Chit Chat Stocks - Drew Cohen Tells Us Whether Software Stocks Are Dead (Agents, Models, Moats, and More)

Episode Date: September 23, 2026

On this episode of Chit Chat Stocks, Drew Cohen of Speedwell Research returns to tell us about his comprehensive research into software's AI risk. We discuss: 00:00 Introduction 02:11 How AI advancements threaten traditional software companies 04:02 In-house AI development and enterprise software risks 07:05 The outer harness and AI interface evolution in enterprises 11:47 Risks of AI model commodification and competitive dynamics 20:09 Impact of AI on marketplace and ad tech businesses 30:06 Future of Google, Apple, and Microsoft in AI era 40:00 Most at-risk software subsectors and niche markets 49:53 Beneficiaries and winners in the AI-driven software landscape Speedwell's report: https://speedwellresearch.com/companies/ ***************************************************** Subscribe to our newsletter, Emerging Moats: emergingmoats.com  ********************************************************************* Chit Chat Stocks is presented by Interactive Brokers. Get professional pricing, global access, and premier technology with the best brokerage for investors today:  https://www.interactivebrokers.com/  Interactive Brokers is a member of SIPC.  ********************************************************************* Fiscal.ai is building the future of financial data. With custom charts, AI-generated research reports, and endless analytical tools, you can get up to speed on any stock around the globe. All for a reasonable price.  Use our LINK and get 15% off any premium plan: ⁠https://fiscal.ai/chitchat  ********************************************************************* Disclosure: Chit Chat Stocks hosts and guests are not financial advisors, and nothing they say on this show is formal advice or a recommendation. Learn more about your ad choices. Visit megaphone.fm/adchoices

For the past three years, interactive brokers individual clients averaged 24.3% annually beating the S&P 500. Lower costs and access to 170 plus global markets matter. Visit IBKR.com slash performance. Welcome into the Chit Chat Stocks podcast, a podcast to help you find your next great investment. My name is Brett Schaefer, and I'm joined by my co-host, Ryan Henderson. And today we bring back on recurring guest, Drew Cohen, from Speedwell Research, and his namesake YouTube channel, which does long-form research in video and written form.

He and the team at Speedwell have recently released a lengthy report titled Software's AI Risk, which is close to 200 pages covering everything from the broad AI risk to specific company analysis. We're going to talk about all this in the report today. We will have a link to it in the show notes for anyone that wants to check it out. It'll go to the Speedwell Research website. But first, Drew, welcome in.

and let's catch listeners up who haven't been following this story. Why are at least some of the software stocks in either massive drawdowns or seeing a huge narrative that they are AI losers? Yeah. Thank you for having me. And this has been kind of a longgoing story over the past year or so, and it really accelerated sometime around February, when AI really started pushing out a lot more advancements in terms of software development. And so kind of the way the first level thinking goes is that if AI is able to create software very easily, very cheaply, doesn't that mean that these software companies are threatened?

And if that is kind of where you stop your analysis and you think that Claude code is going to get to the point where you could say, hey, can you vibe code me a Salesforce replacement and it's able to just do it. And even if it's not great right now, we all know AI is only going to continue to get better over time. And so eventually it should be able to do it pretty perfectly. Doesn't that mean that Salesforce is done for that their business is totally threatened because you can now get a substitute product very cheaply. And so that understandably freaks a lot of people out, especially if you've seen videos of people coding on AI and being able to create software very quickly. And so the idea that there's now going to be kind of a lower bearing barrier of entries or competition,

all of this is kind of one of the main reasons why AI has sold off. And there's lots of other risk in the model we could kind of get into that too. Are they going to become dependent on these AI model providers? And so, you know, what's the token cost going to be? Is it going to be a higher cost to deliver even if the SaaS companies exist, but they wrap it around with more AI features that are costly so their margins get compressed? Or does it mean, you know, even if these existing software incumbents continue to be there, that there's going to be more competition for future customers.

so their customer acquisition costs go up. And a lot of these software companies, they weren't, you know, that profitable. And, you know, they were growing quickly, but they still needed a lot of growth and future profitability in order to rationalize their valuations. And so this AI moment was a little bit like a coming to Jesus moment that this may never happen, that these software companies may never actually have this long runway of growth and ability to actually become more profitable. And, you know, we've seen since then several SaaS companies really have kind of buckled down all. lot. This is kind of a tangential point, but they have really tried to boost their margins and show profits kind of in response to this, and a lot of their stock prices sagging. Yeah, let's talk through some of the specific threats or I guess AI developments that

people are worried could impact software. And I think there's three specifically that come to mind for me, but there's probably several. Let's start with the in-house concern. This is probably the most talked about point that Claude allows Claude AI in general allows enterprises to build tools in-house instead of relying on software vendors. What do you think about this

concern broadly? And I would say it's been going on for about two to three years now this concern. How have we seen it play out? Yeah. So I think it depends what software specifically we're talking about and whether or not we're looking at enterprises

versus small businesses. Because in the report, I kind of go into this debate of where basically the customer user interface is going to move, because that's going to be a key question, is how do you start your work? And where you start your work, that question is going to get very tied up to who owns the outer harness. So who is, if you're not that familiar with kind of the way AI works is you have a model, and then you have a harness that's kind of wrapped around that model. And the harness, it has a lot of rules in it. It has hard logic in it. Sometimes it has its own kind of mini-L-L-LM within that. And the harness is really designed to get the model to actually do what it's supposed to do, provide safety features, permissioning, governance, and all of that. Whoever kind of is controlling this outer harness of the model, it's going to probably be pretty tied to whoever also gets that ultimate user interface. And so when you're asking this question of, are people going to be able to create their own software. They can currently, but it's going to also

