Private Equity Predictions 2026

Death, divorce, disease, depression, the four Ds of lower metal market, deal flow. Now the private equity fund cast is about to begin, so give it up your host, Devon and Jim. Well, who would have predicted we would actually do a prediction show this year, Devon? That's true. Get the couple predictions years. We did. Sure why? There wasn't a lot to talk about. Yeah, nothing's really. Yeah, I think it's really not. It's pretty simple. It's pretty simple. Exactly. But here we are. Here we are. And you're tan resting already for this, right? Rested and ready, but never tan. Yes, true. We're Irish, so this is zero hope. The most things. Bobster is not a tan. So why do we predict things? Do we, yeah, why do people even bother predicting the future, especially in PEG? So we can make money on Kelsey. So the first thing for us is avoiding paying, right? So that's one of the reasons why. Hey, if we know what's going to happen, can we get out of the way of the train coming? Right? To think it's reasonably important. We're in private equity, so the other way is to make money. You know, if we can see a train coming and we can capitalize on it, that's a great thing. If it's not, if it's not all possible. And then we want to try and take advantage of, you know, any opportunities to come from that chain. So sometimes you can get out of the way of the train. Sometimes you can make money. Sometimes it's, you just don't know yet. So it's how do you minimize your risk or how do you stay close to the trend, but not get run over by it? That's really what we're trying to do in private equity, right? And anybody who knows anything about predictions and people who make predictions know that the more you know about something, the worse you are predicting it. Yes, 100%. So to your right, we know nothing about anything. So it's very easy for us to make predictions. Exactly. Well, you know, you think of the sports guys or the famous, uh, in finance, you know, you send out a newsletter to 10,000 people, you know, 5,000. You tell them the price is going to go up. 5,000 tell us going to go down and 5,000 people are going to think you're a genius and 5,000 think you're an idiot. You keep doing that. You might be able to make up money to retire by the time you get down to like the final two. Why are you giving away our secrets to how we attract capital? That's our job here at the private equity fund cask where you know, we're demystified. Oh, right, right. We're peeling the curtain back. I think we should we dig in should we go right into predictions here? I think we've stalled it off on our predictions. I'm going. Let's do it. So what are we predicting? I'm going to focus mostly on private equity in the private equity industry. You're going to focus mostly on, uh, blueie, I would imagine. Yes, well, since they've been a choir, you know, that's, isn't he acquired blueie. So I've got, you know, are they going to give it to the Australian accents? What are they going to do with it? Dad, you know, well, they keep my favorite, my favorite episode, keep you up. I just found out that blueie is a girl two days ago. Really? She's, they were both daughters. That's a, I had no idea. This episode of blueie is called keep you up. Okay. I'm going to do mostly tacking. Okay. All right. So I'm going to, you want to go first? I'll go first. Just like prior years, I have no idea what your predictions are. And you have no idea what mine are. Did you know why that's, that is? Why? Because we did write them down. We used Google docs, which were not really that. So it's two artists who more and try to find each other's docs. Exactly. We realized you were on your Gmail account. I was on my Parker Gale account. And we're really good at technology here. Exactly. So yeah. So hey, let it rip. What's your first prediction? Okay. So my first prediction is, you know, AGI, which is artificial general intelligence. That means it's really thinks like a human. Yeah. Is five to 10 years away. Not some of these trends, like the year of the network, PC network. That was like five or 10 years coming, but it got here. Yeah. I just don't think AGI is anywhere close. So this is like full self driving at Tesla. It's always five to 10 years away. Well, full self driving at Tesla. These are really far. So full self driving and other companies that are good at it, like Waymo, right? But they're getting close. But it's, again, it's not the agent itself isn't making all the decisions, even a waymo. In that case, you've got an AI engine, but the other sensors are not feeding into that same system. So you get cameras that are being read by AI, but you've got other sensors using different prediction models. So why are we 10 years away from AGI and how are you so confident to predict it? So first and foremost, right now, everybody thinks of AI as LLMs, large language models. And they're really just textual predictors. They're really good at conversation. And then we're going to talk about this. This is an unlock. I think the fact that they're really good at interfacing with humans means it's a new interface for traditional systems. If you can talk to the system and it can talk back to you, it's incredibly powerful from a user interface to you. Not teaching people to put their hands on keyboards anymore. The problem is that you've seen it over the last year or two, the more money that gets spent on these LLMs, they're not getting that much better. So it's kind of the 80/20 rule. So the poor billions of dollars into the next version of the model. And it's slightly better, which is going to tie to one of my follow-on predictions. So you have to tie it to other AI technology, which we've been using for years. So if your credit card company calls you with a fraud detection, that is a AI model under the cover is now. Is it an Asian model, is neural network, doesn't matter? It's a very tightly controlled model as compared to an LLM. And you're going to have to bundle all of these things together. So if you're really part of the brain of the LL, part of the brain of the LL, part of the brain of your own, oh, yeah, it's not a human. It's not the pieces of the puzzle. Is it a capability issue? Is it a technology issue and know how issue, what's the bottleneck in HGA? Yes, it's all of those things. And so I think we're going to get better over the next year or two at integrating LLMs and other AI technologies that are more open. So these tight models, like fraud detection, is a very, very tight model. Or if you're using a prediction agent on a piece of machinery and a factory, very, very tight algorithms. So they can't make mistakes because there are no people around. The robot, the robot driller, or the thing making the car, it's doing the work. And it has to do it exactly right every single time. So combining these technologies is going to be somewhat