Loading summary
Transcript127 lines
- [00:00]
A
Foreign.
- [00:07]
B
Welcome to the Synopsis, a business and investing podcast. This is take two of. We were already 10 minutes in. We were having a lot of great, great content going on, and Drew had the wrong mic port, and so it sounded horrible. And now Drew's making me redo it, and I'm not really in the mood to redo it. I feel like we already kind of. I gotta make the same jokes. I can't recreate the magic. Drew, I don't know what you want from me.
- [00:30]
A
I don't want anything other than, I guess, the sound to be there, because apparently the people care about the sound more than the content.
- [00:37]
B
And, you know, maybe we can hear about that sound or content. I'm more of a content guy, but the sound bothers me. You need both. It is what it is. You need both. Can't miss one.
- [00:45]
A
I mean, we are a podcast. I feel like the audio should work.
- [00:48]
B
Yeah, it's kind of like, man, the movie was great, but it was just, you know, it really was blurry most of the time. It's kind of tough, right?
- [00:55]
A
All right, so go ahead, make your jokes about how you can't pronounce Atlassian.
- [00:59]
B
Okay, so here's the synopsis of what's going on. In the synopsis, we're talking a little Atlassian, we're talking a little dual lingo, Constellation software, and we're talking a little Warren Buffett, who's still kicking in there. 95 years old, Becky Quick still interviewing him. And I'm worried about Becky Quick's career. For those of you guys who don't know, Becky Quick was the exclusive interviewer of Warren Buffett and Charlie Munger. I. I think that's really the backbone of her career. And as you know, she sadly lost 50% of her interviewing duo here. So, you know, with Charlie Munger's passing, so I'm a little concern about her longevity after, you know, Warren Buffett inevitably goes that direction, but he's 95. He's in good shape. We'll see where it goes. He sounds pretty good on four times speed. You know, he sounds pretty coherent. Anything less than four times, that's too rough.
- [01:43]
A
You're coming out here real hot. You know, you're. You're throwing shade at Becky. You're throwing shade at Buffett. Buffett doesn't need to be listened to any faster than two times. Two times and he sounds good.
- [01:53]
B
And that's it. That is a man blinded by his love for Warren Buffett. The guys lost it a little bit, you know, and listen I've got a special. I've got a special place in my heart for, you know, the original, the original Oracle of omaha. But at 95, I don't know how much he has left to say. I'm going to be honest, that's a hot take. I don't think it's that hot of a take. I think it's a pretty reasonable take.
- [02:14]
A
No comment.
- [02:16]
B
No comment. No comment. And today we're talking about. Let's get into Atlassian a little bit, which we're going to mispronounce five times because Drew's done three things that I don't like here. He's GoT A software, AI software Thematic topic where we can get all these thoughts out. So that's step one. This is the ninth conversation we've had on this. This one's kind of interesting, though. And then secondly, it's a business that I don't think would have made it past any of my initial screens. Forgetting of the AI, it just seems like a. Not a very high quality company. But there's some interesting. I would say I did have some emotional roller coasters because I'd hear a fact and I go, ooh, that piques my interest. Then I'd hear another fact and I go, why do I know anything about this company? So that's step two and step three. It's hard to pronounce. So now through this whole podcast, whole discussion, it's going to be at Lesion. Is it Atlassian? Is it at. You know, it's going to. I'm going to mess it up eight or nine times and, you know, I'm going to get someone in the comments mad at me. So that's what you got going for me today, Drew.
- [03:10]
A
And right off the bat, he's already setting his conditions for failure, saying he's not going to even try to pronounce Atlassian. And then on top of that, you're right, though, that this is a business that was picked not for its business quality, but because it seemed kind of like a crappy business, to be honest. So right now, Speedwell Researches is working on this thematic report, and we're looking at all these different software companies. And the report's basically trying to figure out what software companies are at risk of AI and what is the ramifications of AI overall for software. And I kind of specifically, in order to kind of fill out the research more, wanted to find a software company that I thought was lower quality. And I'm sorry to say Atlassian was the one that I picked And I think even through my research I kind of confirmed that that was the case. And so that was the real reason why I was picked. It's not that there was a choir of people sort of screaming for me to cover Atlassian. There was many, many, many, many other businesses that would have been picked ahead of that. If I let the YouTube audience pick Andy.
- [04:12]
B
And we're going to get into it, but first, what is, and before we start talking about it, because I'll just make this point quickly, which is I don't think this is a bad business because of AI. I think that this is just a bad business for other reasons. But it is interesting in the context of why you chose this as an example of a company that could be disrupted from AI versus all the other software companies. But why don't you get into a quick overview of what Atlassian does? Because I don't think it's a household name by any stretch of the imagination.
- [04:40]
A
Yeah, I mean, first of all, I'll say right off the bat, in some ways it's competitively a challenged business, but in other ways it's a good business as it is a software company with some of the best in class gross margins, 80 plus percent gross margins. You have retention rates that are 98, 99% net revenue retention like 120. So there's a lot of stats, a lot of financial stats that are going for it, other things in competitive factors, less so, which we'll get into. But just kind of a quick overall brief introduction of what Atlassian is, is it's basically work productivity software. So one of their core products is called Jira, that started as basically helping software developers coordinate software development. And so you'll have different tasks to assign within a project, you'll have different coordination. You'll have, you know, basically making sure certain subtasks within a project are getting done, certain things addressed. And so that's their core product called Jira. Over time it moved out from just being on software developers to being more for general business use. Now there is a lot of competition in that segment, a lot of alternatives and competitors there as well. And then they have several other products. Confluence is the name of one Jira Service Management, which is going to be IT Ticket Management, very much a direct competitor to ServiceNow's IT Service Management. They have Trello, they have Loom, some other kind of acquisitions that they've done, but everything is within this family of work productivity software. And it's worth mentioning that this is an older business. It was founded in 2002 so you know, a lot of the software is still on premise. So that's another part of the story is they're still transitioning people from on prem to SaaS, cloud based software basically. And all of this is, you know, kind of just another kind of factor that supports this idea that it is kind of legacy software in a way. A lot of customers complain that JIRA doesn't always work very well. Not a lot has changed in the past decade other than some of the new features they've added, haven't worked very well, some integration issues with some of the other apps and all that. So it's a product that it shouldn't be understated, that it's very hard to rip out and it is critical to the workflows of people who use it every day. So there are software developers, people who are general business employees right now, the splits about 1/3 software developers to general business or intelligent employees or whatever they call it. And it's critical to their workflows. And so you have this whole history of all of your work you've done, you're coordinating across all your teams. And so once you have projects based on there and you're doing it constantly, it's not that expensive, doesn't make that much of a sense to rip it out or replace it. And so that's kind of the reason why even though, you know there are a lot of competitors, a lot of alternatives, if you're already on it, kind of pretty unlikely you're going to switch. It's not that expensive. So you know, they do still have a little bit of pricing power there. But to your point, was there a lot of risk in this business before AI? Yes, there was, but I think AI compounds that.