be tied into how they're going to think about it interacting with a lot of their existing software pieces. So if you're an enterprise, for instance, and you're on Salesforce or Service Now, I don't think there's any risk, literally no risk. This is my opinion, that an existing company is going to say, we're going to replace our Salesforce by vibe coding our own version of this or our service now vibe coding our own service of this, and we're going to fully get off of that. I just don't see that happening at all. What I do see happening is that they may say that we're lacking some features from this. And so let maybe us create our own interface. And this interface will interact with Service Now. It'll interact with Salesforce. And both of these companies,

they have APIs, MCPs. They're allowing this interaction. So for these software enterprise companies, it is a little bit like a hedge of not knowing exactly where the future is. And so even if the software work isn't going to be done on our interface, we still want to be involved in that process. And so the way this could look like is if you're picking a company like Nike, let's say, and they are on Salesforce, let's say, internally they may create their own user interface for employees to start their AI work there. And then they have their own sort of model that can dictate where to send the work out to that can say ping the Salesforce API to pull the records. Salesforce is still the system of records, but over time, the actual functioning of the software and what it's doing, it may become less important. On the other hand, though, it's also possible that it just continues to be kind of a pipeline and it goes into the background. So maybe they lose some pricing power. They don't have an ability to roll out new features going forward, but it's still kind of their integral to the work. So there's a lot kind of, I know you asked specifically just on whether

or not the existing companies are going to be able to create software in-house. But that question is so tied to all of these different other elements within a business, whether or not it's going to be able to interact with different software. It's also going to matter whether or not we have a couple or one frontier model that's much better than all of the other alternatives. And whether or not they're kind of allowing API access to that or decide to productize more of their verticals. So right now, though, if you're asking that question and what we've seen, existing enterprise has not moved off of software by and large. Very rare instances of it. We hear a couple, you know, kind of prominent stories, Starbucks for instance. They're trying to create their own internal

CRM system, I believe, and they also had an inventory system. The inventory system ultimately didn't end up working. Part of that was because it used more like AI vision and sensors, and they're saying, oh, it's a different scenario than just vibe-coded software. But we haven't seen any real success of this at like an enterprise level ripping out software, replacing it. If you move down to like SMBs, I feel like there's a lot more experimentation, but it's not like all of a sudden, you know, you're seeing QuickBooks go away either. So I'll kind of leave it right there and we can kind of follow up. Oh, yeah, I loved in your report the existing examples of what has generally not really work too much.

A lot of the experiments and although also historical analogies, I think one, you know, the pitch is for software stocks is that they have high switching costs. an example that always stuck with me is I believe it was listening to the acquired history on AWS where they came out with a competing, I want to say database, something. You know, that's my expertise in software right there that was supposed to replace and compete with Oracle. But it took Amazon 17 years to actually get off Oracle for its internal systems, even though it was selling a competing solution to third parties.

Is that part of the reason here? why it's maybe even if there are some risks, it's been slower than, you know, the perm—not the perma bears, the very bearish people have claimed over the last couple of years. Yeah. I mean, ultimately, I think it helps to just simplify everything in terms of thinking of what benefit does it bring to the company to switch, right? So if we're thinking of benefits, the most top of mind, one for a lot of people, is cost. So if I create it myself, I'm not paying it.

someone else, you know, some of these companies, they're paying millions of dollars a year in software. I think in the Starbucks story, it said they're paying like $400 million a year on software. So if we're able to create it in-house, then won't it be a lot cheaper? Well, there's two sides to this. There's the direct development cost, which maybe they consider to be upfront and non-recurring. So they'll say, yeah, we'll shell out more money up front, but then we own it and we don't have to develop it anymore. The problem with this is going to be twofold. One, they're going to find out that the software doesn't work exactly the way they intended, which is we're already seeing this with a lot of AI-created software.

there's still more bugs. It takes someone to upkeep it and maintain it and all that. The second thing is you're going to need someone who is constantly updating it basically for new features, who's performing maintenance on it, security updates, all of this. And so then you're going to have to have an internal team that's dedicated just to this software development. Then it ultimately gets to the point that how can it possibly be the case that an internal team of 10 people, whatever it is, is able to more efficiently create software and maintain it in-house, then a company specializing just in that and then delivering it to a lot of other companies. And that's why companies specialize in things initially anyway. And the Starbucks story, the way it's pointing to it, kind of already seems like they're running into some headaches trying to develop their own in-house software. And there's other stories of people that have done something similar. And then they find out it's actually a lot more expensive and harder than they expect. Getting it integrated with all these different databases, all these existing workflows, all these existing applications, getting the permissioning, the governance

layer, and all of that. And so the hope is that these things, the hope I would say, I don't want to say hope. I would say the people that believe software is going to disappear are the people that believe all of these things are just going to become better and better over time and really not issues and you're going to be able to get pretty close to a one shot. That's not where we are today. And if we did get to that point, then I would imagine you would see that the software companies, they probably would lose just a lot of pricing power. But I still don't think that they would go away. There would be other software companies that would come in their place in the future that would be able to offer these services externally and for cheaper because it just doesn't make sense for every

business to specialize in all of these different areas within the department. That's why business is modularized out in the first place is because it's more efficient to outsource certain things to other businesses. And so then it becomes a question of whether or not they need. need something to be that customized, that it makes sense for them to keep it in-house. But even then, all of these different software platforms like ServiceNow, Salesforce, they all have the ability to customize within it already. So this is kind of a long way of me getting back to the point that I don't think that companies are really going to want to maintain a lot of their own software in-house on the enterprise level. There's one exception to this that I want to talk

about having to do with the outer harness. But I think like the core software systems, I just don't really see it. Why don't we talk about this outer harness now? What is the one risk from, you know, I should mention again, you and the team spent a lot of time into this research. You weren't just reading a few tweets. The report is very lengthy. What is the one risk you kind of saw that actually could materialize here? So to me, the big debate is where is AI going to ultimately be used by an individual, by a business? And if you think about it, there's kind of multiple different places where AI is being infused. If we're thinking of a typical business right now, you could see that on the operating system level, they usually have co-pilot.