difficult. It'll be powerful, but somewhat different. So when we talk about AGI, we're basically a computer that can think, yeah, it's Lieutenant Commander data. As you can see by Tesla's Optimus robots that fall over, one of the guys taking the mask off, nothing falls off the back. We're years away from this stuff, right? Which ties into my second prediction. At a portfolio company level, bigger models are going to lose their charm because the data underneath the cover's winds, right? So right now people are using a tropics cloud or Gemini or Chatchy PT from OpenAI. And those are very, very large general-purpose models that are really good at interacting with you. And they're really good at digging under the covers and going out and researching other data. But in order for one of our portfolio companies to win or any software company, they're going to have to have tighter models that are very specific to their own domain. So having these large language models that are public, phenomenal from a, hey, will this even work. But at the end of the day, you're going to want to take your own data and build highly, highly customized models. For example, at hybrid, as an example, if you're doing image recognition, it's very specific to auctions. And so using a general-purpose model that could get confused by outside data is not the solution. The solution is a very tight model trained in their own data, which might be used in a level large model. We are now at a point where that's actually possible from a computing standpoint from it. So I'm going to cost you a fortune as a smaller company. So I think you're going to see a lot of the companies that test it on large language models from commercial vendors start moving to their own open source model. And this is a control and cost largely. It's really three things, latency. So how fast is that model working, answering you? Because in a real-time system, you can't wait 30 seconds for this thing to be thinking. Right? Second is cost. So today, all of the commercial vendors lose their shirts on the chat GPT interface that you and I will sit there and bang away into. Right? They make their, you know, they make the money back on the API. You can't really build a system using just their interface. You got to use the API and it gets very, very expensive. So cost becomes a problem, especially when you don't have people in the loop. AI is doing the work for you with your customers. You could start burning millions of dollars in token costs, without even knowing you're doing it. And the third is predictability. Right? If you don't have tight bounds around the engine, it's going to hallucinate because it really wants to make you happy. It really wants to give you an answer. So those are the three things that are really good. So let me try and tighten up that predictions. Sure. Who wins in a world where models get smaller on corporate data? Well, who wins are the individual companies that are, if you're a system of ruins of a provider to that? Well, those are going to mostly be open source models. Okay. So those isn't open AI or, yeah, it's going to be llama or mistrole. It's, you know, there's, you know, anything that you find from hugging face or and in fact, some of the vendors like, you know, Amazon, their bedrock product is specifically designed to, to eliminate the interface between your code and a specific model. And that allows our portfolio companies or anybody to basically use a layer and plug in different models as they go. Amazing. I'm very optimistic about this, but we are years away from, you know, the Lieutenant Commander Data talking robot that does everything for you. Gotcha. All right. Keep rolling. No, no, man, you're up. You're, you're, you're, you're, I, I, are we doing two predictions at a time where we just do, well, you want me to throw all the AI? Because we'll burn through this whole show, much. Right. Uh, you know, let me do that. Track by knuckles. I, you know, good bye, dude. So my first prediction for the overall private equity market is liquidity and grooves. But mostly from the secondary market, secondaries. Uh, so somebody, uh, he won fund buying an interest in existing funds, uh, portfolio companies, maybe one company, maybe multiple companies, maybe the entire fund. Um, so not one private equity firm buying a company from another private equity fund, or a strategic buying a company or an IPO. So think of it still a sticky still too many companies in the private equity world. They've got to find a way to exit, uh, right now, uh, the secondary market, which was kind of a nichey little market. We've done podcasts on this before, uh, secondary providers 10, 15 years ago, we get hung up on. Hey, I've got this fund that buys secondary interests and private equity funds. It's like, click, not interested, right? I buy, improve and sell to another P E fund, or a strategic or I take the company public. I still think the secondary market is going to drive the bulk of the liquidity this year. And I think most of that is front and loaded. Because we get closer to the summer as we get closer. So the predictions coming in and the polls coming in for midterms, everybody hates uncertainty and the world, while it's wildly uncertain today and that's been a big impact on 2025, it's not going to get any more certain from June to November as everybody waits the results of the elect. Well, plus we always know that every investment banker who claims they're working their tail off in the summer is actually, it's caused some board meetings from, you know, let's see school ends in late May, the kids go to camps for the first four weeks and then they're off, you know, from all of July and August and then rolled back in September, right? Yeah. So I think CVs are under pressure. There's right now a lawsuit of an LPs suing to stop a CV from happening. I'm not sure that's a continuation vehicle. Basically, one fund creating another fund for one or several of its assets so they can hold on to it longer. Basically because they really believe in the business more cynically because they can't get the return they want if they took it to market. So they basically sell it to themselves. But I think that still drives liquidity. There's been massive amounts of money plowed into secondary funds. They have huge incentives to go deploy that capital just like a private equity fund has huge incentives to play their own capital. PE funds I do believe in 26 will probably be more active putting money to work and we can talk about that in future predictions. But getting the money back is largely going to be through these nontraditional means. To that point, I do think the AI gold rush starts to thin out, right? So apps without differentiation are going to fall apart. And this is, I think this is going to look exactly like the original internet meltdown, you know, in the 2001 timeframe