- [07:38]
B
And yeah, again, initially looking at this business, to me it's just, it seems there's two sides to this, right? Which is there are an immense number of very competent competitors in the space, right? And so I look at it and I go, man, it's hard to win incremental business here. And then you look at it and you go, okay. But it is also on the flip side, very difficult to rip out if you have training and all the systems and all the projects on a certain platform that is difficult to switch. So that's one side and then the second side is I look at, well, it's hard to win market share. But then I go, well, they grew revenue 5x net retentions, you know, over 100%. So not only are they maintaining customers, but those customers are increasing their spend on an annual basis. So then I'm kind of like, okay, well high level, it seems like a bad business. It wouldn't get into my filters because I don't really know what the durable competitive advantages are other than that it's kind of a pain to move. But people aren't exceedingly happy with it. But how unhappy with it. It's kind of the consolation thing, which is it's a mission critical vertically integrated software that is very important to them, not a lot of money. And how much better is the next software company that might be 30% better. Right.
- [08:45]
A
So now here's a great point because we get to talk about the difference between vertical software and horizontal software. And when you're dealing with Constellation software in specific, the verticals are very niche and as a result of that there's not a lot of alternatives that exist. Whereas when you're dealing with horizontal software, the market's much bigger and there's much more reason for a competitor to want to enter that business, which is exactly what's happened. And you know, they have different products and different companies compete across different products. But you know, competitors include Monday.com, asana, which is now my turn to mispronounce a company name, Maybe linear Microsoft ServiceNow notion. A lot of other businesses are competing against their core markets and a lot of probably new startups will continue to as well. And there's more of a possibility of someone at, you know, an enterprise company, something like that, using AI in my opinion to create some sort of in house solution for this because it just doesn't seem like it's going to be that complicated and maybe it's cheap enough that they don't bother to do it, but it just seems kind of possible because then you could customize it exactly to your workflows and get exactly what you want where. Whereas you know, if you're dealing with vertical market software and Constellation software, you can't rip it out because it's mission critical. You have all your data there, you have all your workflows there and whatever customization you would theoretically want, you would just contract Constellation to do that. And they were doing that before, you know, AI Vibe coding was a thing. Constellation software is the professional service revenue which was them going in there and building out revenues for them. And so, you know, the fact that you do for Atlassian, you have the workflows, you have the work history, it makes it a sticky product. But I just don't think many new up and coming, you know, startups are going to gravitate towards it. They have a lot of other options to go towards it. And so that does make it kind of like a legacy business hanging in there, doing the best they can from their existing customer group. It just doesn't seem like it's going to be a real share gainer there.
- [10:42]
B
Well, good point. In my misuse of vertical market, I guess I was using it vertical in the sense that it's a, it's a kind of a narrow, I guess not that narrow. I mean I know a lot of people who use Asana in the project and that is like really their main workflow. I mean a lot of it stems from that. So you're right, vertical was just the wrong word to use there. It's definitely a horizontal platform and again, makes sense they attract more competitors and things of that nature. One thing which I agree with you, when you have a legacy component and you have a lot of new, motivated, well funded startups moving in Asana workday, some of those things you mentioned, but the narrative that it was kind of a melting ice cube to me, which was where immediately my mind went, was negated by the fact that their revenue growth is so significant and has been for many years. So how do you reconcile those two things? And then I'm going to get into the AI argument with you, but first, just on their business level, how do you reconcile the melting ice cube argument with the revenue growth?
- [11:35]
A
Yeah, I mean that is good pushback because they are still getting new customers. It's not like new customer growth is zero. They talk about total number of customers and most recently that was like a mid teens growth rate. If you go back though, you know, to around, let's say pre Covid, you know, that was 40% and a couple of years before that that was, you know, a high 20s percent, a 30s percent. So it's definitely been falling a lot. And then they also have this other number which is number of customers with over $10,000 in ARR. And that's been consistently decreasing as well. You know, that was 20% if you go back a couple years and now it's about 10%. And so it's not like they're not gaining any new customers. But you know, it's a, it's a growing market still for, for software, for work productivity software and all that. I imagine they're losing market share and others are growing faster than them. But a lot of the revenue growth too has just been price increases because it is a pretty sticky product as well. And then they also have had Some acquisitions where they may not have necessarily acquired revenue, but once they acquired those businesses, they had these cross sell opportunities. So it's selling different products, more products to existing customers. So yes, there, there is still some customer growth, but it just hasn't been a ton. And I imagine moving forward it just continues to get harder as well.
- [12:53]
B
I mean, fair, fair, I could see that. I think the AI argument though is still. I'm not, I don't.
- [13:01]
A
Let me give it to you, I'm not seeing it. Yeah. So right now when you're coordinating different project tasks and stuff, you're starting your workflow in Jira and the question is, what if you start that instead in AI? So instead of going into this manual doc manually, doing these subtasks, seeing what other people are working on and all of that, you instead go to your company's AI and you say, hey, like can you pull up information from this project? I'm working on what's remaining to be done. And the AI is the one that captures all of that information and it has that repository and it talks to the other employees and all of that information is being pulled through the AI. So maybe the AI works with Jira and Jira is kind of recording where all of that data is held, but maybe they also don't need JIRA at all and maybe it's just able to do it without that. So that's kind of one form of it is the form function basically is changing the user interface where the customer, where the user interacts with the software, it moves to the AI. And so now it could be an AI talking to someone else's AI. Hey, I completed all of this stuff in this project. As I complete it, just go ahead and continue to alert the other employees in the project. What's remaining to be done? What I've done. Can you pass along these couple questions or something like that? When this other employee goes and sits down for the day and asked about this project, the AI has all this updated information. It's not someone needing to go into jira, type all of this in manually. And so maybe it's a little hard to picture right now, but that's one of the AI risks. There's other AI risks too. One of the bigger ones I think is also that if you are a company like Microsoft and you have copilot, you say, hey, this kind of seems tangential to the thing we're building. Why don't we also go ahead and build something like this or maybe an individual company, a lot of these different software development platforms like ServiceNow, Salesforce and Microsoft, they have the ability for an enterprise to create their own application on their kind of platform. And so I think that's something else that can happen. Maybe it integrates very easily with teams. And so now it's, you know, you're chatting into AI, you could chat to teams and at the same time it's recording and pulling a lot of this information. It's coordinating with Outlook and it's able to say what we're doing, progress has been done on a given project. And so all of these are, they're kind of open ended risk. But it just seems like the vectors of competition could be changing over time and it's just not that like you need Jira and there's no alternatives. It's not that, you know, great of a pool of a product that I just couldn't see workflows changing. And there's different ways around this. And by the way, there's a lot of freemium competition too, which is I guess a little, little tangential to that. But that's another kind of factor floating in there where now you have an AI and you could use like a free tool tool to access that AI.
- [15:42]
B
You know, it's kind of reminding me of, I, I think it was last week where I was kind of like, oh, they're gonna have this AI product. And you're like, what is that? And I'm like, well, they could, you know, and you were kind of pinning me down on that. And it feels like you're kind of making that argument right now, which is,
- [15:55]
A
I feel like I described what it could look like though.