That's one AI layer that's being infused. On the application layer, if you're using Salesforce, or you're using Atlassian, even Adobe, any of these businesses, they all now infuse AI within it, right? So that's at the application layer. You also see that there is going to be AI. Then there's also the AI models. Do you go directly to Claude to use your AI? Do you go directly to chat GPT? So that's three different layers of AI that currently exist right now. Now, if you are an enterprise business and you want your employees using AI, but you want to be able to sort of standardize it, systematize it, make sure that these AI agents are doing things that they're supposed to do, what I think is going to end up happening for a lot of enterprises is that they're going to want to

create their own sort of custom AI interface and become the ultimate outer harness. And, you know, the cloud infrastructure providers like, you know, Amazon with Bedrock, Microsoft with Foundry, they provide tools that help enterprises do this already. And I think this might be the way a lot of AI work goes because a business doesn't want to just use Claude, because then all a sudden they're reliant on Claude. They're just using one model. They're beholden to the cost of that model of Claude. It's a little messy using AI in every individual application, individually permissioning it. It would make more sense if there was one central place, you could start this work, and then you could say, I want to do, you know, I want to look up sales records for this one

individual client, and then it can go ahead and decide, recruit other subagents, pull the data from different software rather than you individually going into each software app, right? So it's changing kind of the interface of where the work starts. And then the there is the OS layer, which would be like using copilot. I think some businesses will go that route just because it's the path of least resistance. It already ties into like the Microsoft ecosystem, which has its own, you know, permission and governance, all that with Entra identity. And so it's kind of a path of least resistance, but I just don't think people like Microsoft products that much. I don't think that the co-pilot product is that great, that a business is going to say this is

going to be our full like AI strategy. And so I think what could end up happening is if for a lot of enterprises, they're going to create their own sort of dashboard interface, their own internal software that this is going to be our AI. This is going to be the way that we use it and you start your work there. Now, it's not impossible to say that maybe Microsoft is the person that's actually serving that on behalf of the client, but it's going to be the client that kind of owns that outer wrapper and can customize it more. So they have the outer hardness. And then within that, it's recruiting all of these other AI models individually. And so this tool also is going to be dependent on how commoditized AI can be, which could be a future topic we talk about.

You research your investments, you analyze markets, you manage risk, but have you researched your broker? For the past three years, interactive brokers, individual clients averaged an annual return of 24.3% compared to 23.1% on the S&P 500. IBKR's lower trading costs, competitive rates, efficient execution, and access to 170-plus global market. markets help investors keep more of what they earn and put more capital to work. Over time, the broker you choose matters. Interactive brokers member SIPC. If you care about performance, find out why the best informed investors choose interactive brokers at IBCR.com slash performance. Again, visit Ibkr.com slash.

Okay. When I sell my business, I want the best tax and investment advice. I want to help my kids and I want to give back to the community. Ooh, then it's the vacation of a life. I wonder if my head of office has a forever setter. An IG Private Wealth Advisor creates the clarity you need with plans that harmonize your business, your family, and your dreams. Get financial advice that puts you at the center. Find your advisor at IGPrivatewealth.com. Yeah, I guess a couple things. First of all, I think right now there's the discourse going on about the frontier models needing to slow down development.

I thought your comment was funny around co-pilot because Microsoft came out and, like, yeah, we agree. Everyone's like, what, you haven't already been slowing down? Everyone's got to slow down. Yeah. The other part, going back to your earlier point, I think the whole in-house argument has never really made a whole lot of sense to me. Because I think the biggest myth about this whole thing is that software is anywhere close to set it and forget it. Like anyone that's ever worked in the software organization knows that every day you're coming across bugs. Every day you're coming across customer issues trying to solve specific use cases.

And it's just like the whole idea of one-shodding something and having to just persist forever and be a useful replacement seems far-fetched to me. I do want to talk about another angle. And I think this was primarily, primarily became a concern after the Citrini report a while back. around agents performing sort of replacing the consumption habits of a user. So one of the ones that was used as an example was Airbnb. And potentially instead of going to Airbnb and looking for alternative accommodations, Ryan, I have my Ryan travel agent go and find the lowest cost possible stay for these specific things I'm looking for. You know, it has to have this kind of Wi-Fi, this many bedrooms.

and it's going to scour the entire internet, not reference Airbnb or maybe try to avoid Airbnb, whatever, and that these consumption habits would hurt the aggregators. What do you think of that risk? And do you think this could have a real impact on some of those Airbnb-like businesses? Yeah. So I think that was an awful example he gave because the AI agent risk is I see it. And we could talk specifically about the aggregators. but as we're talking about, you know, software, kind of more enterprise software as a whole,

is goes back to the question of where the work starts. And this gets back to this question of whether or not you're interacting directly with software or it kind of becomes something that's happening in the background. And the AI agent is recruiting the software on your behalf. So I'll get to what you were asking more directly. But just to kind of loop back to what we were saying earlier, if you do have a Salesforce platform and you do have all your data on there, what will that could end up happening is the AI agent, it basically goes to Salesforce, it uses that API. It tells Salesforce what to do. Salesforce already has its own logic, deterministic logic and all that. So it knows how to calculate things to pull different sales reports. It has all of the data already

secured in there. And so it's the AI agent using Salesforce and then delivering that back to the user. Now, the question two is going to be, what does that interface look like? Is that user then going to be interacting with Salesforce through a sales force? Salesforce AI interface. So, you know, Salesforce does provide that service now provides their own AI interface. It looks a lot like chat GPT, very similar. Or instead, is it going to be a company own sort of interface? Or instead, is it going to be directly on cloud or something like that? Because you can connect Salesforce to cloud. So in this circumstance, it is still the AI agent that is kind of disintermediating a little bit, this interface, but it's still critical to actually