when you had pets.com and this just a lot of VC money has been chasing variation on AI tooling on AI and you've got three or four really large vendors that have an unbelievable head start. In other businesses, like another trends client service computing or mobile. Okay, maybe Apple had an advantage of mobile. Nobody had an advantage in just web browsing to begin with. But Google opened a Microsoft and Thropic. I mean, they hold the lion share of the data and they've got the dollars to drive these models, right? But below that, all this tooling that are really just features for AI, not products, I think I start to fall apart because those features get built into the core LLMs and other AI kind of, yes, large players. Yes. So they kind of look as a, oh, this is, we're not going to get to that yet. Some app is now filled the void and that the next release of whatever these models will include that feature and kind of wipe somebody out. Yeah, in particular, a lot of times these smaller features are built using AI as well. So they're sort of super fast generated. So you're not going to see a bunch of acquisitions because the larger vendors like, look, I don't need to pay a hundred million bucks to do some kind of aqua hire for small guys building a feature when I can probably that feature in myself for in a month or two. An example of that would be two years ago, you built an image generator off of open AI or some other product and now open as released and Google has released their own image generator off their own models and so all those intergenerators are gone. Yeah, there are going to be no reason for them. Right. So I think you're going to see a lot of these tests things start to fall apart. And again, for us, we have to be careful about letting up portfolio companies adopt a feature set or adopt a new product or module based on these one off technologies that are going to be around because the large corporation buys into some technology or some product from some smaller company regardless of the dollars they spent they don't care. But if it falls apart, they just start again. It's not really going to hurt them. It's going to kill us. Right. So if one of our portfolio companies rolls in, builds in a feature from a very small company that goes away. We're in trouble, particularly because these are technologies have been around for a long time. Even in the case that you get source code from these things, we had an escrow agreement with some of the some of this is going to be in a model that they built that's largely a black box. So knowing that this is going to stop, it's great to test these tools. But we have to be rigorous about, you know, it's the company going to be around. Can we take the risk of putting it in? So you get to the point where, yeah, it's got to be super valuable in order to take the risk of building a small company. Yeah, in my opinion, I would have disagreed with that because if you look at all the software companies that were funded in the ZERP era, the Zero Industry period, when this huge run up in 1819, 20, 21, maybe beyond, I think there's 1200 unicorns or something out there. Companies worth over a billion dollars. But I think really we're better still private. Sure. We're funded evaluations over a billion dollars. There's a lot of chat right now that maybe five percent of those are going to make it. A lot of them are stuck in different ways. But it takes so long for them to go away. These companies have 50, 200, 250 million of revenue. They just been priced out of a good exit. Well those are not where it's up to. Right. But the same things happen in AI, which is the money has flowed in so quickly in such huge amounts. When we were getting started, if you raised 10 million dollars, that was a lot of money. And software was harder to build back then and more expensive to build than it is today. There weren't these vibe coding apps and all these tools out there. Now raising a hundred or a billion, a hundred million or a billion dollars isn't that big of a deal. The difference with the AI companies is it's so expensive for them to do what they do. They burn money so much faster versus a vertical SaaS business that raised $200 million in 2020. Wildly overvalued, maybe growing 10 or 15 percent, which isn't going to get them the return. They thought they were going to get, but they can grind that thing down and live off that $100 million. Maybe forever. Yeah. Where the AI companies can. It's kind of go bigger, go home. Right. And that's the problem is we can't afford them to go under. Like, if a company's wildly funded and isn't really growing, it's not, you know, you put a couple hundred million bucks in it. It's going to find a home, right? See, I'm not worried about the home. I might be worried for my portfolio companies about the contract terms when it gets sold. You know, if something gets sold to Oracle, they jack the price through the moon. You know, they, in the old parlance, they c8 us, they're just jacking up the price. We can't afford that in our portfolio companies. But at least it'll still be there. It gives us a runway to get off the road. But otherwise, I think betting on small companies around the AI edge, I think it's going to be, it's going to be a risk. The other part of that is I think AI starts to become a feature as opposed to companies. It's not a standalone. It's like, how are we using AI technologies in our existing product? Right. Because at the end of the day, I'm not sure it's its own standalone category. It's fundamentally, if you think about it, what is the killer app for AI and LLMs? We've had AI spent around for decades. And nobody really talked about it. Well, the killer app is, you can talk to the computer. It's just like, yeah. Right. And it's really understands English, really understands it. I mean, we've been around a long enough that trying to do simple things like voice recognition was very, very difficult and very, very expensive. All that stuff is super easy now. And you can, you can con, converse with the software. So when you think about an LLM layer on top of the existing product, you're in a world where you hire, let's say you're a manufacturer and you've got an ERP system that you're running. You're hiring, you know, your senior people are retiring. You're hiring younger people that are managing those models, supply chain stuff. A lot easier to opt to the application than it is to try and figure out these interfaces. No, it doesn't work for everything, but it works for a lot of things. And I can say for sure, medical is going to be completely changed over time, even though they're a laggard because when you think about it, you know, medicine starts the historically medicine with a few tests and the doctor asking you a bunch of questions. And at the end of that day, it's, it's rock solid for that kind of stuff. So I think it