- [15:57]
B
Yeah. But to me it just seems like, why, you know, so you're saying, oh, the, the whole workflow is gonna go away because you'll have this AI agent. I mean, but all this needs to live somewhere and you want it to look up.
- [16:06]
A
Right?
- [16:07]
B
So why would they not layer on an AI, which they're already doing. Like why would they're doing that?
- [16:11]
A
So own AI agent that they're layering on and they'll say, you know, people use the AI agent, ARR is twice as high as those that don't. The question is, how does an employee's interaction with their work change across time? Is every employee going to an individual app and using that individual app's AI agent? Or is there going to be one company level AI agent that orchestrates a lot of this? Or is it going to be Copilot? Is it going to be Claude? Is there going to be one AI Agent that is connecting to all of these other apps and can actually connect to other AI agents. It's capable of doing that as well. I don't know, is what I'm saying. And that's a new risk. I would imagine that if you're thinking of what company is going to really own an AI agent, someone like Jira is just in the weakest position. I see that someone like Microsoft with Copilot a much better position. Even ServiceNow, I think their idea of being, you know, the AI control tower and the platform that makes more sense to me, or a Salesforce, someone like Jira, it just seems to me, is in a weaker position for that. I get that a lot of people do a lot of daily work in that. It just doesn't seem like that's going to be, you know, that AI agent that is going to be that go to one and it's going to then send work to other AI agents. I don't even know if you're capable of doing that right now with their current AI agent, which, by the way, is. Is called Rovo, is capable of doing that. I know, for instance, that, you know, Claude Cowork is capable of plugging into Salesforce API and vice versa. Salesforce's agent Force is capable of calling up Claude. And so these are things that make more sense to me as one possible way you're dealing and interacting with the AI layer. I just. Maybe I'm wrong, but I just don't see Rovo as being like the center of workflows and directing AI work to other sub AI agents. I see them as being more likely to be a sub AI agent, which means there's another layer on top of Atlassian, which means reduce visibility. Reduce visibility, reduce customer interaction. The potential for your workloads to basically be circumvented over time. Maybe you don't actually need Jiro. Maybe it just becomes kind of, you know, a system of record or something like that, where it has a bunch of history, but it becomes used less and less over time as more future work gets stored in, you know, this AI's new AI's memory layer. And so I don't know. This is kind of how I'm describing it.
- [18:28]
B
So here's again breaking this down to me, the AI argument is always complicated because one, you have these software interactions who can just layer on AI. And this is the other thing. You look at the billions of dollars in R and D and stock comp. Right? That's an argument. So we're making the argument that An AI agent, Claude or internal is going to be able to completely replace the entire stack of billions of dollars of software hours. And again, if that's the case or that's the base case of AI, I just feel like the issues in society are so much broader than atlassian being down 20%. Like the AI would have to be so good to complete.
- [19:04]
A
Well, let me put this question to you. Let me put this question to you instead. Maybe this will help. Do you see Atlassian being in a different competitive position than ServiceNow or Salesforce? Because the question you posed is too blanket in my opinion.
- [19:17]
B
Do I see Atlassian in a worse
- [19:20]
A
position, a different competitive position than Salesforce or ServiceNow? Because you're painting the picture like these are all the same software companies and it's the same argument. But Atlassian's in a worse position.
- [19:31]
B
But Atlassian was in a worse position because it's a worse business than those businesses. It's already in a worse position. I just don't.
- [19:37]
A
And you think AI doesn't change that though. There's no significant thing worth pointing out though with this new AI vector.
- [19:44]
B
I think it wasn't a very compelling business before AI. It's not a compelling business after AI. And I'd say yeah, it's at least compelling. It's less compelling now. But I don't think it's like that's the reason it's not compelling as AI. That's not the reason to me, but it's less like. I don't think the narrative should have been oh, Atlassian was an investment, but now because of AI, it's not. To me that's. You get what I'm saying? It doesn't seem like that was the straw.
- [20:09]
A
I don't disagree with that. I think that AI still made it a worse business though. I think that if you were looking at the way the stock price traded and all that before, it did assume a lot of growth in the future since it's been down some 75% now, it trades at an EV to sales of four times. So I guess before that would have been closer to a mid teens to around 20 times EV to sales figure, at least on trailing. So yeah, that did assume a lot of growth and that was probably the wrong valuation just given how competitively challenged this market has been. A lot of new entrants and not a great business in that regard. I hear you there on the other end, you know, you could point to the fact that, you know, it's still at 80% plus gross margins very High retention, high net revenue retention and all of that looked fine. It just seemed like it was mismanaged in terms of the P and L, R and D has always been very high. And I think that still though AI has made this a worse business, nevertheless I think that this is a new competitive vector that challenges them in a new way, has to make them address it in a new way and it makes them a worse business because it's less likely they're able to continue to grow and get that TAM and I know this AI agent thing in way of working, it's still kind of theoretical, it's not here yet. But I do think that there's a real risk that it basically gets subsumed where people are not going directly to a JIRA dashboard to do all this and instead it's happening on the AI agent layer, which makes it easier in the future for people to either switch to someone else or to never have them in the first place. All of that becomes more of a potential over time. If you're interacting with this AI layer and it's pulling all your information, it has all this data and maybe four years from now all of the data and information is already stored at the AI layer and they're like eh, it doesn't really matter what we did that much before four years on jira. And by the way, there's new ways that we could move that data over if we really wanted to save it. And it's just not important to our workflow anymore or the way we're working is changing. And I get that part's a little theoretical and I get your point that it was a bad business before. I still think it's a worse business after AI.
- [22:03]
B
I'll concede that it's worse. It's worse. I'll give it to you that here's my rule of thumb now for AI, which I think is reasonable, which is there's kind of two outcomes here. One is that we reach superhuman level intelligence in that an AI can recode everything Atlassian spent billions of dollars and all of his effort on and that AI can recode that platform and everyone's going to have their own. Okay, that's one outcome. That's a draconian outcome for humanity. I don't know what that looks like, right? That's like superhuman level intelligence at every layer of a engineers and they can build whatever in that world. I don't know, I don't know what we're investing in and all that. The second kind of outcome here is that to me, the software companies that are in trouble are software companies that if I had a full time human sitting next to me who could solve certain problems for me, then that would be a problem. And I think of chegg as a prime example, which is if I had a hyper competent tutor in every subject sitting next to me, what's the point of chegg? It's a homework help. Okay, same thing. I look at QuickBooks as an example of something that I thought would be useless, but as like a CFO or someone entering data. If I had a full time human doing that, and I'm actually the one who needs to has the knowledge of the expenses and needs to categorize and I'm interested in how money is flowing in and out. A full time human is not really going to necessarily help me. I need to sit with the data, I need to look through the profit and losses, I need to look at what statements are coming out and how am I going to do that. QuickBooks is a good example of how I'm going to do that. And I don't necessarily see if I had a full time CFO to the right of me, maybe I'd interact with it on top of QuickBooks, but I wouldn't rip it out. And so to me that's kind of my heuristic about software. Atlassian it seems like if I want to see what Joe's doing, I still want some type of software interface to determine, well, where am I in this project, where is this happening? I don't need to necessarily be talking to someone the whole time. I want, want a platform to see everything and again, whether that's at last seen or something else. But that seems like if humans are involved at all, there needs to be somewhere that this is interacted with.