getting the work done. So the risk there is that over time, what ends up happening is that as you keep doing these different queries, sales data polls and all that, more and more data maybe gets stored in the AI model and it doesn't actually need to recruit out Salesforce. Then over time, this AI model company like Anthropic could say, hey, we're going to build out our own CRM2. And it could start to try to shift work into its own internal tool rather than relying on Salesforce. In addition to this, Salesforce is now not in a good position to launch any new features because it's kind of just this API. So any new features, new markets, growth, and all of that, it's now kind of being

disintermediated. It doesn't have this direct user interface that could say, hey, go ahead and don't just use us for sales. You could use us for IT ticketing or something like that. Now instead it is this model company that is in front of that, that has a better position to push new services. So getting back to how you started this question, though, on, you know, know, aggregators and whether or not they're kind of screwed because of AI agents. And I said

the Airbnb was an awful example because one, Airbnb has a lot of exclusive inventory and an AI agent can't crawl Airbnb without Airbnb's permission. So the way that this is more likely going to work in the future. And Brian Chesky, CEO of Airbnb has talked about this, is that they are going to have the AI within their app that does this. So you go directly to Airbnb and you say, hey, I'm looking for you give the prompt that you gave, and then Airbnb is going to be the one to surface this. Because if you go to chat GPT, they don't have access to Airbnb. And if they did get access to Airbnb, it would only be because Airbnb allowed them to because they wanted them to get access to their exclusive inventory in that case. Now, maybe you could get a little farther out and

say, oh, well, you know, a lot of these people that list on Airbnb, they, a lot of times they do a list, they'll have a property management company. What if they create an MCP, that then gets exposed and then these different, you know, Claude ChatGPT can actually go in and pull from that. And then you have the listing right there and you could book it right there. And that's true. And that is kind of a risk, except it's still,

you're going to want, if you're, you know, booking an Airbnb, a random hotel, you're still going to want to know that Airbnb stands behind it, that you have the support that stands behind it, that you have the reviews,

the ratings, all of that. So it's still going to, you're going to want that kind of trust layer that Airbnb provides. For that same reason, I don't think that AI agents are going to be that disruptive to Amazon. Because if you imagine you are going to chat GPT to look for stuff, and it shows you like eight different options, and I've actually done this before where I've tried to buy something from chat GPT, and it brings up all these results. It's all these websites you've never heard of. You've never heard of them.

You have no idea what they are. And then you're going to give them your money. And you have no idea what customer service is like, even if it says delivery is quick. If it doesn't come, what do you do? it's just really annoying. And then I would ask chat GPT, can you find this on Amazon? And right now it can't really do that that well, because it's not integrated too well. It gives you listings that like, you're like, this doesn't look quite right. But let's say that it did integrate very well. And you could ask an AI agent. I would still want it to shop on Amazon for me specifically because I know that if there's

anything wrong with the item, I could just get it returned. I never have to worry about the shipping. I know where everything is coming from. There's a certain trust layer that they've already provided as a result of that. So that's why this AI agent disinger mediation, I don't think it's going to actually be, you know, that worrisome for platforms like Airbnb, Amazon. I don't think so. I think that people will use AI agents to book these things, absolutely. But the reason why you go to Amazon and Airbnb is more than just, hey, can I save like 20 seconds, you know, searching for something. Plus, Airbnb specifically, you actually want to look at property photos and compare reviews and stuff. Where I think it could get more disruptive is something like Uber.

where now if an AI agent can book an Uber for you, price check between Uber, Lyft, now maybe Waymo and Tesla's, you know, a robo cab or taxi or whatever it's called. And it's price checking these four. Cyber cab, there we go. And that, to me, it seems more disruptive because you don't have any particular loyalty to needing an Uber versus a Waymo or a Lyft. You don't really care. You just want the quickest one, the cheapest one, and all that. Maybe if you're in a loyalty program, you care a little bit, but a lot of people aren't.

and so it could still be pretty disruptive, plus the fact, and I guess this does also weigh on Amazon to a certain extent too, is that you get disintermediated from the advertising portion. And so Uber, a lot of their kind of operating margin leverage has come recently from having a lot of ads shown within the Uber app. And so that's only possible if people are opening up your app. If AI agents are disintermediating that because they're just calling RideHale through an API and then creating a custom. dashboard, kind of like Apple does right now, when you pull out your phone and you call a car, has that little new custom dashboard. They could just create something like that on the fly for their own version of ride hail. And so you're cutting up that ad revenue. That to me, by the way, is one of the biggest and most understated risk across all marketplaces. And I said this as early as

like when I wrote my co-star report, because they have a company called Apartments.com. That's like one of the most popular places for people to find apartments. But they make all their money through sponsored ads, sponsored listings. And what's going to go away if all of a sudden you have an AI agent doing your searching for you and it could search 200 pages is that it's not going to care whether or not something's a sponsored listing. It's just going to ignore it and it doesn't matter that it's top of mind because it's not lazy like us and it's willing to look through 100 pages anyway. And so a lot of that, a lot of these platforms that are monetizing with sponsored ads

and are not going to be able to command a user to directly go there, I think that's going to be problematic for them. You know, in Amazon's case, though, if Amazon has a decent enough AI chat box that I could just type in there and it gives me recommendations, I'll just go to that rather than chat GPT. And so if chat GPT doesn't get access to Amazon, or maybe they only get it on the condition that you provide multiple listings with one being a sponsored ad, something kind of like how Google or Gemini is showing up right now, then that could be a way they kind of remedy that. But there's going to be a lot of businesses.