starts to become a feature to all of our products as opposed to, you know, a lot of standalone companies that are just saying besides the model guys, we're an AI guy. 100% agree. I mean, I'm not cynical about AI at all. We're seeing it and we're using a ton of it. But this was called machine learning 20, 30, 40 years ago. This was called process automation, 20, 15 years ago. This was called robotic process automation five or 10 years ago. And now it's called AI from this agentic AI. I think robotic process is going to, is going to be a problem because with the MCP protocol, the ability to have a large language model, talk to other software systems and coordinate it, you don't really need, you know, something like a Zapier or robotic, you know, RPA kind of functionality. We can do it with the LLMS, which I think is, you know, it's pretty interesting. Highly impactful. We've all played with the agentic. Doesn't work so great when you add that second step in the workflow. I think you're right. It gets built into the product, just like other technologies get built into the product. And we're talking mostly B to B, yep, software, enterprise software, system of record type things. And I don't think you can charge much for it. No, I agree with you. I think it, the customers expect like, no, I'm paying you an annual subscription. You're supposed to use new technologies to make the base product better. Yes. I'll take small price escalators every year, but that's for the benefit of you, you know, cost going up and you investing in the product. That was the deal with SaaS, right? And not buying it once and then just, you know, letting it atrophy. I'm renting it and you're going to improve it. The difference here is, you know, in SaaS, you know, if you had a, you know, multi-tenant solution and it started to get slow. You could spin up another component. in some place, relatively inexpensively. Not necessarily cheap, but relatively expensively. You can lose your shirt implementing a large open AI, a lot, with your product using their APIs. If you don't have some colors, so that's going to be the biggest challenges. It's very expensive. I think the number is-- the predicted number is if you search for something on Google, it takes about 0.003 cost. So call it less than a penny. It's three cents minimum for the same kind of query on AI. So it's 10 times as expensive. So the challenge, of course, is now that your customer could actually burn zillions of your tokens and you don't even know what asking questions. Like the best examples are when you have hackers come in, and they try and steal something. The best guys on YouTube keep the hackers going. They have no intention of doing it over and over. Your customer could do that unintentionally with your support application. So getting the colors around it is one solution. The second part is this is why these open source models are getting so much attention because they don't cost you the $1,000. The real expense is training models, as opposed to just running them. You can run an open source model on your own infrastructure pretty inexpensively. So that's why we're starting to see this. So what is open AI do? There are great test bed. For 20 bucks a month, we can test anything out. That application may cost us hundreds of thousands of dollars a month in production. And I'm just going to go to an open source model that might be good enough. And there's an interesting geopolitical play there too that if China really wanted to mess with America, they could just flood our market with open source LLMs that are really good and just crack the potentially crack the open AI dominance, the cost structure, the investments they're making where and the predictions they're having for revenue. Yeah, keep an eye on that. I think that actually might be not that we're predicting geopolitics. But why wouldn't they? They flooded the world with cheap goods. Yeah. Why wouldn't they flood the world with cheap software? That works. Yeah, so the problem is that that works part is going to be the problem, right? Because it's at the end of the day. And again, this is one of the biggest problems with LLMs and commercial LLMs is they continue to update those models. So that black box is going to change underneath you whether you want it to or not. So that's the problem. If you write a piece of software and it's tightly written, you know what the inputs are and you know what the outputs are. There's nothing that's going to change that. In this case, you're sort of handing it to the, it's Tom Hanks in big, asking the genie a question, you're handing it to that genie and that genie keeps changing. And so that's the difficulty with using an offshore model that you really can't interrogate. That's going to be the challenge. And I think that's again, why these AI systems aren't going to exist. We got to take what LLMs are good at and start to combine them with other tighter technologies and predictors in order to get a solution we can rely on that doesn't change under the covers for us. All right, here's my next sound fundraising. So I think fundraising stays brutal. Sorry, everybody's talking. And it gets even more bifurcated. The barbell effect. Nice use of bifurcated. Well, you know, thank you very much. The barbell effect is here to stay. Meaning the large, large funds, you know, the Walmart's and targets of the financial world, meaning Blackstone KKR Carlyle. Just keep getting bigger and bigger and bigger. They keep adding more and more products. Wealth management channel and investment advisors will have access to those products and more mom and pop investors will have access to those products largely through those very large asset gatherers. That's where most of the money is going. You look at just dollars today and it's been accelerating that way. There's also so much more money in the industry. They're just getting their fair share. So the bulk dollars moving up market are pretty staggering. And then these small little funds that have a niche that are really good at this one little thing. Do something esoteric. Do something in a market that's not so crowded. Have an angle, have an edge. I think they thrive too, but they're running around the world looking for nickels, dimes and quarters to assemble enough capital to go do that. If you're in the middle somewhere and I don't know where the middle is, but call it like that two to four billion dollar fund and you invest in a bunch of different industries, it's your track record from the last fund in the DPI in the last 12 months. That's what I think you how you differentiate yourself to maintain your fund raising edge. It's not, hey, I'm playing the world moving up market and safety and diversification. That's a big guys way. Niche specialists, boutique shops, custom suit makers that have always existed in the investment industry. They