- [24:04]
A
Yeah, and I agree with that. And it remains to be seen though whether or not these AI agents can spin up dashboards, graphics tables. It may be a recurring user sort of interface exactly the way you want it to look like. And it could just stabilize that and keep showing you the same thing over time. You know you're capable of doing that. And that gets a little bit more into the argument of, you know, kind of vibe coding your own software. But that's still kind of one of the ways I think that they could be at risk I guess, you know, when you are dealing with anything you've used a lot, you get used to the UI and so that's definitely protection against them there. So I think generally speaking it's more a new competitive vector that Makes it harder for them to win new customers, win new market share and also get into new businesses as well. Whereas an old legacy business that is just used to using it, pretty unlikely it switches or changes anytime soon. But it's just a new competitive vector for new businesses, something weighing on growth even more.
- [25:02]
B
And look, I'm with you on that. Like I said, I would concede Atlassian as a worse business after AI or not a worse business necessarily. But this is my other argument. With all these software companies, the distribution outcome has widened and I think that it's widened in a negative way for them and that there are more scenarios in which things don't work out with the advent of AI. But it doesn't mean that their future necessarily is going to be horrible. But I think it's gotten wider the
- [25:30]
A
distribution, yeah, and we keep calling it a bad business, but I feel like there's a few other things, facts we need to surface in order to really make that claim. So I'm going to take over as host and direct this part right here, which is that if you look at their P and L, it is pretty gross. So this is a company that is 24 years old, this is a 24 year old software company and their operating margins are negative. They have a negative 4% operating margin. They were just a little bit profitable if we go back to 2022 and then they haven't been profitable since on a GAAP basis. If you look at R and D as a percent of revenue, it is 50%. Now I don't like to be the guy that's, you know, looking at a company and saying, oh, you're spending too much money on research and all that. But I kind of am going to say that in this case because customer, you know, critiques are that the product has barely changed. So I don't understand how you're spending 50% on R& D. And some of the scuttlebutt we got from AlphaSense with these expert call transcripts was that they for whatever reason got this idea that they needed to hire a lot of people from Google and Meta and Microsoft, like Fang Talent. And as a result of that they were paying a very, very high premium of that. But also the only people that were leav were like not very. The best performers will say. And so they not only paid up a lot for them, it also may not have been the best talent. And either way they're spending half of their revenues on revenue, which is just an insane amount. You had no operating leverage on the R and D line item, basically ever. If we go to 2017, it was still 50%, now it's 52%. So it's jacked up even higher. If we look at marketing and sales, also no negative operating leverage on that line. They went from 21% to 23 from 2017 to the last 12 months. So you would want to see a software company that goes from $600 million in revenue in 2017 to over $6 billion in revenue. So 10x in revenue show at least a little bit of operating leverage. But you didn't get any of that for the business. If you're looking at stock based comp, it is 25% of revenue. So very, very high. So all of those. It just seems like a very poorly managed business from a P and L perspective. I'm not sure what they're spending so much money on R and D for because customers don't seem to be that ent product developments. And on top of that, they've made some dubious acquisitions. Most recently they made an acquisition of a browser company.
- [27:56]
B
But yeah, it is funny because a lot of the software company guys will say, oh, mature margins. And we got it. We've been into this argument before, you and I, about mature margins and whether they ever materialize and a CEO, especially a founder CEO is ever going to admit, oh, we're a mature company now and it's time to show cash flow. Even though in your video, which I think you made a great point point, most other software companies kind of heard the writing and you know, saw the writing on the wall that hey, it's time. 2021 was growth at all costs. 2023, really after the rate hike was show money because people want to see the money now. Now with AI, I guess people don't want to see the money again. They're kind of over it, you know, oh, blow all your money on. As long as you're blowing it on Nvidia chips, I guess that's a fine way to spend money. But that's the only exception.
- [28:40]
A
Yeah, I mean if you're looking at even Salesforce, which for the long time was kind of, you know, the stock the market made fun of for being like an unprofitable mature software company because they're also a pretty old business even they went from, you know, 2022 from 3% margins to now they have 20% margins. So you know, they, they got the note. They've also clamped down on stock based comp as well, you know, and in terms of, you know, stock based comp they went from a company that used to have, you know, stock based comp as a percent of revenue in like the mid teens, which is still by the way lower than where Atlassian is today to like mid single digits, maybe mid to high single digits, something like 6, 7%. So still high but you know, that's progress. Whereas you just didn't see that with Atlassian, despite the fact they had this revenue growth. And so all of that was, you know, kind of problematic on the mature margin discussion. Not to rehash the whole thing, but in theory this is a business that should be able to generate solid GAAP profitability. I don't see any reason why that's not the case. And I think it's because they're kind of of wasting money, if you will, on quote unquote growth expenses that don't actually lead to real growth or that are unnecessary. That's why they're ultimately not profitable. That's, that's my suspicion. Now the browser company acquisition is interesting because it's kind of to me a sort of evidence point that what I was saying about the fear of AI was correct because the reason why they want to own a browser is because they're saying, oh hey, everyone accesses ChatGPT and Claude in, in their, you know, Chrome or Internet Explorer browser. And if we own the browser then we would control that customer interface and we could create an AI browser that is tied to Jira and our other apps and then, you know, enterprises we'd be able to sell these browsers to. And that way we don't have to worry about being displaced by an AI agent which is what their strategy was with that so that, you know, they spent over $600 million on this thing. And I really of don't like making a lot of confident predictions, but I'm pretty confident this is going to be a write off of an acquisition because it just makes no sense to me very simply because this is not something I don't think customers want. I don't think a customer wants a brand new Internet browser. You know, it's notoriously hard to get people to use a different Internet browser. You know, ChatGPT just failed with Atlas. That hadn't gone well at all. And so I don't see that Jira or Atlassian is going to be the one to all of a sudden break that mold. And I don't see people switching because they want to help solve a business problem. Atlassian. And this is kind of getting back to one of the key mistakes that businesses make all the time, which is that they are thinking what helps them and what helps them from their strategic position, from an issue with their business. They're not thinking about what is an issue the consumer needs to solve. They're thinking, what's an issue with our business can we solve? And because of that, and not understanding that distinction, it leads them to bad strategic decisions. And I think that's what this is. They're trying to solve a business problem. They're worried that they're going to lose the direct customer interface. They're worried that they're going to lose their direct customer relationship. And so they bought this browser to try to solve for that. This doesn't solve for a customer problem, though. And I predict it'll be very hard to convince enterprises to try a new browser, not least of which is because of security reasons. But also customers, but also employees are just used to using an individual browser. And by the way, all these different AI features and the ability to have a sort of AI browser and all that, if there's any advantage to that, I think that it's going to be Chrome and Internet Explorer with Microsoft that are going to better be able to embed all of their tools into that browser. And so the idea that somehow this is a big benefit to a customer if you're using Jira through a browser and it's tied together, I just, I don't see it. It doesn't make any sense to me.