Apartments.com specifically I'm a little more worried about because there's a lot of of different places you can pull a listing of a rental unit for, and it's not proprietary inventory. And so that's why it's going to be harder for them to say, you know, demand people to go to Apartments.com. I think I just talked for like eight minutes straight on it. So go ahead. That was wonderful. That was wonderful. Yeah, I was trying to think about follow up here, but I think for consumers, and it's different than enterprises, if something can either save you a lot of time that's just busy work, for example, hunting for an apartment or hunting for a place to rent or buy. It takes a lot of time and you can maybe use some of these AI agents to search

on Zillow Apartments.com for you or something that you don't want to that you have to buy, such as a plane ticket or maybe a train ticket where you had to fill in these forms. If you want to do it online, you don't know if you're doing it right. Is this, this is going to take me 10 minutes? That could technically be used by the AI agents. And I think it will. I've tested with one of them. I bought a train ticket. We'll see if it works. If it totally botches it. With chat chippy T or? I used the one that we were talking about before the show, instinct, and it seemed to work.

I got an email receipt. You have to give it access. Of course, the access piece is interesting. Where like someone like Amazon. It's all your data forever in perpetuity. Yeah. And whether some of these platforms are going to allow agents to log in for them, but I don't know, maybe, maybe the long term they do.

But I kind of think this may be, and you can correct me if it wrong, could lead to an increasing worry for Google because generally in the past, you know, if I was a need to train ticket in Europe, you would type that into Google search. But if you have an AI agent do that for you, are they going to, you know, click on sponsored links and things like that? That's kind of what I'm trying to get up with as follow up here. Is the growth of agents from meta, from Gemini, well, that's part of Google, but from Apple, Siri, from all these other ones, open AI cloud, of course, does that increase the bear case on, you know, Google? I think so. I think it's, to me, I think it's pretty obvious that Google's search business, if we're looking at over 10 years, it's going to be pressured. Now, they have a very natural kind of transition to people using Gemini and more AI instead, where they could show sponsored ads. And so they could turn out to be just fine. But it seems like it's kind of obvious to me that people are going to be doing traditional search less. They're going to either be going directly to an AI chatbot or, you know, with the Surrey update and more. And more. It seems like it's kind of obvious to me that people are going to be doing traditional search less. They're going to be going to be going to be going to be going. Or, you know,

AI agents, it's going to be a lot more voice, especially when you have AR, VR, VR glasses, the new AirPods are going to have AI capabilities in them. So this is replacing a lot of the things we used to use search for. And, you know, of course, Google search, you know, 90% plus market share. So this is going to be disruptive to them because the less times you're going to Google search, the less times they are showing keyword searches, the less clicks those get. And the less this is, you know, a business for them. Now, then the question is, okay, well, are people going to then transition more to AI work yet. It seems like that. They own Android. They could have their own sort of AI assistant within Android that is then commandeering a lot of this activity that used

to go to Google search. So they could be fine in that respect. And a lot of people do, you know, still use Gemini. They use a lot of other Google products that have AI infused in them. But if you are thinking of the most monetizable searches, which is, you know, I want to buy this item. I'm looking for running shoes. You know, what's the best running shoe? I think a lot of that is going to AI and it's not all going to Gemini. I think it's getting split across all of these different providers. And now, you know, even Muses is kind of coming up in there, which is Meadows. And that's had a pretty good launch and is good specifically for shopping.

And so to me, that's always kind of been a scary headwind for Google is that with the search business. Now, on the other side of that, you know, their cloud business has been absolutely booming. You know, YouTube is, you know, totally dominant. very hard to see them ever getting displaced. So they have a lot of other kind of businesses within that. And then, you know, can they roll out something with agentic AI and, you know, tying their AI intelligence to their cloud service? And can that more than replace a lot of this Google search? I think that that's possible. But to me, I think it's people's eyes are a little bit closed on what's going to happen with Google search in the future just because it has been good so far. Right. The first thing that happens when you have better AI is it makes all everything better, right?

It makes the searches better. It makes the ad matching better. So now people are clicking more ads because they're more relevant. And it helps them with generative ad creative at the same time. And so I think that's why you've actually seen, you know, AI search actually be very strong. But eventually, once the user habits change, you're just not going to be able to monetize that the same way. So I don't know. Like if you can imagine we have, you know, a new iPhone in a few years that actually is really good Siri that's all AI.

Maybe it is powered by Gemini. I know they have an agreement and all that, but they're not getting paid the same way. They're paying them a billion dollars right now for the rights to use that versus how much they're monetizing on search. So many folds of that. So to me, it seems like in the future more kind of the search exploration activity. It starts with these AI agents directly. And I think that this is moving the touchpoint away from search. And it is disruptive to search.

I think search activity drops a good amount. yeah the it seems like apple and and alphabet slash google have the the best right to win you however you want to call it to build the best agent uh because they have that at least for consumers but i i have been disappointed in kind of their product roll out so far i will mention for the profitability on google search i think i read one time that the most profitable searches are for car insurance might have been home insurance but i think it's car insurance providers So that is another example, I think, of what an agent can do really well with. Those are all confusing on what is exactly the best one.

You kind of go, hey, search around for my specific use case, which one is the best provider and you're not going through Google search. But there's not a Google pie or one more fall up there, Drew. Yeah. I mean, to me, that's such an interesting thing because then the AI models, unless they're like taking money to recommend certain things, I don't see how they monetize that search. Because if you're asking an AI model, what are the best insurances I can have? you want the information to not be tainted by them collecting any money.

So maybe they give you three options and then there's a fourth one that's sponsored or something like that, but it's clearly the worst one. I just, I don't really see how that model works the same way. Whereas before it's like you're searching and it's just pushing up, you know, different links to you at the top. Maybe for like something like, you know, you're searching running shoes and it says this one's the best and then it could have a sponsored out of the retailer that sells that shoe.