thrive, but and they can win on thematic or people just saying, hey, you know what? I'm done with the big guys, I'm coming down market or I have money around the edges of the desk I'm gonna give that to some niche little funds to see if there's some alpha down there. But if you're in between, you're kind of like, well, are you really much better than the big guys that can come down market or have a down market product? Or are you differentiated enough to make a big difference in kind of how you source deals, how you operate deals, access to your network and all those things you can get at a multi-stage, multi-asset type firm. That can make a very compelling pitch for that. So these tweeners, you know, crazy that you could be a couple of, oh, I know, it's a big dollar fund and be a tweener. tweeners, it's hard unless you have outstanding recent returns that look repeatable and have handled succession planning well. Big guys are gonna thrive and there is the wealth management market moves into private equity, they will thrive because of that. And then small people can differentiate. And also, I think you can start funds, which we'll talk about later. The world is looking for niche things. Sure. A certain part of the world is looking for a niche thing. If you're a sovereign wealth fund and need to deploy tens of billions of dollars, you're never coming down market. But if you're a family office that used to invest in the brand names and have gotten disappointed in those brand names ability to generate DPI at the pace you would like to it, you're probably looking down market. And maybe for the first time trying to find these niche little players or maybe kind of going back to your roots to say, hey, you know what? I mean, I did really well when I found that emerging manager who had this angle. And then I stayed with him while they grew. And now maybe there's been some a couple of generational shifts and it's time for me to go look back down at the lower end of the market. So overall fundraising, if you look at the charts of the years, you're going to see a lot of flat bars here for a while until the liquidity is cleared up, which I don't believe is going to come in some huge flood in 2026. So fundraising's still going to be brutal next year. I think historically would you say, would you agree with this that when these large funds were not public companies, you know, they didn't take small investors, wealth mayors have brought involved in them. So a lot of smaller, the middle of the barbell got some dollars because you couldn't get into those bigger funds. Now these bigger funds are not turning anybody away. So I think that's where the challenge is, right? So there's these people, you know, lump in. If you're $100,000 or a best from a well-billion dollar black rock funds, you know, you know, you're not meaningful to them and they simply don't care. Be very meaningful to, you know, a firm that's smaller in the market. Well, I think there are so many more people with $100,000 than their 401K that could be invested in private equity now than there are people with $10 million to invest in black stones fund. So they're just playing the numbers game and it's a smart game to play. We can have a whole other podcast about whether wealth managers should be pushing people 401Ks into a private equity or how that world is even shaping up. But I think that has a diminimous impact in 2026 on fundraising. I think long term, it's a very big trend that largely the large brand name, you know, of asset gatherers will benefit from. Higher hard to see how a 401K dollar winds up in an emerging manager. That seems pretty risky and evening as a small fund, I probably wouldn't advise somebody to make their wealth through that given the potential volatility. Sure, so. All right, what do you got next? Just finishing up the AI trends, just a couple more bulletins, jobs are not going to disappear. They're just going to change the way they do it, right? People have figured out that, yeah. The current state of AI with LLM being, when everybody's talking about, you can't do it without humans. Now we'll get tighter controls around it. Humans going to be part of those. I think we're still going to be in the phase of a bunch of people saying it's useless. And a bunch of people say it's the greatest thing in the world. And it's neither one of those things, right? It's going to be somewhere in the middle. And I still think that's going to take, you know, through 2026 to wash out. Like what is the ROI of any particular use of AI? And so what is the ROI? There's no such thing as the ROI. Is it productive for software developers to use to help generate code? Is it helpful for writers to say that idea is, it's, you know, it's those pieces, right? And so what's the ROI for developing? It's going to vary by organization. I've seen studies that have said, the best use of AI or the most, the strongest ROI is young developers using the LLM to generate code. And I've seen other studies say that's complete, you know, you know, disaster. It's not true. It's the best developers. It makes them 30% or 40% more productive because they don't have to do the kind of, you know, typey typey work that they used. So the answer is, I don't think we know. So, you know, jobs are going to change, but they're not going away. And this is going to be done differently. It's a fun time to be alive. It's a fun time to be alive, man. I think AI becomes plumbing, you know, you're going to have to learn to depend on it, which means The next phase of development are going to be all of the logging companies, data dog, observify. Any of those guys are going to want to get in the game to manage models because now these things are going to be plumbing and in production. When they break, if you have a piece of code that you have today that you wrote that breaks, you're likely, I hope you have, I know, all our portfolios have. We have a really good log of what the hack happened and you fix it. A lot of times it's, hey, we ran out of memory, ran out of disks. We had a bug because some data came through that we weren't expecting a number, a text going to number yield. We shouldn't be doing that, but it happens, right? In this case, you have no idea, especially with a commercial model, it just stops working where you thought it was going to work and it says, oh, because chat GPD is enhanced 5.2 1.7 and now it's answering things differently. We don't know how to clear the memory and the context. So as it becomes plumbing, we're going to start to see a lot of the traditional manage, software management vendors get in this game to figure out, hey, can we instrument these systems so they are reliable and when it breaks, you got some idea how to fix it. So I think it's going to be a big change for a lot of our companies. Smaller software companies don't really have good plumbing to begin