- [32:28]
B
You know what I'm seeing? I'm seeing the makings of a portfolio or portfolio that is software companies that have been oversold, not Atlassian, because I don't. I think there's better software companies out there that have been rerated because of AI and then almost another portion of portfolio that is real world. Kind of the anti AI trade. Not the anti AI trade, but just real world stuff. I'm thinking Flor and Decor. I'm talking Copart. I'm talking those types of things. And that's a nice merger. Because if AI goes superhuman, well, we've got some real world stuff. And if AI doesn't go superhuman, still great companies. And then we got the software up. So side. So, you know, this is what I'm seeing here.
- [33:06]
A
Drew, I, I have no idea how you went on that tangent from what I just said, but we'll go.
- [33:10]
B
No, no, this is completely unrelated. I, I heard what you had to say. They're talking browsers, you're talking capital allocation, but this is where my mind is. How are we, how are we going to Take some action.
- [33:19]
A
This is why we're, we're a good duo because instead of working together, we're just off in different directions.
- [33:25]
B
Completely off. I, I listen, I stopped listening to the Atlassian discussion as, as soon as we started having it. So to me, I'm onto the next thing.
- [33:34]
A
All right, well maybe that's a good time to then move on to the next thing.
- [33:38]
B
To move on to the next thing. No, I'm going to wrap this up with Atlassian. I think overall I've been kind of harsh on it. That's unfair. I think that it's a company that's been around since 2002. There are some compelling stats, there are some uncompelling stats. It's trading, as you mentioned, at the valuation. There's a reasonable thesis to be had around Atlassian keeping its place or layering on AI and not being this kind of, you know, complete casualty of this AI transition. Right.
- [34:10]
A
Well, there's actually one other thing I want to bring into discussion, last point, which is that there's CO CEOs of Atlassian and one of the CEOs recently stepped down. It was kind of reported that probably was because of strategic differences, but, but also because apparently one of them bought like $100 million home and then the other one bought, you know, 80 or 100, whatever it was, million dollar home right next to him and then like renovated this, this a hundred million dollar home without like getting his permission or something. And so they fought over this and that was another thing kind of floating around there. And at the same time this CEO who, whose name is Mike Cannonbrooks, he's also been like very kind of active with social issues and stuff. And you know, this is at a time where, you know, the business kind of could use a little more attention. And so that was kind of another weird thing. I didn't like to see, see CO CEOs of over two decades like get in a fight over, you know, who's, what should happen with their hundred million dollar properties that just happen to be next to each other.
- [35:11]
B
Yeah, another kind of layer to that saga. Right. So it's an interesting, it's an interesting company for sure. But let's get into our next topic, which is Duolingo, another, another company where I gotta say, like I just mentioned where if I had a full time, very comprehensive Spanish tutor that could AI straight directly to me that seems better than Duolingo. So in my little heuristic I've developed, Duolingo is not in good shape. But what's Our.
- [35:39]
A
You know what's funny? I've never, I don't actually ever recall you being, like, excited to talk about any of the topics we've talked about on the podcast.
- [35:48]
B
I mean, you know, I get a little. I get excited about Copart. I get excited. We haven't talked about Gary in a while. What's Gary been up to? I mean, where is he on.
- [35:55]
A
That's Gary Friedman of RH for those wondering.
- [35:58]
B
You're right. Thanks for Gary. Where's Gary been? I mean, has the luxury housing market still down or what's up?
- [36:05]
A
That's all you want is unhinged CEOs of real world businesses.
- [36:10]
B
That's what I want. The. I want the Cavaliers. I want the Trailblazers. But okay, Duolingo to me, using my heuristic of. Would I rather have a full time custom Spanish tutor sitting next to me or have Duolingo. I'd rather have my Spanish tutor sitting next to me. So Duolingo.
- [36:27]
A
Yeah.
- [36:27]
B
So that's an awful great company. Using my.
- [36:29]
A
Yeah, that's just an awful take. Awful take. So let's break down. That's just such a bad take. So.
- [36:35]
B
Really?
- [36:36]
A
Yeah, I think that's very bad. I don't think AI is actually a risk to Duolingo, least at all. I don't think that's the reason why they're down.
- [36:43]
B
You and I, we're gonna have to meet outside after this because we're having some huge disagreements that are not getting resolved right here.
- [36:49]
A
Well, you know, I thought that's what the podcast was for, but I guess, I guess we're entering new levels of the synopsis in, in season four.
- [36:57]
B
New levels? Yeah, UFC Ultimate Fighting.
- [37:00]
A
Maybe we could get it outside the White House too. So, yeah, don't agree with that take. And, and here's why. So if you're thinking about what Duolingo is actually serving, I don't think it. Learn an education. I don't think many people download the app with the explicit purpose of I'm going to learn to become fluent in this. This app. I think instead a lot of people, they see the marketing for it. Maybe they want to brush up a little on a language, learn a few words. Maybe they have travel coming up. And so they download the app, they play around with it. It's like a gamified social learning experience. Right? There's all of those elements going on. It's not like you're. You're sitting down with like a book or, or maybe some AI version of this to really learn a Language instead. I want to look at some graphics. I like the noises. I have score streaks, I share them with my friends. It's kind of a fun thing. And by the way, the top 10% of users spend under three minutes a day on the app. So a very, very small portion of your day is spent on that. So you're not going to learn a language or become fluent spending a couple minutes a day yet that's the vast majority of usage. And so I, I think that understanding why users actually use Duolingo is kind of key to this. And I don't think most of them come with very serious intent to learn a language. I think if they do, they kind of learn after some period of time that this isn't going to be the best way to do that. I think instead it is more because it's gamified, there's a social element to it, it's kind of fun and you can feel good. You know, you learn a couple words or something like that and that's your experience. Now that's why don't think AI is actually competitive to it. I don't think this idea that, you know, oh, I'm going to go spend my time in ChatGPT with, you know, talking to AI or something or have it create me a custom course. I think the person that was going to do that was not going to go and think of Duolingo as a replacement. So that's the first thing. The second thing is, yes, AI allows you to create more apps and all that, but there's already a lot of language apps out there. The real problems, distribution. And somehow Duolingo has been able to crack that with their superior marketing strategy, the so called unhinged marketing, which I'll talk about in a moment. Respond.
- [39:05]
B
The unhinged marketing strategy. I just, I fundamentally just don't understand how you're going to make this argument about Atlassian, about, oh, we're going to have a general AI, we're going to have internal spin ups and it's, how can you not see this company getting disrupted? And then Duolingo, which is the most basic app, I just, I don't get it. It just that feels like the app, app that's very easily taken. Oh, it's fine. You know, everyone's got their opinion. Right.
