Our Perspective on Software and AI

Jeffrey Stammen
Hi everyone. My name is Jeff Stammen, Global Head of Investor Coverage here at Antares in New York. AI is changing how investors think about software. It's all over the headlines, and investors are asking what this means for private credit portfolios. Antares has underwritten software for decades. Software is meaningful but measured in the portfolio with historically resilient recurring cash flows. The key question isn't whether AI creates change. We all know that change is coming. The key question is how you underwrite software companies responsibly in a period of such dynamic change. The conversation that follows is with Antares Deputy Chief Investment Officer, along with senior underwriting leaders on our team. They'll outline Antares’ approach to evaluating software and AI risk and approach grounded in investment team frameworks that guide underwriting and ongoing risk management. So, without further delay, over to Shannon Fritz, our Deputy CIO.

Shannon Fritz
Welcome everyone. My name is Shannon Fritz, and I'm Deputy Chief Investment Officer at Antares. I'm joined today by two experienced and long tenured Antares senior underwriting professionals, Dan Landis, Managing Director, and Phillip Smith, Senior Vice President. Today, we look forward to having a discussion on how Antares thinks about enterprise software in an AI enabled and evolving world. So, let's get started. Dan, I'll begin with you. With all the recent headlines, why wouldn't an investor or advisor avoid software exposure altogether just to be safe?

Daniel Landis
Hey, Shannon. It's a great question. I think it's a really important question. And starting with the conclusion, we do think that having exposure to enterprise software companies is important for investors. So, software companies comprise about 4 to 5% of U.S. GDP. They are recession resistant businesses, they have recurring revenue, high margin and high cash flow profiles, and then importantly, our book of software deals have low loan to value and that really matters when you think about all the valuation shocks, right, that we've seen in the public markets in recent times. And just to put some numbers to that, if we modeled a 20% reduction in valuation of our software portfolio, our LTVs remained at call it 40 to 45%, which is a very manageable number. And then moving on from that, if we think about actual performance of software businesses. So, as I say that, I mean sales, you know, the historical sales and EBITDA performance of these companies, we've seen software deals perform favorably to non-software deals over time, so the percentage of our software deals that end up on our watchlist is lower than the percentage of non-software deals that end up on our watchlist. And if you do that same kind of thought process for actual losses, it's the same story, software being lower than non-software. So, what our view is, is that software is a very sizable part of the U.S. economy. It's an important part of the U.S. economy, and we don't think that investing in software has to be an all or nothing proposition.

Shannon Fritz
Thank you Dan. Very helpful perspective. Claude, Anthropic’s next generation AI assistant, released new agentic AI capabilities. How does this impact enterprise software? Does this mean the end of enterprise software?

Daniel Landis
There are lots of predictions out about this, as you know, reading the news, everything ranging from on one side, AI is going to take all jobs and eat enterprise software, all the way to the other side, where some are saying AI is sort of an overhyped pattern matching tool. And like most things in life, at Antares, we think the long-term reality is likely somewhere in between. So, there are some use cases today undeniably being materially impacted by AI, thinking about things like customer support with AI powered chatbots, software development with AI powered vibe coding, and we think anything that is consistent, repeatable tasks, is at risk of being automated by AI. So, that said, though, we think most businesses are going to continue to benefit from enterprise software for use cases that are complex, that are very well integrated, and very importantly, they're delivering a strong return on investment, they work well and they're AI-enabled. So, our view is that AI presents risks, but it also creates some opportunities, and AI is not the end of enterprise software.

Shannon Fritz
Great. Thank you, Dan. Helpful. There's clearly a spectrum as you outlined. Next, talk through a few ways that AI could impact existing software vendors.

Daniel Landis
There are three core ways that AI could impact an enterprise software vendor. The first is a client finding a new, cheaper, sort of more efficient AI first software vendor. So, a startup, right? The second is by building it themselves. So, sort of an in-house, you know, option. We think in both of these scenarios, established software vendors have an edge. And you know what's the why behind that. The why really is that all of these players have access to the same underlying AI models. The existing software vendors have preexisting client relationships, so they already own those relationships. And very importantly, the existing software vendors often have the data that is proprietary to train the AI, and they've developed complex workflows over time that really makes switching difficult. So those are a few reasons that if, you know, we think about a hypothetical world of unlimited time and unlimited financial resources, absolutely companies could certainly automate many of their tasks with AI but of course we live in that world of finite time, finite financial capital. So, our view is that we think successful companies are going to continue to focus that finite time and finite capital on their differentiation. Then they're going to leverage enterprise software vendors who have an AI enabled solution to help them automate those portions of their business that aren't related to their differentiation. And then to touch on the third point, I'd mentioned three. The third way that AI could impact an enterprise, an existing enterprise software vendor, is the potential for pricing model changes and seat reductions. So, vendors, of course, they’re incorporating AI as it exists today, and they're really working on how to communicate the value and return on investments to clients. So that whole topic of how do we monetize AI? It's still in discovery at this point. So that's really what we're talking about when we say pricing model changes. And then AI of course, presents the risk of reductions in force because of efficiency, and considering that software is often priced on a per seat basis, that could impact software vendors as well. So, what we're doing is we're keeping our finger on the pulse of both seat-based risks and pricing model changes, and over time, we'll react accordingly as needed.