with. They're just getting their hands around, sophisticated logging because of expensive vendors are expensive. It's costly to store the data. It's costly to manage the data. So this is a case where, hey, I can even help with that part of it, but these instrumentation vendors, you know, are going to have to adopt this technology and we're going to have to fit it into our systems. So it's plumbing. What do you got next? Last one specifically for our portfolio companies, I think ultimately AI will become an invisible infrastructure within the products. And that's all three phases, right? You're, well, I should say four phases. Once again, we're back to the expansion position. Using it for sales at marketing, like handling customer calls, analyzing the data that comes for those generating marketing materials, back office operations, simplifying, processing of invoices or sending out, you know, bills, whatever you're doing, helping software develop, develop software and then being part of your actual product, right? Those are the sort of four big buckets that we put these things in. In our portfolio, one of 26 is going to do the year of, you've got it in the products and you figure out how you're using it in your engineering organization. I do not want to see piloting AI in 2026. We did that in 2025. Why? Because it's like cloud computing. It's easy. How complicated is it to get started with chat GPD? As long as you got a debit or credit card, that art. You know, and I think that's what allowed people to pilot. But we're out of the pilot phase. I think most customers have said, "Hey, we think this is, we like it." We think this value here, just like mobile devices. But really the take up on this has so much faster. You know, get the exact numbers. But, you know, Netflix took like three or four years to get to some set of millions of customers and, you know, chat GPD did it in like six months or so or a year. Some crazy. There's no friction to the adoption here. So, you know, in our portfolio, it's like we've had enough talk. So we played around with solutions. We know where the pluses of minus art. Now we got to get them into art. We got to fit them into the actual operations and fun to watch how fast the companies we don't actually did it this year. Yeah. So we went from pilot to production pretty quickly in lots of cases. And there's still a lot more to go do. But it's easier to adopt this fall companies right because we don't have the committees. We don't have the politics, right? So a large, you know. I'm sure every large portion 100 C suite is saying we're all in an AI and there's loads of pilots going on at the individual personal level. Probably ones the company itself doesn't even know people trying individual things. But the bridge between what the CEO thinks it can do and what it's actually doing at these large companies is still huge. It's we could be much faster at our stage. Not the politics. We got relatively small teams. We can move faster. All right. So I talked about liquidity. I talked about fundraising now. Let's talk about deals getting done. And I still think the lower end of the market still has lots of activity. Founders still need to sell. People are not getting any younger. They're looking at 2026 and saying, Hey, do I want to hang on for another year? She's time to sell. Should I get out before the midterms or things going to change? I'll argue that 2027 I think will be even better than 26 because I have a sense we're going to have gridlock after November. And what do markets love? What do you mean? I think markets love gridlock. I agree. Nothing happens. And right now nobody knows what's going to happen as John Malini used to say there's a horse in the hospital. And nobody knows how that horse is going to act. And when the horse is out of the hospital, at least the horse is tied up somewhere where you know where he is, you get to actually transact and predict things right now. So these big, big deals, they have lots of global exposure. Who knows what's going to happen next? So everybody's going to hedge hedge hedge who wants to buy the company and the seller's going to say, Hey, I can't take that price. I understand why you're making these assumptions. We disagree. I can't take that price. So the bigger the deal is, the more macro exposure you have. This is a chaotic macro environment. And while sellers want to sell, they're not going to sell at the price they don't want to sell. That's why I think secondaries become the place where a lot of the larger deals transact. So you see deal volume, the dollars will be geared more towards secondaries and CVs and other and a non-traditional liquidity paths. While the number of deals will be largely in the small end. Why? Death, divorce, disease, depression, the four Ds of lower middle market, deal flow. Or kids don't want to run it. You kids don't want to run it. Yeah. It's like, Hey, dad's a big one. I'm working at a hedge fund now. I don't want to buy your little saw. You know, I'm buying lower middle markets offer companies for a private equity firm. I want to come run your lower middle markets offer company for you, dad, which we see all the time. Yeah. So that's never going to change. And individual sellers owners of businesses who are looking to sell are not optimizing for enterprise value generally. But well, they're not optimizing for macro trends either. It's micro trends. Yeah. They are optimizing for how do I want to live my life? How do I spend my day? Do I feel like I've got too many eggs in the basket that my kids want to run this and software companies weren't built to be handed down to somebody's kid. They were built to be sold eventually. Can we just get you know, one thing I get to predict you show, but let's give a little advice here. If you are the founder, selling your company for a good number this year, whatever you do, do not buy a boat. That too many founders, I'm going to buy a boat. It's just a bad decision. I would say buy a boat before you sell it because then you're like, wow, I'm spending all this money on a boat. I need to sell my company to pay for my boat hobby. Well, that's why we send yachting magazines and founders. Yeah, if you they said yachting magazines in any time in the last 20 years, it probably lost from us. It probably came from us. So I think you've got founders don't wait for macro trends, it's all micro. There's an event that has precipitated a reason to sell the business, right? Local lenders, small lenders, structure you can get with founders selling deals, prices you can get with founder selling deals. That's going to keep the engine running. There's 15% of the money chasing 85% of the companies, right? So those 85% of the companies in North America are going to transact, whether the midterms go one way or another, whether the GDP is higher low, whether inflation's up or down. And