- [39:29]
A
So the thing is that it's not that AI can't make this app, it's more that there's been a lot of alternatives to language learning apps currently. Right. And then the question is, why did one kind of succeed while the others haven't. And once you get that level of distribution, you get the brand and all that, it does become a little bit
- [39:49]
B
of a cycle because you've tried to learn mandate. It hasn't gone well, but you've tried.
- [39:57]
A
This is a contentious podcast. You know what I blame? I blame it being 100 degrees in the valley. That's what I blame it on.
- [40:04]
B
Yeah, there you go.
- [40:05]
A
That's making us feisty.
- [40:07]
B
Yeah, it's making us feisty. But you try to learn Mandarin. Duolingo was an option, but what did you do? You went to hire a Mandarin tutor.
- [40:16]
A
So I used duolingo.
- [40:17]
B
What was the better experience? Mandarin tutor or duolingo?
- [40:22]
A
So this, this makes my point. In my opinion. I thought duolingo would help me learn the language better, and it didn't. It would teach me like a couple random words. And it was. It just wasn't a great way for me to learn. And I, I preferred to use a book in kind of a more traditional means to brush up. And so that, that's to me making my point because I don't think the average person goes to duolingo to become seriously FL Fluent.
- [40:48]
B
They do it as some. For some token. For some token language before they land in their foreign country for their vacation. That's what you're saying?
- [40:55]
A
I. I think there's an element to it where the fact that it's gamified, it's fun. It's a little bit like a game. You. You learn a little bit, which can feel good and can feel rewarding and you get a nice sound when you get something correct. Then you have the score streaks, which is a big reason why people continue to do it every day. They like the score streaks. And I do think there's, you know, they did a good job with these notifications, the apps getting some people to return to it. But the fact that people are spending under three minutes a day on it. The top 10% of people are spending under three minutes a day. Some. Somewhere I've seen even two minutes. This is coming from the rap data that suggests that the vast majority of users are not serious in trying to learn a language. You're not serious trying to learn a language if you're spending under 15 minutes a week on it.
- [41:34]
B
So I. This is where I'm not. So you have a bunch of unserious, unmotivated people as the core user of your apps who, like, how is that a good business? You have. Like we talk about.
- [41:45]
A
I'm not making an argument. It's A good business. I'm making an argument that AI is not going to replace that.
- [41:50]
B
Oh, okay, sorry, we're having nuanced discussions here whether AI. You're right, you're right because your whole
- [41:58]
A
thing is if you have a person next to you that can do it, then it's going to replace it. And I'm saying that that's the wrong model here because people are not using this app in order to, to replace a tutor. They're using it for these other reasons.
- [42:11]
B
Okay, well if we're going to say the distribution curve has increased. Okay, so you can say that, you know, AI can make a human tutor or we have medic glasses that can do live translation or something of that nature, or an AI agent who can do live translation. Would you agree that AI has made Duolingo a worse business?
- [42:30]
A
Again, I just, I don't think that people are, are going there because, oh, I need to learn a language, to learn how to translate something quickly before I go on. This stuff existed for like years. You know, there's the ability for Google Translate. For a long time they had the headphones that would automatically do that. There's, there's portable translators you could get that would translate stuff. You know, I saw those when I was in Japan and those were, some of those are like a decade old. So again, I just don't think that's the core reason why people learn. You know, people learn a language because they want to speak it. They find something self gratifying about it. And those are the reasons why you learn a language. But you don't necessarily go to duolingo to learn a language. You go for kind of these other reasons that, that some of them, they overlap a little bit with wanting to learn a language, but it's just not that serious of, of a person who's there. And I don't mean to offend anyone. This is just based off of my experience and my friends experiences who have used it, where they'll use it for some period of time, they'll pick up a couple dozen words and if they're really serious about learning language, they'll do something else. If they're not, they're going to churn from the app.
- [43:30]
B
So let me ask you the golden question, Drew, which is if you're drawing a distribution curves of outcomes for Duolingo, will you not admit that AI has given some increased outcomes, increased negative outcomes to duolingo or you think there's been no change?
- [43:48]
A
I'm sure there's some, but that to me is not the problem with the business. I think the market narrative and the reason the Stock's been down 75% and a lot of people in the market think it has something to do with AI. I think it's just a more prosaic business problem having to do with the fact that your customer acquisition strategy kind of broke as you churn through a lot of users and then you had existing users leaving and now you need to change your value prop in order to get users to come back. Because you're out of users that have basically never tried the app before and now you need to try to reacquire users that have tried the app before. So you need to change your marketing message, but you also need to change your product. Product. I think this is a more prosaic issue that happens with businesses and it's not the AI related thing. Yes, this is a far out risk, but this hasn't happened yet. Right? And we're already seeing deteriorations in growth. We're already seeing issues with the business today. So it's not the case like when the software companies where, even with Atlassian, you know, you could say they're, they're still growing and all of that. Right now what we've seen is a slowing of growth, a big deceleration where they were growing, you know, bookings 40% a year a couple years ago. Now they're guiding to, to 6% next quarter. And so the question is why that deceleration? I think it's a more prosaic business problem. It doesn't have to do with AI. AI could be a potential future risk, but that's not what we're seeing here now. And it's kind of muddling what's going on when people confuse the two.
- [45:06]
B
This is kind of a funny discussion we're having in general because you're kind of like, oh, it's not a bloodbath because of AI. It's a bloodbath because it's just not a good business. Which is kind of like, oh, they didn't lose the war because the general made a mistake. They lost the war because the lieutenant made it. I don't know, it's what, what's the, what's the point?
- [45:25]
A
I think these things matter. I think these things absolutely matter. Because if you're under the assumption that it's because of AI, then you're stuck disputing the AI risk case and you're totally missing the underlying business deterioration. So I think that trying to understand a reason to attribute to a business's deterioration is absolutely critical, especially if you're counting on a turnaround, because there's probably a lot of people out there right now that are maybe owning the stock thinking, oh, I'm not worried about AI. Hi, well, let's point to this. Are you worried about this, this other thing going on there? And you know, that's not to say it's a good or bad stock or any of that. We could talk more about that. But there is, I think the key issue with Duolingo is not actually being surfaced, which has to do with the fact that their customer acquisition strategy is not working. Their customer acquisition strategy is changing. They talked about the fact that they're not going to do unhinged marketing. They're not going to do that in domestic markets anymore. They kind of give this reason that, oh, it's not exactly what we want our brand to be known as, but at the same time they're still using that strategy internationally. So that does not comport, uh, instead I think it's really because getting someone to download an app on an unhinged sort of marketing strategy will work once, it won't work twice. And so once someone's already downloaded the app and Duolingo didn't really give them the experience they wanted and they churned, they're not going to re download the app because they get an unhinged marketing ad. And so now, now they have to change their message. And when they change their message, they have to change their value prop too, which is exactly what CEO Louis Von Ahn has talked about. He's talked about the fact that they need to actually make it basically a better educational app. And if we are talking about AI, I actually think AI for them could be an enabler of this. And I could see it kind of in a simple way where you know, you're, you're talking to one of these characters or something in the app and it's all AI led. And so for the ability for AI to actually have you converse live in an app, I think is a very cool new feature to actually help teach and educate. They've also talked about how AI has helped the number of courses grow and so a lot more lessons and all of that. So that's one way AI could be helpful. But again, the big theme is that I don't think the key issue with what is going on with Duolingo has anything to do with AI or anything to do with competition that currently exists today. I think it's just, just a more prosaic go to market Strategy issue and value prop. Misalignment.
