen that market raise more capital, and start to do bigger deals with bigger businesses.
Our markets have really converged, and I would say at the upper end of the private credit market, the business profiles aren't necessarily all that different. These are some really large businesses that, you know, sell, like in the case of software, enterprise grade software, that's highly complex, et cetera. So, you know, that's pretty similar from that standpoint.

Greg: That's interesting. Yeah, I mean, I know on our team, on the private credit side, they're, you know, they're one of the more conservative operators in the industry. They're, you know, senior secured lending, middle market companies, kind of, you know, right down the middle of the fairway type of stuff, and you've been interacting with them, I think a lot.

Brad: Yeah, I've had a lot of dialogue with their team year to date. I mean, we're talking pretty much on a weekly basis, I would say. And a lot of that is just to compare notes on what we're seeing in our markets. It's helping us get that perspective of what's going on. For smaller businesses with different sponsors that maybe we're not exposed to in our market, are we seeing similar trends with our businesses that they're seeing with their businesses?

They're looking to us to be able to get help in thinking about pricing risk in their market. Given how fast things are changing in our market with where things are trading and so, there's a lot of collaboration back and forth, and it's a big advantage of our platform, I would say, to be able to go across liquid, illiquid, US, other geographies.

We get to see the entire market, which I think is, you know, really important and it's something to really take advantage of.

Greg: Yeah. That's great. And yeah, I mean if it seems really important, honestly, especially in moments like these.

Brad: Yeah. And the one other thing I would add there too, is that there's also opportunities. It's not just about comparing notes and thinking about priceing risk. It's also about, you know, we're at a place now where borrowers go between markets, so there might be an opportunity for a private credit financing, because our market is not there, and it might be an issuer that we've got extensive history with, and we know really well, we feel very comfortable with the risk profile. So that could be an opportunity for them. Or even vice versa. It could be a company that they've financed for a long time. You know, the business is changing hands with a new sponsor. They're gonna come into our market, and we've got a chance to, you know, participate in that. Or even with our Capital Solutions team as we get into more kind of stress distress situations,
there could be opportunities for us to do deals with them, so getting their perspectives, and collaborating I think is really key.

Greg: Yeah, that's really, that's a great point. All right. Enough of a Barings commercial here, let's talk about where things go next for AI. This is a very broad question, so answer it however you want, but I'm interested to hear where you think things go next. What are things you're gonna be watching?

If you wanna talk about agents, that would be interesting. If you wanna talk about, you mentioned new models, you know, obviously Anthropics Mythos models causing quite a stir. Anywhere you want to take it, but anything that you think's coming down the line, you're gonna be watching, maybe not fully on people's radar screens.

Brad: Sure. So I'm gonna stop short of making predictions about what happens next in AI. There's so many opinions out there. There's so many people that are way smarter than me that have views on where AI is going and what's gonna happen next. We don't know what's gonna happen in the next 2, 3, 4 years. And actually I think that's really important that we all have some humility about this. Because I, frankly anybody, I'm skeptical of anybody that is emphatic about, hey, this is what's next, right. Here's what's gonna happen in, in two, three years from now.

But I, you know, I would say that there's a lot of things that we're paying attention to, you know, as you alluded to before. The first thing I would say is I'm very focused on the CapEx spin. That's been, if you look at software performance, and you look at the hyperscaler performance there, there's really been kind of an inverse correlation with the acceleration in spending and what's been happening from a software standpoint.

So watching that, if that continues to grow and to accelerate, I think that that's gonna be a real you know, important indicator for, you know, what's gonna happen in terms of price volatility within the market going forward.

Greg: Are you watching the CapEx of companies as well in terms of borrowers? Because one thing that jumps out to me is like, I think there was some headlines a week or two ago about Uber's, I can't remember if it was their CFO or CTO, coming out and saying, Hey, we've already kind of blown through our entire token budget for '26 and it's only April, whatever. And so the concept of like, everybody's rolling out AI, everybody's super bullish on AI, but then at some point it's gonna sit on the CFO's desk and they're gonna say, wait a minute, are the productivity gains showing up.
Brad: Yeah. And in some cases they'll be increasing budgets, but in other cases, companies will have a hard time increasing budgets as they want to invest in AI. So one of the things that I'm watching for is, okay, if you want to allocate more resources to AI, where does that come from? If you can't just grow the budget by 10%, the budget's the budget, where's that gonna come from? Is that gonna come from, you know, it's gonna come from either other parts of your IT spin.