that's why we live here and we're permanent residents down on the center of the market. But I do think the lower mental market holds up. Again, there will be valuation gaps, there will be challenges, there will be decisions. And they're always on whether should I sell. But buying platforms from families and founders and carve outs at the low end, sub hundred million dollars of enterprise value, it will be an engine. Again, given how the bifurcation of the private equity market is, it gets dwarfed by the big, big dollars getting raised, the big dollars being put to work, the big dollars being liquidated. But most of the volume happens down here. And I think that continues in 2026. So that price arbitrage is the high end though, it's largely a negotiation between people who come from the same market, right? So it's really the founders, investors, whatever going back at point, at our end of the market, a lot of times the founders price in their head has nothing to do with the market market. No, they decided 20 years ago that the business worth, when they get to here, it's worth 50 million bucks. And so they're not going to take less than 50 million bucks because that's very important. They don't have to end up in the market and maybe worth a lot less. Maybe worth a lot more, but that's their number, right? Yeah, totally agree. And again, there's a lot of pent up demand in the low end for selling because, or certainly for private equity owned businesses because of that horse in the hospital, nobody wanted to take a chance on a process in the back half of 2025 because the diligence was going to be really rough. The opposite tariffs, exposure to macro trends, yada yada. Everybody's looking at 2025. Oh, it came in pretty good. Let's launch. Where you were never going to get credit for 2025 until December 31st. So why run a process before then? So I believe the low end, where you don't have as much macro exposure, there'll be a decent amount of volume. I think again in that first half of the year, I think summer gets quiet. I think as there's more certainty in November, there'll be a push. Rapid fire, other cyber security becomes number one and kind of diligence processes. Not because it is important and not because everybody is important. It's super easy to diligence. So we've seen that in every one of our deals. The third party is that come in, they start with cyber because it's easy to find problems. And that gives them plenty of time to start press, but tearing you down on price, you know, while they do the rest of the diligence. So, you know, we started being super tight on cyber security, you know, two years ago, but I think it's going to come to the forefront in terms of, and it's going to affect multiple, so they're going to use it as a way to knock down the price. It's a great, it's a great excuse to cut price that the partner of the private equity firm doesn't understand. But understands it's an existential risk. Yes. I think cloud costs get sober. We went from the Build Your Own Data Center to public clouds to private clouds. I think people are now figuring out that they just can't throw everything in the cloud. Is it local to them? It's the hybrid cloud thing starts to take off me right here in Chicago. The base camp guy is famously ditched all of AWS and built their own servers. I'm not sure that's right from a cost or management standpoint. I think the hybrid model is it, but I think this constantly escalating cloud costs, that's going away. We're starting to re-sync whether they need everything in the cloud, doesn't need to be public cloud, doesn't need to be private cloud. And there are huge advocates on both ends of this thing, but I think people are starting going to start to rationalize this, which is going to put some pressure on the Google clouds, the AWS is of the world, the Microsoft's of the world, because they're going to see workloads potentially start to come down. Great for our gross margins. I'll give this in a couple of the funny categories. Automation, Eats, PE, admin work. But everybody thinks it will. So if you see the number one pitch that we get from anybody is, hey, we're using AI at these five big firms that we don't care about that's cutting all sorts of operational stuff out of these PE firms, which is not exactly the way small PE firms run. We don't see thousands and thousands of Sims, you know, deal books a year across industries, across products, across geographies. Yes, we don't. Right. But I think that's become one of those things. Hey, we could actually cut tons of cost out of it. And I think this is going to be the year of that rationalization, where it becomes a bit of a mess at these larger PE firms, where they think they can just ditch a bunch of their associates and use AI to do all the work. And Sims are going to be automatically analyzed. And you know, you can ask a vendor to find, you know, here's what the kind of deal we're looking for in the scour the market and find you off-market deals that are right in your wheels. Second half. Well, Robert Smith from Vista famously is public saying they're cutting a lot of employees because AI is doing a lot of the administrative and other lower level employee work. Not an inspiring message to send your junior team, right? And I think largely untrue. Maybe they just have a lot of people. I don't know the firm very well, but we've looked at all the products. Yeah. AI products for PE workflows. There's some interesting things happening. 100%. Not significantly more interesting than a pro version of chat GPT, if you know how to prompt. Yep. They seem to be wrappers on a technology rather than game changers. And I do have a sense of like just kind of it's just a reversion to the mean if your AI model is analyzing a deal, it's going to show you the deals that are the most obviously most interesting. Yes. And that's not how the best deals get done. Does everybody know if it's obviously interesting, right? You got to pay the price for that deal or you got to have an angle. If you can teach the machine to look for an angle that you uniquely have based on your expertise, your network and other things, oh, that gets interesting. We haven't found that it can do that. That requires again a human and human. A human knows something. It's not just a, you know, it's not just a, you know, a paper pusher. It has to be someone who understands it in the middle. And I think lots of people are running in that direction. And how you notice it's happening is you get a lot of AI slot from these larger companies. You're getting slop AI slop email messages that look like they've analyzed your website and talked to you, but you can get a feel for it as the same feel to it is the same not snarkiness. It's it's slop. Yeah, but that person we're all in on on AI and playing around. How did we say? Yep. We're using the hell out of it across