Shannon Fritz
It's helpful overview of different considerations that are important to think through. Phillip, we'd like to move to the next question for you. Software is an important sector for Antares. Talk about how we select or separate the winners from the losers.

Phillip Smith
Yeah, it's a great question, Shannon, because there will be winners and losers here. I think first of all something that’s important to point out is that we have our finger on the pulse of AI through discussions that we have internally, with our experienced software vertical team, who meets at least monthly to discuss this topic, as well as conversations that we have externally with various experts in the field and consulting groups. The next thing we did was develop two AI risk frameworks. So, these frameworks are specifically built by Antares, so we don't show them externally, and they cover both software and non-software companies. So, we're on software, some of the key areas that we are focused on trying to better understand is number one displacement risk by competitors and adjacent vendors. Number two, what is the potential for existing clients to bring this development in-house using AI? And third, what is the vendor pricing model and how will that change over time? Historically, software companies have largely used a seat-based revenue model, and we anticipate that this will change over time and focus more in on success based or usage-based models. As we have reviewed companies and industries at scale, we've noticed a trend where we view more AI risk in companies that focus on marketing, DevOps, customer service, and data analytic applications. For our non-software company frameworks that we use, key views that we focus there on are anything that's repetitive, human centric models. So, think about something that AI can be faster, cheaper, and with the same quality output as a human based model. We also view there to be potential shifts in supply chains and competitive landscapes, and this factors into our framework. Examples of at-risk industries for non-software include different areas within business services as well as data info services. So, I would say that our frameworks help us do three primary things. First of all, it helps us to surface issues that deal teams may not have thought of on their own as they engaged using this framework. Secondly, it helps us to have a consistent approach across AI evaluation for all of the deal teams within Antares. And third, it helps us to make a solid AI risk adjusted decision for both our software and non-software related transactions.

Shannon Fritz
Great, thank you, Phillip. Really highlights the experience, the expertise, the importance of discipline in evaluating these types of risks. Final question to pose today and so I will ask both of you to respond to this. What differentiates Antares in an AI-enabled world? Phillip, we’ll start with you.

Phillip Smith
Antares has been in the market for nearly 30 years, and we evaluate AI as part of the same disciplined approach we used across multiple waves of technology over the past three decades. From the dot com era to evolution of enterprise and cloud adoption, our focus has remained on underwriting durable business models with appropriate capital structures and strong downside protection. Antares has a diversified portfolio of nearly 500 companies, and this provides different diversity at every level. I think it's also important to dive the next layer down. Our software exposure represents about 15% of Antares’ total capital under management and is spread across more than 100 borrowers. These borrowers have strong credit KPIs. First of all, the majority are first lien, senior secured position. As Dan mentioned earlier, they generally have low loan to value ratios and substantial equity cushions, and together that is designed to absorb more meaningful valuation or portfolio volatility for principal at risk, as Dan highlighted earlier. Historically, our software and high-tech exposure has demonstrated stronger performance than the broader portfolio, including lower levels of watch list exposure and non-accruals. Also, as we've evaluated the AI risk within our portfolio, we view there to be low single digit exposure to elevated AI risk, and we will continue to monitor that on a quarterly basis going forward. Lastly, I would just point out that we lead or co-lead more than 90% of the high tech direct lending transactions. This gives us an opportunity to interact directly with sponsors and borrowers on a regular basis, and to potentially identify any potential issues in the future as things continue to evolve and change in the AI world.

Shannon Fritz
Great. Thank you. We'll turn it to Dan. Dan, what differentiates Antares in an AI-enabled world?

Daniel Landis
Our experience. Our portfolio diversity. We also have a dedicated team of software experts that both Phillip and I lead, and we underwrite our software deals, we manage the portfolio, we stay on top of evolving capabilities of AI as well. And, you know, Phillip spoke in depth about our AI risk frameworks, and we've now run our entire portfolio through those frameworks, and our conclusion is that the way that we've historically underwritten software, it puts us in a very good place to manage AI risk.

Shannon Fritz
Thank you, Dan and Phillip. I believe this has brought together a very insightful and informative discussion. Clearly, AI is evolving quickly. We expect that evolution to continue. Our focus at Antares remains the same. For 30 years, we've underwritten evolving risks. We've built portfolios to withstand headwinds and changes, disciplined underwriting, thoughtful portfolio construction, and a structured approach to evaluating technology risk. We believe that experience, diversification, and a consistent credit framework has positioned us and will continue to position us to navigate change responsibly. So, thank you for joining us.