So, it could come from existing software providers, that's something that we're paying attention to. It could come from labor savings, right? So maybe you're cutting back on spend in terms of headcount. Or it could come from, you know, other third party spin that you have right there. So, kind of, what are the wallet share shifts, if you will, in the enterprise in order to be able to allow companies to make the necessary investments. To be able to be ahead of the curve.

Greg: Yeah, I kind of wonder from that standpoint, if that's the next phase. It's like the first phase was get it out there, get everybody using it, and then like phase two, I wonder, is more of an ROI focused phase.

Brad: Yeah, that's a great point. I'm very much focused on ROI, because the way in the conversations that I have when I talk with management teams, when I talk with third party experts. The way that this is sort of played out, is that everybody thinks that AI is a transformational technology.

So, every business is gonna be impacted in some way, shape, or form. So every board is asking every management team, what are we doing with AI? Every management team's now going to the next level down and saying, what are we doing with AI? So nobody wants to get left behind.

So there's a rush to adopt it and to say, okay, let's experiment with this. Let's find, you know, early wins and that's happening. But then it becomes, okay, how do we really not only drive like automation and cost savings within our business, but how do we actually drive, like, you know, revenue and real business value?

How do we, how do we transform processes? And that's really where the ROI is gonna come in, and that's something that is gonna be closely followed, and that's what we're focused on. And that comes down to, you know, kind of good old fashioned talking to the management teams, talking to the sponsors, talking to third party experts, all the different folks in the market. It takes a, you know, a lot of kind of manual effort in order to be able to kind of put that mosaic together. To get the view on what's going on.
Greg: Alright, so CapEx is one thing you're watching. Sorry to interrupt you, but what else is on the.

Brad: You know, open AI and anthropic, the model providers, they've been releasing a lot of models. Which we're, you know, obviously paying attention to and following.

Greg: I mean the pace is unbelievable.

Brad: The pace is unbelievable. Don't forget, those businesses reportedly want to get IPOs done, and so, I think that there's a lot behind releasing the models and the pace of releasing the models.

Every time they release a model, they get a ton of publicity and a ton of press. They get a ton of downloads, and a lot more usage of the models and, you know, it creates a real buzz. And so, there's a little bit of a mystical element right now with those businesses. I mean, what's the run rate revenue? What's profitability? You know, we don't know. They've self-reported some things and we've heard about funding rounds at different valuations, but I think it's gonna be really important for software more broadly for those businesses to get in the public market for us to be able to see, okay, what is the strategy?

What is the product roadmap and what does the business model look like? What does the cost structure look like? What is the ROI on this stuff? There's a mechanism where you got, when you have 50 equity analysts asking the same question every single quarter, that it creates this level of transparency and accountability that I think will help, kind of, demystify things a little bit.

So we're definitely watching that. And then the last thing I would mention with software companies themselves, I think that there's a lot of things that I'm focused on. First and foremost, it's how are they embedding AI capabilities into their products to drive incremental revenue, I guess protecting their base. But also, can they actually drive incremental revenue opportunities? What margin does that come at? So what is the ROI on that? I'm watching very closely how pricing models evolve within the software space. So most folks know that SaaS software, a lot of the pricing is based on seats. So actual people, clearly with the rhetoric around, you know, what might happen in terms of AI potentially replacing jobs, there's gonna have to be an evolution of the pricing model.

Greg: Yeah. I mean, I kind of wonder like, is there a world we're gonna be living in where the employment stuff is overdone, and what you end up seeing is more demand for all this stuff, you know, driven by agents. And because
companies end up wanting to do more, rather than do the same with less people. I don't know. Who knows? That's a kind of a, to your point around having humility, we just don't know yet how that's gonna play out.

Brad: One last thing I would mention, in terms of software companies, and this is saying that I, this is something that I don't think a lot of people are talking about right now but it's something that I'm thinking about for software companies, and it's this idea of deterministic versus probabilistic.

So I, I think a lot of folks are aware that. These models are probabilistic models, so they're taking known information that's out there in the world and they're making kind of a probability assumption on what the right answer is given a, a prompt or whatnot. The advantage that software companies have is that they are, they can be, you know, for systems of record and really kind of core mission critical software.

They can be a single source of truth, and so as they embed AI capabilities, do they have the opportunity to kind of become like a deterministic AI provider? Meaning that there's like factual guardrails around what's taking place as people or agents interact with the software? And the data, and the information, et cetera, so that you don't have, like these hallucinations that you're hearing about with the model or, you know, work slop or what have you.

So again, that's something that I think is like, you know, not necessarily tomorrow, but I'm focused on that with our software companies and saying, hey, okay, how are you embedding AI into products? Like, can you differentiate your AI capabilities, not just because you are the software and it's easy to buy from you and bundle it into your existing package, but you actually offer something that they cannot simply offer with their model.