the portfolio. We're using a lot of it inside the firm. We just haven't been compelled enough to pay a third party who's built a wrapper on top of a public model to read our read a data room and tell us what we should pay for a business. Maybe it gets there and we will be there when it comes. It's allowing us to punch above our weight as a small firm across the board, phenomenal. But not yet third party. Let's pay a few hundred thousand dollars for a product that's going to do something that we want to do ourselves actually. We think we can do it better. Yeah. I mean, I think where will happen is we've got a troll of information, mostly in PDFs. If you look in a private equity firm and you deleted all the PDFs and XLS X spreadsheets, firm would be crypto. It'd be like an atom or hydrogen bomb. Yes, they killed all the. A neutron bomb. Yeah. What's the bomb that kills all the trouble but neutron lands all the buildings. You're trying. Yeah. It would be a neutron bomb. Yeah. So, but the searchability of that stuff when you think of, you know, LLM's bind with this MCP, which allows LMS to talk to other, let's talk to your data, talk to your files. The combination of those two with just a commercial LLM is a great solution to lower middle market firms. We don't need that pipeline thing that isn't going to work with volume. That's not good enough yet. Totally. All right. Now, I'll let you with the, the, you know, the ones we know we're going to see just because these are predictions because we can see it every day, right? Every sim that comes in is going to say AI and it, no matter what, every sim we see is going to have AI. Yes. We've seen this already. It's going to keep happening. You're going to see this pitch of like AI for diligence, you know, with the engine that's just really just offshore analysts and a slack channel. So we're still going to see that stuff with the fake AI, the mechanical Turk. Board decks are going to get longer because they're going to be, you know, generated a lot of, you know, AI content in sheets and they're still not going to get read. Even though we say, please read the board deck before you attend the board meeting, they're still not going to do that. I have a, yeah, generating the board decks, Devon. And then a board was going to AI reading the decks for them. So this is going to get interesting. You say, I, how's that, that looks like looking in the rear view mirror than it does looking, out the windshield. That's already happening all over the place. I would say, advice to us to, to whoever puts the board decks together out there, probably the CFOs don't use AI generated images. I don't need a AI guy at a desk with a, with a Patagonia vest on in the board deck. Just show me the charts with all three fingers on one hand and the spelling is wrong. Yeah, 100%. Yeah. Seriously, stock photography is not expensive. Folks, he's the put that in. Right. That's it. That's my predictions. I think you get amazing. Bring us home, brother. Bring us home. You left me, you let me wrap it up. This is amazing. So here's my software sector prediction is, and I've been saying this for a while. So maybe I'm finally right, which is software is a mature market. It gets invested in as if it's this growth industry that's still chugging along at these amazing compounded growth rates. The world is over software. It's easy to get. It's easy to buy. It's easy to make. It's easy to make. So everybody's got plenty of software. And the last 10 years, so many new companies have been started going after very similar problems. And the big guys are getting aren't losing sight of the small problems that are out there. They're gobbling it all up. So I think in 2026, people finally look at from the investing side is just because this is a software business that's got a great growth rate. It doesn't mean that's going to continue in perpetuity because software is hot and everybody needs software. And everybody had this false positive of COVID, of how much people started spending on software. I think that was a one time event. If you look at it, it's kind of come back and still on the same trajectory similar to kind of like the e-commerce bump and other things, everything kind of reverted back to the trajectory it was on. So I would sort of call it like the consumer peloton directory where everybody bought a peloton when they were home and couldn't go to the gym and then everybody just used it to hang clothes on. So it's a great drying rack. It's a great one. It's a different place. Much better than the Nordic track in my opinion, which is that kind of L shape. You go, oh yeah. And then you have like a treadmill. There isn't a lot of that. There's an other surface area in a treadmill, the Holy Roman and if it's got round edges so you can't nothing. Plus you don't want anything that low. Your dog can get to it. Like it's got to be up. It's got to be elevated. No, peloton's great. One of the great drying devices out there. I have one and there's stuff hanging on it right now. And so I do think that like, Rothe, you look at it like, hey, this is a steady, nice little grower business. It grows more than GDP. Invest in it like it's industrial, consumer or healthcare services, business services. And this, the world wants more predictability than opportunity. Right? Like, oh, this could be great. And I think the people who invest in software, super disciplined, heavy operations, buy it well and operate the heck out of it will thrive. I agree. And there will still be, oh, these guys crushed it on this amazing deal. It didn't look obvious. It was a grower and, you know, and there's massive multiple expansion. But we're back to three yards and a cloud of dust investing. Just good because that's kind of what we do. So it's a self-serving one. And here's my final prediction, Jim. For the people who are watching this, they might have noticed we're three-dimensional now. Yeah, that's true, actually. And they get to see us. I don't see you as three-dimensional. But I trust you. You get to see us rather than just hear us. And there's something afoot and the private equity fund cast. So 2026 aims afoot. This is basically the easiest prediction we've made because it's already happening is that 2026 will be the biggest year for the private equity fund cast since the year we launched it 12 years ago. Yep. So 300 episodes in 12 years, two million downloads. We finally have succumbed to video and we're all in on it. So if you like what you're seeing, you're going to see more of it. You're going to see more people from the firm doing it. We're going to experiment. We're going to play around. We're going to have some fun. But we're going to continue to peel the curtain back on private equity and try and share what we know with the people who are interested to hear it. I love it. Nicely done, brother. Bye for now. Bye for now.