- [47:44]
B
Well, I see your argument, I do see your argument as to the decline matters, I suppose. I suppose it matters to me. We're going back to Atlassian and then I've already mispronounced it at Legian, whatever it's called. I'm taking the opposite take, which is I think that all the things you mentioned about capital allocation and stock based comps and where's their growth coming from and the competitive dynamic, to me those are again fundamental business risks. More than the AI risk. You're taking more of an AI risk opinion. To me, Duolingo is a little bit of both. But I do think that the overall statement that AI has made Duolingo a worse business. I don't think I would die on the Hill that it hasn't. Obviously you are dying on the Hill, which is fine. That's fine. But I do.
- [48:29]
A
So what's your. I don't see how you make your argument.
- [48:35]
B
No, I, I just like you're making the argument that, oh yeah, well, AI, you can vibe code all these things and there's all these, you know, new, you know.
- [48:41]
A
Are you talking prospectively, like in the future about it becoming a worse business because of AI or currently, as of today, it's a worse business because of AI?
- [48:50]
B
Are you. Well, again, let's go back to that at least Atlassian discussion.
- [48:54]
A
So Atlassian because of AI. Atlassian is a better business today, but I think it's a future risk. I think AIs had no impact on Du Lingo's existing business and negative impact. And it could have had a little bit of a positive impact because it's allowed them to roll out a lot more categories. And I think in the future I don't see it as being a big risk.
- [49:16]
B
That's a take. I disagree with the take, but it's a take.
- [49:19]
A
What's your take?
- [49:21]
B
My take is similar to what I said, which is I think that the outcomes for Duolingo have worsened. I know you're saying, oh well, there's all these other language competitors, but if you look at. At what in the future.
- [49:32]
A
Sure, in the future. I can't. We can't argue that that's a. We don't know what's going to happen in the future, but today. Do you think Duolingo's business model got worse today because of AI? No, but I think the market financials were hit, not. Not what the stock's pricing in, like the actual financials have been hit because of AI, which was what I was saying.
- [49:48]
B
Unless, again, this is where I don't follow duolingo closely enough, but I would assume, unless there's some computation, increase cost from all the AI features that they've had added. If it's a lower margin service, then
- [50:00]
A
I guess it would be. I'm sure it is. I'm sure it's weighed on like, gross profit margins, like by a point or two, but that's very minimal. And then you could push back on that at the other end and say, well, that helped increase retention, they were able to charge more for all of that, et cetera. All right, well, it's just good that you have very, very strong opinions on a business you've self professed and never have looked at. But I could be wrong too.
- [50:24]
B
Well, that's one thing you got to know about me, which is I can have strong opinions about anything with delimited information, but that's what makes me a good host, Drew.
- [50:32]
A
That's. That's what makes you a good host. All right, well, I, I'm going to take over hosting again. I'm going to move on from the Duolingo topic. And let's get to these, these last news items of Waymo, Uber and Buffett.
- [50:45]
B
Yeah, good, good hosting, Drew. Way to, way to take it over for me. I think I needed that. You know, you're right. The heat, the heat's getting to my head. Getting, getting me feisty over here, here. But this Uber news doesn't make it any better, which is. So the news came out. Uber is favoring autonomous vehicle regulations to protect Uber drivers. I don't know what legislator is t. I mean, that I, I can't see a more shameless backing of trying to just entrench your business and regulations. And if that's kind of Uber's, I don't know how much eggs they're throwing in this basket, but the, hey, we gotta, we gotta nationalize the Uber drivers for job protection if that's kind of the route they're going. I mean, I'm feeling, I don't think they're feeling too optimistic about their future if that's. I don't know how many. That's their strategy at the moment.
- [51:31]
A
You know, that was kind of my first take, to be honest, is that if they're supporting this legislation to protect Uber drivers, you know, part of the legislation is you can't like, have an autonomous vehicle without like, someone inside of it or something like that, and obviously not good for av, at least in the, in the short term. Maybe that legislation is, and that was my take is like, it can't be a good thing that they're supporting this. It kind of suggests that maybe they do have a crack or they're worried about AV and all that in their business model. My now current take though is like, if you're Uber, like, why not, like, yeah, why not support that legislation and, you know, push back AV if you can. And if it works, like all the benefit to you, if it doesn't work, then, you know, you'll deal with what's there. And so it does suggest they are, you know, a little worried about autonomous vehicles. For sure. Sure. But the other side of that is like, the status quo is pretty good for them right now. And so if something helps entrench the status quo, then, you know, it makes sense. Why not support it?
- [52:29]
B
I mean, I, I get that too. Yeah, why not Shoot or shoot, right?
- [52:33]
A
Shoot or shoot.
- [52:34]
B
So if it, if it helps, great. I respect that. You know, I think it's prudent of them, again, not giving me a lot of confidence in this narrative that they're going to somehow integrate all of these AI, you know, integrate all the autonomous vehicles onto the Uber app and sit on top of them, which is a future, you know, I, I find to be unlikely. So, yeah, we'll see where that lands with Uber, but not giving me an immense amount of confidence in that business. So another Uber news, since we're hitting this. They acquired Delivery Hero for 15 billion, so that's giving them kind of some access to European food delivery markets. Any takes on that or just what, you know, diving more into food delivery?
- [53:10]
A
Yeah, I mean, you know, it makes sense. Know they need to basically get more volume, more customers, more restaurants onto their Uber Eats platform. I don't know much about the specifics of valuation or anything like that, but, you know, they're trying to catch up to Doordash. There's a lot of other competitors. Markets are consolidating in that I think food delivery is a better business than Ride Hail is just because it's a three side marketplace instead of two and less, you know, risk that there's all of a sudden and become suppliers that become aggregated and have some sort of power leverage, which was the whole AV discussion that we did prior. So really, no notes.
- [53:48]
B
That's the highest praise you could receive from Drew. No notes. That's enough of our Uber Quick update. Let's get into Buffett, who we alluded to. So Buffett kind of spoke about how he was the main perpetrator of the Google investment from Berkshire. So if Buffett was the main perpetrator of the Google perpetrator. The main. Yeah. Is that not right? I think that's not bad. I guess it usually is a negative connotation but the main instigator, individual commits
- [54:16]
A
a crime, illegal act or harmful or immoral deed.
- [54:20]
B
All right, we'll say that that was the wrong word. We'll say main instigator, main decision maker, whatever. He was the main guy behind the Google investment. Does that change anything for you? That this was a Buffet, not a Greg Abel or I guess Todd doesn't work there anymore, right? Or Ted doesn't work there. Which one doesn't work there?