Greg: Okay. Yeah, that's a really good one. Okay, cool. Let's finish up with a couple of maybe we'll call , lightning round questions.

I noticed a phenomena, I don't know if I'll call it like AI anxiety, just this idea that like, this thing's coming at us so fast. We've already mentioned like how many headlines are hitting and all that kind of stuff. Like how are you personally kind of like keeping your wits about you as somebody who is tasked with analyzing these and trying to figure out how it's gonna affect all the business models. How are you keeping your wits about you as these, as the headlines hit?
Brad: Yeah. Well, I mean, it kind of depends on the day candidly, but I think what it comes back to is having a very disciplined, repeatable process. So we have a framework that we use when we evaluate software credits, or just technology credits in general. That framework, certainly we update that and, you know, we reassess it as new information comes out, but it's a framework that, you know, is an all weathers framework. You know, whether the market's up or whether the market's down or whether, you know, new headlines are coming out. So you can't just react to the headlines, and chase the headlines.

Certainly, you know, I wanna be paying attention to everything that's coming out and be on top of everything, but it's really about having that disciplined, repeatable process that's tried and true. That, you know, works in good times and in bad, and that's really what keeps me sane.

I mean, if I didn't have a process that I was going through, I would just be reacting, right? I would be, you know making decisions out of fear. But that's what the process, I think does, and that's why I think it's so important, you know, for investors to be able to have.

Greg: All right. Cool. All right, last question.

So net everything that we've talked about, all the conversations that you're having with management teams, all the conversations you're having with sponsors, where do you come out on AI's impact on credit markets, let's say over the next few years? Are you more optimistic or more pessimistic?

Brad: Yeah, I'm gonna answer that question, maybe it'll sound like a hedge, it's not meant to be a hedge, but I think it's multifaceted. I am pessimistic in the sense that I think that we're gonna have continued price volatility in the market, and we've seen it already this year. We don't know exactly where this is going in two to three years. The advancements of the capabilities is fast and furious.
New models are coming out and so, there will continue to be headlines I think that cause volatility in the market. And right now, you know, there's this idea of you can't disprove a negative when people are talking about what's gonna happen in three and four years and speculating on that.

Who's to say that that's not what's gonna happen, right? So I think we're gonna continue to see volatility and we're gonna continue to see stuff you know, trade in kind of funny ways. I do think, though, and this is where I'm optimistic, is that I think that there's gonna be real opportunities on the back of that. I mentioned before, not all software is created equal. I think that there's gonna be winners and there's gonna be losers. There's gonna be companies that innovate
that incorporate ai, that they become stronger as a result of it. And then there's gonna be companies that don't, right?

And they're gonna fall behind. And so being able to identify those companies, and try to pick through those winners and find those guys. I think that there's gonna be real opportunities in, you know, the midst of volatility and market dislocation in order to be able to find really good risk adjusted return opportunities.

So, kind of taking advantage of that dispersion if you will. But what I would say on top of that is that it, like the fundamentals are key. Just, and it kind of comes back to old school, you know, what we've always done, which is, you gotta know the management teams, you gotta know the sponsors, you gotta know the market and the market participants, and you gotta be able to put that mosaic together to be able to make the calls. And you know, kind of find the opportunities and then, you know, I've talked about kind of having some humility, as it relates to it, but diversification I think is really important, right?
Running diversified credit portfolios because we don't know what's gonna happen in the future.

You don't want to be overexposed to any one name. It's about credit selection, not just, you know, sector allocation, but it's also about not being overexposed to one particular sector. So you need experience, you need a well-resourced team, you need a process that's repeatable and disciplined and patient.

And so, I'm optimistic that all of those things will actually result in and drive, you know, good performance going forward for our clients.

Greg: All right. I love it. Very nuanced answer, which I think is appropriate given that I think a nuanced take on all this stuff is probably appropriate as opposed to, some of the following, some of the hyperbolic rhetoric out there.

So yeah. Brad, this has been awesome. Really appreciated. Let's get you back on here at some point because as we've talked about, things are changing constantly, but I learned a lot and I really appreciate your time.

Brad: Sounds good. Thanks Greg.

Greg: Thanks for listening to or watching this episode of Streaming Income.

If you'd like to stay up to date on our latest insights across alternative asset markets, from credit and real assets, to Capital Solutions, please follow the
show and leave us a review on your favorite podcast platform. We're on Apple Podcast, Spotify and YouTube, and don't forget to follow Barings on LinkedIn where you can also subscribe to our monthly newsletter Where Credit is Due.

Thanks for listening and see you next time.