It could work in that case. but even then it should be in your interest. And so I know that it's going to be hard to make AI really expand a lot and get huge usage, huge usage on a subscription model. But that to me makes more sense. Like I don't want my AI collecting ad money and like recommending me stuff based off of the ad money it's collecting. I agree. I'd rather pay 20 something bucks a month.

And if they can save you so much time in a week or a month, I'd be willing to pay even more. But this is not just a Google podcast with many other things. to discuss. There's another risk you outlined here. I maybe combine, you talked about this one a little bit, but I'll combine this with any other risks that you think are important to highlight. I think you had eight, maybe seven or eight within the report.

What about basically AI giving the existing software companies and startups much easier capabilities to say, hey, we can 10x our product suite overlap with everyone else and that drives down increases competition

and drives down, let's say, prescription prices or however it's charged. You're saying an existing incumbent does this, or the AI model company? Existing incumbents, not, let's exclude the AI models. Maybe that can be mentioned too, but the existing incumbents or say any startup in the sector. Yeah, this to me is a key risk and it's something that's already happening where all of a sudden it's a lot easier to make software. It means it's a lot easier to expand out into adjacent verticals.

So if you're looking at service now that started an IT, Now they're moving into CRM. Salesforce started in CRM. Now they have an ITSM product. You know, there's examples in the report of Atlassian expanding out into different adjacent verticals as well. So this is something I think is the most likely to happen. And it is already happening because they already have a direct connection with a customer.

They have a paying customer. And so they're going to this customer and they're saying, hey, we could do more stuff for you. And we know software is pretty cheap to deliver. And so if they're able to roll out these new products, new modules, they could undercut the existing player, especially if those much cheaper development costs. Because a lot of the hard part in getting a software customer, in getting a software customer is that initial distribution. Because you have to go find the customers, you have to onboard them, you have to get them familiar with your offering. But once you're already there and you know them, it becomes easier relatively to get them to just

add more features. Oh, you know, we already are synced up to all of this. We already have, you know, data on all of the sales that you're doing. Why don't you hire us to also do more marketing for you and we could connect this to analytics and we could continue to just do more stuff for you. So it becomes a little easier to continue to move out into adjacent area. So I think that that actually is going to be a very formidable source of competition. I think in terms of startups doing this, less so just because it's still very hard to get that initial entrance into a company.

because if you're thinking about the P&L, all of what AI does is it lowers the R&D costs, but it's not lowering really the S&M cost. It still is pretty expensive to go out and find all these customers. And you might be thinking, well, we could use AI. Like, who wants to talk to an AI voice bot? Like, you're not going to pick that call up. That's not getting a sale. And so it lowers only a portion of the cost to actually get this product out in front of a customer.

And, you know, this idea that you're like, oh, well, I created the whole thing with AI. Like, customers don't want to hear that. That scares them, if anything. You know, we have such a small team. So all of this is to say that I think startups will be most disruptive on people that don't have any existing software currently. And so there's no incumbent that you're battling it out against. And if you are, you know, a software startup, and if you are a startup, you're more likely

to entertain new software offerings that maybe are, you know, a lot cheaper. And, you know, you feel like they're more feature forward or something. Okay. I think we've hit, well, I guess were there any competitors? competitive threats from AI that we didn't discuss that you want to raise? So in the report, I list eight of them. We're not going to go through all eight of them. I think the kind of one that's lesser mentioned, though, is customer journeys move.

And it does touch a little bit back to kind of the customer interface question, but it's a little different in this respect because it's more about workflows changing. So, for example, let's say that you used to be a photographer and an e-commerce company used to hire you to take photos of all. of their products, which is a very common thing. So you want to list 100 products on Amazon. You hire a photographer to stage it all. And the photographer would then go into Adobe Photoshop, fix all the photos, go to Lightroom, make them all look good, and then it'd give you back all your photos. Now, though, what you're deciding to do instead, though, is that you're saying, hey, I'm going to actually take these photos on my iPhone. I'm going to upload them to Meta, and Meta's going to automatically create my own ad. Then I'm going to upload it to Amazon, and Amazon's going to do

an AI generated ad listing. And these are, you know, very strong products. It's good enough, a very trivial difference if we go out and we hire the photographer. So instead, we don't need that photographer at all. So the photographer now is losing work. He doesn't need Adobe anymore either. And so this idea of kind of AI transforming the way work is done and moving these like customer journeys, I think is something that's underappreciated. And I think we're going to see it on a lot more areas than we can even think about. For instance, I was looking up and there's this real estate company that used to stage all of these different apartments or homes for sale, because if it's a staged home, it's more likely to sell. Now instead of them hiring this company out who would go out,

they'd stage it, they'd take photos, instead it's all AI generated. So again, you're losing another, you know, photographer job, basically. And then, you know, you don't have a need for the staging company. So it's disruptive in these ways that we wouldn't have been so obvious kind of beforehand. And I think that is kind of one of the bigger risks because it's changing workflows in ways I feel like it's just kind of hard to understand out ahead of time. It's very easy for us to grasp the fact that AI is creating software. Oh, is that good for software? It must be or is it bad?

Like it's kind of a very obvious front and center thing. But these kind of other quieter things that are hitting on the end markets of the software users, that shows up as a structural decline very slowly. And it may not become apparent until that later, until much later, what's actually happening and why this end market eroded. Okay. When I sell my business, I want the best tax and investment advice. I want to help my kids. And I want to give back to the community. Ooh, then it's the vacation of a lifetime.

I wonder if my head of office has a forever setting. An IG private wealth advisor creates the clarity you need with plans. that harmonize your business, your family, and your dreams. Get financial advice that puts you at the center. Find your advisor at IGPrivatewealth.com. Part of the Beast is the must-see adventure of the year. It's just you and me, boy. One of the most extraordinary on-screen bond you'll ever see.