- [54:35]
A
I don't remember.
- [54:36]
B
Do they both not work?
- [54:36]
A
Todd Combs left.
- [54:38]
B
Todd Combs left. Okay, so Todd and Ted, the end of the Todd Combs era. But regardless, does that mean anything incrementally to you that it was Buffett himself that made that investment?
- [54:47]
A
A little, I mean I, I don't know how Greg invests really and it's kind of a little odd that now they have like more capital than ever before and it's going to be someone that we've never seen actually allocate capital. And I, I guess he's been running the business and the arguments, he's been allocating capital organically as he businesses but we've never seen how we invest in investment stakes in public markets. And so it gives me a little more confidence but it was also a little odd. I don't know, it just seemed a little weird to be honest. Buffett usually doesn't buy companies that you know, are trading near all times highs. When he was doing that, you know, the Google stake was already up a lot. It kind of felt like he was sitting on the sidelines a long time and it was just weird timing to all of a sudden be like, okay, they're doing an equity offering, like we'll take a big bite of that. I don't know, that just all seems a little weird to me. And then he's also talking about how they're going to deploy so much capital and he's saying, you know, a great business is one that can deploy a lot of capital at a high rate of return. But if anything is kind of in question right now, I feel like is what is the rate of return of some of this Capex and you know, more so I think for maybe Google than Meta, that's more my opinion. Certainly, you know, Google has been a great business. You know, they have cloud they have, you know, the AI models and they're in a good position and all of that, but I don't know, it's still a lot of money that they're spending and they're spending it at an increasing rate. And it's not that I have anything wrong with someone taking that opinion. It's just a little odd that you're going to say Google wasn't obvious for maybe the past 20 years and then at this moment today, it all of a sudden becomes an obvious thing for you. That's the weird thing to me is I think it was a much cleaner case to be made, you know, 10 years ago or something than it would be today at this, this price and valuation with all of a sudden moving to a very, you know, heavy capex spend.
- [56:32]
B
So you're telling me the man born in 1930 is not necessarily making the most apt technology investments in 2026? I think he had. Again, I don't mean to sound disparaging towards Buffett, obviously he's an icon in the investing community. I shouldn't. But I just feel,
- [56:51]
A
I just, I
- [56:51]
B
don't get, I'm not taking signal, I'm not taking the same signals that I was from him, you know, looking, you know, watching the annual meeting and, you know, the 1998 annual meeting or the 2000, like, he's just not the same. It's just not the same anymore. I don't, I don't know what to say. It's not the same anymore.
- [57:06]
A
I just. For someone who for so long considered technology outside of the circle of competence and that's fine, you know, I can respect that. It just seems weird that in this moment right here, he considers it now to be within his circle of competence. And that to me is just odd because it seemed like, you know, this would have been a better business for him to understand even five years ago, because five years ago he was also talking about how much GEICO spent on search ads and how much, you know, the return on ad spend was for that business and how it was a big mistake, but even back then it wasn't trading at that high of a multiple. And so I don't know why, you know, he passed on it then, but now it's like when they're going into negative free cash flow and they're spending all of this money on capex and AI such a big question mark that all of a sudden it's like, now the future is clear for me. And look, he's, he's the best. He will always be the best investor. I think it's something like, you know, Berkshire could go down 99%. He's still, still beat the S&P 500. So no one can say anything against him. But I don't know, it's a little weird to me.
- [58:05]
B
Say one thing also with the Buffet, you know, Buffett people live by these Buffett axioms. And he will be the first, you know, he'll always say, don't buy commodities. And then that one a. Oh, I spent 200 million on silver. Because I think that a supply and demand mechanic, you know, the supply and demand was off. And so he always does these things, which are, I think people live by what he says a lot more intensely than. Than he does. And so.
- [58:28]
A
Which, by the way, I think is the right thing for him to do. I think, you know, you have general rules, but sometimes, you know, situations change and you need to make exceptions. And I think that's actually the right way to do it. But I don't know, it's a little odd too, because I remember his interview when he's talking about Paramount when they made that investment and someone was asking him on the investment thesis there and the whole, you know, time he's talking about it. It's basically why it was such a bad investment investment, and this is while he owned it, why it could be a bad investment, why competition, you know, is picking up, why it could be a bad industry. Content spends hard customer acquisitions, expensive, yada, yada, yada. And yeah, that ended up being right. And it was just, I don't know, I thought that was weird because he spent so long talking about, like, why it wasn't a good business. And then, you know, I hear him talk about, you know, Google right now where he's saying, you know, great business is one that could deploy a lot of capital at a very high rate of return. And then, you know, of course a lot of people, the question they have is what is that return going to be? We don't really know. You know, historically it's been pretty good, but it seems to me a little bit more skeptical, especially when a lot of the spend a lot of these companies are doing are like, we know, on like kind of artificially priced, you know, memory and semiconductors. And, you know, they're hoping to get a premium in order to continue to sell more compute, you know, to anthropic and to OpenAI. And, uh, they need to buy that compute because they need to continue to generate more revenues and they need to generate more revenues to show more Growth because they're trying to go public because they want to continue to get a higher valuation. But all of this, it's still a little bit of a question mark to me. And I, I fully, you know, it's not like I don't believe in AI or anything like that. I just think when you have multiple companies that very quickly get, you know, trillion plus dollar valuations for, you know, being AI companies, there's still a lot to be proven out in terms of the economics of that and the, that means downstream of that, the infrastructure providers of all of this. There's also a little bit of a question mark in terms of how sustainable their economics are. I don't see how all of these companies can just continue to build out, you know, data centers with basically, you know, off the shelf purchase parts. And not that it's not hard to do it, but there's a lot of players that are able to make data centers and just by virtue of creating one, you're able to get, you know, a pretty high margin business. That just doesn't seem sustainable to me. And so at some point there could be another shoe to drop there. And so, you know, I, I don't talk about really what I own or anything like that, but I'm not negative on Google. I could disclose that I have a position in Google, but that's just kind of a little pushback on kind of what seemed to be a little odd from, from Buffett.
- [60:57]
B
Yeah, I mean, again, I'm, I'm not, I'm done hanging on his every word. That's all I'll say. I still love the man. He'll always have a big place in my heart, but I'm not hanging on his every word. And at this time in, in my, in my career. But I think we, we covered a lot here. Talked a little at Lesion, at Lazy, and I had her mispronounce it two more times before we left. And we did a little Duolingo, little buffet, little update. So I think we'll leave the listeners there and.
- [61:22]
A
Well, this has been a more contentious podcast recording than usual, but, you know, maybe that makes for entertainment. Who knows?
- [61:29]
B
Well, you know, it is what it is. It's 100 degrees, the air conditioning is not working super well. My Internet's cutting in and out. We had to rerecord this. It's been, it's been an uphill battle this episode. So sorry, I got a little testy, but I regret nothing. And Duolingo has a lot of AI concerns and I'll die by that.
- [61:46]
A
All right.
- [61:46]
B
But until next time.
- [61:48]
A
Until next time.