I'll get you home. Brad Pitt, Part of the Beast. September 25th. Get tickets now. Yeah, I like that. That's definitely something that's not as thought about, And that the AI staging, I saw a ton of those examples where it's just basically an empty room and you've got an AI couch. Some of them are kind of funny and misplaced or whatever. But for the most part, you can get a sense of what a room would look like.

We've got a couple other questions. We've talked a lot about the broad threats to software. I'm curious what subsector, or you can even name individual companies if you like, of the software industry. do you see as the most at risk to AI disruption? I think if we can go broader, I would say anything in consumer software I see as being particularly at risk because with consumer software, which includes stuff like apps, there's the option of when a platform provider like Apple or Android decides to implement an AI feature, they could kind of force apps into it in a way.

So it's an easier way to kind of blanket, intermediateate a bunch of apps all at once. Whereas if you're on your computer, in Windows is a more open system, you know, they could put co-pilot on it and all of that. But it's not really going to ever be like, okay, you have to use copilot in order to interact with all of this. Whereas, you know, with Apple and Siri, it's going to be, you know, a button right on the side. It's going to be pushed a lot. And it's going to be in search. And now it's going to be Siri. They could say in order for you to be an app on the platform.

has to be able to interact with these apps through the backend, something like that. Android could do something similar to. And so I see that as potentially being kind of a kill shot for a lot of consumer software because, again, they could just do it at the platform layer and kind of force companies to opt in. And then they could create their own apps, too, if they don't want to, you know, be that strong of a force. So that's one that I worry the most about.

Other than that, it's going to be a lot of like very niche vertical SaaS apps. I, you know, could give you example. of one that I use in for my work, where I created my own website. And when I did this originally, we had a SaaS provider provide something that it was super simple. It was just a collected name, some emails for people that wanted to reach out to me who are interested in investment advisory. And it was a super simple SaaS app. And I was able to have Claude, basically vibe coded out of existence for us. And so I think that when you're switching over to the small businesses, there's a lot more freedom with how a small business can operate and what they could do.

And I think they will be able to kind of knock out some of these smaller, very niche apps if they want to. Now, not everyone is going to go out and vibe code their own apps. A lot of people for a long time are going to continue to just pay for it and all that. But, you know, I think, you know, in time when you have the people that are young now that are more AI native. And, you know, when they're in business in 10 years, 15 years, they're not going to say, I'm going to go and pay for something. I know I could prompt Claude to do and do it in an hour. So I think that that is going to happen kind of more and more in the small business side.

And on the small business side, it's also going to be more likely that, you know, Claude, a chat GPT, that is kind of your main interface for your business work. So they are going to be the ones that are kind of pulling in data from all these other apps. And so generally speaking, I would say most at risk is consumer software apps. And then second to that, it's going to be very niche, like vertical specific kind of apps. And one of them I think is kind of at risk, a bigger company is going to be Atlassian. You know, in this report, we talked about over a dozen different companies. I'll tell you right now, I don't think any of them are that at risk. I'm singling out atlasian a little bit just because

even though I know that the people that use them currently, it's very integrated in their workflows. I just think in the future, the way people are going to be doing work is going to change a lot. And it seems to be moving a lot more to these, you know, chat boxes with AI and them. and that's going to have more of an ability to track what's happening and create custom dashboards. And so I think that part of their business is going to be a little bit more at work. I know Jira, the development platform is probably a little bit stickier in that respect. But that that's one where I was kind of wondering exactly how that looks like when you have people changing basically, how they're working over time. And, you know, I know, I'm Blassian would like to say, oh, we're going to be the ones that accommodate those changes.

And, you know, you can coordinate with your AI agent through our platform. but I'm a little bit more unsure of that because it seems a little disruptive to the workflow process. Okay, let's talk. So, yeah, I really like that answer. On the flip side, winners of the software industry, I guess the outer harness space, if we want to call it that. I do find some of these AI terms kind of funny, but the outer harness, who do you think would be some of the biggest winners or beneficiaries, software companies specifically, or you could go for broad if you want, afford that outer harness layer.

I think service now, they have the best shot of being a big beneficiary. They seem to just be at the right place. They seem like their developer stack is the cleanest. It's a unified platform. Salesforce and contrast is like this amalgamation of, you know, 70 plus acquisitions. And they're only now going back and, you know, saying we have to clean up our data warehouse and make it all clean, so everything's integrated and AI ready. That's not to say, you know, Salesforce is at trouble or anything.

I just, I feel like service now is probably going to be the one that is best positioned to integrate all this. They feel like a cleaner, newer platform. They talk about being this AI control center because this is kind of the other joccing that's happening between all these software companies is they all want to be the one to wrap all of this AI activity in their own software or platform. And I don't think they're all going to win. I think that you might still have multiple providers within an existing company, but I do think

that a lot of work will go ultimately to either one kind of software platform or it'll be like an enterprise created version of it because they don't want to be beholden to one. I think that service now, though, does have a pretty good shot of being that company where a lot of this AI agent activity is kind of being monitored. And that's in part because, you know, they're very strong with IT service management. So that means when there's issues with anything that's happening, IT-wise, they're able to to monitor it. They have workflows. The other thing is that they have a lot of different kind of detection and sensors within a business to know when something is wrong. They have something called

configure device data management, something like that, CD. Yeah. And that's basically means it has it has a whole map of all of the devices of an individual's laptop, phone, work computer, all of that. And it's kind of a natural place to add AI agent monitoring on top of that. So having those things, it's kind of important to be able to monitor everything that's going on in the business. And that seems like a good place to kind of be when you're rolling out all these new AI agents and you want to permission them. You want to make sure an AI agent is only doing what it's allowed to doing and governance and making sure that it's not accessing data. It's not supposed to access. You want an identity on top of that. You want to know which AI agent is doing which thing.

You want audit trails. So you can understand what's happening. Why did it do what it did? You know, who deleted all of these files? was this guy's AI agent. And so having all of that, I think it benefits service now pretty much. I think service now could be a big beneficiary of a lot of that. Now, that's not, you know, an investment recommendation or anything like that. That's just someone who I think is pretty well positioned. All of, you know, Microsoft also seems pretty decently positioned with the caveat that I don't know exactly what happens to the office suite in the future. It seems like they're going to be fine. And Word, Excel, all of that has somehow continued to exist despite being like.

like, I don't know, pretty much a very average meth product that still somehow able to raise pricing on it every year. My one concern with that is that I could imagine someone like chat GPT saying, hey, why don't we just integrate our own word processor in Excel right here in our app? Because it's a lot more convenient than having it go the other way around because the AI integration in Excel is super awkward. You know, you're in this tiny little box right there. It's kind of clunky and you're telling it to do stuff and it's like, am I allowed to do this? And then you've got to select, you know, these certain things and all that. Whereas if it was the other way around and the chat interface was the main thing and then you were writing your

document within that and or Excel, I just feel like they probably got a shot of getting a lot more usage like that. And they can make it free, right? The same way Google Suite made their work productivity apps free. And they have now an advantage there because they could integrate, you know, AI functionality across all of that. So that, that's, That's the one aspect, and that's a big business for Microsoft. I think they'll be fine, but that's a risk that people don't talk about quite as much. And yeah, I mean, I guess I'll leave it there. But there's a little bit of a risk too for all those marketplace businesses that don't really have a good Waldgarden to like demand people to actually show up to the site. I think that's a good point around that transfer or that export from the AI layer to Excel or word is a bit off. It's kind of clunky. You can kind of layer it on with these Excel plugins and stuff, but then it's like kind of plug-in Inception. If you have like a data provider, it does get a little messy.

Okay, kind of shifting gears a little bit. You've talked on, at least on your YouTube channel quite a bit lately around ad tech. A couple companies specifically, the trade desk who's had a rough go of it. Reddit meta as well. What impact is AI having on these ad tech players or the ad tech? space generally is, I know that's kind of a broad question, but. Yeah, surprise topic. Let's take it.

So generally positive. If you're thinking about for meta specifically, AIs allowed them to match ads to individuals at a much more accurate rate, improve conversions as well, and track improved conversions. So there's two sides of any ad tech thing. It's, is the ad we're showing them? Is it the right ad? And then the second side is the analytics on that. did they actually click into it and can we track?

They clicked into it and got that sale because even if the ad matching works, you don't always know whether or not it actually worked because that's a separate sort of issue. So AI helps in both those respects. The AI also now is helping generate the actual ad copy, which is huge because one of the barriers to entry for a lot of advertisers is simply creating the ads or having enough ads. And what this allows them to do is twofold. One, it means much more advertisers are able to advertise because there's less of a barrier to entry. of creating the ad. But the second thing is the more versions of an ad that exists, the more they could A, B, test it. And the more they could see which is better ad copy and performs better,

and then show that one to more people. So it's improving the ad matching as well. So very big, good stuff going on for meta. For Trade Desk, I don't know that they have incorporated AI that much to the extent that it's, you know, this big positive. In that video I had on Trade Dusk, it, the theme of that video was basically to have issues tracking whether or not ads are actually working, specifically on connected TV, because that's a hard thing to do, especially when you don't have a closed loop. So you have to get all of these different players to cooperate with each other, integrate with each other, share data with each other, and it doesn't seem like they've solved that. So theoretically, once they solve that, then AI could layer that up even more, make it even

better. But as of now, I'm not so sure. I know they just rolled out a new platform, that's much more AI forward. But the last AI features they've rolled out, people actually complained about them because what ended up happening was they alacart charges for data, basically, because they don't own the data. They have to buy it.

And the AI was telling all of these customers that we could get you better and better results if we use more data sets in the tracking. And so customers were seeing their price basically go up because it kept buying more data packs to improve the targeting, which if you're on a performance, you know, looking at performance, it did mean that, you know, performance was probably better. But a lot of

these buyers instead just saw a much higher, you know, advertising price. And now they're like, why are we spending even more money on this? I don't really know what it's doing. And so they kind of scoffed at that. So it was kind of a negative in that respect because they weren't, because the performance that Trade Desk is providing to, it's not usually return on ad spend, which is what meta does. It's other, you know, kind of lower down metrics, which are going to be stuff like, you know, how many website clicks you're getting or something like that. Or, you know, it's not a full, you know, how many people are actually buying this product. It's maybe showing a little bit of a lift or something like that. So, yeah, I can't say that it's been a huge beneficiary for trade desk so far. In theory, should be,

though. They're having other issues, though. You got to check that video out to fully understand that one. And then you asked about Reddit. I could imagine in theory it should help, but Reddit's bottleneck right now is creative copy. Because if you try to advertise on Reddit, you might notice like an ad that does well on Reddit. It's like a very different sort of formatted ad. It usually is kind of in like the same sort of copy language as the way people talk on Reddit, different like subcultures and all of that. And so that's kind of been their sticking point is getting people to like have that sort of copy that works well on Reddit.

Because if you just run like a basic video ad you're running on Instagram, people just score right by that because it's like a primarily text platform. And so that's kind of been their issue. is basically getting more advertisers onboarded and then the right ad copy. But once you do that, then yeah, sure, in theory, AI helps in that regard too. Met is the only clear beneficiary, I think, of AI. And I won't just say for ad tech, I'll just say in general. They're the only clear player that has implemented AI and has showed scaled returns on it so far. Now, you can make an argument whether or not those returns are high enough for how much they're spending in CAPEX currently,

But you do see a clear lift, basically, to revenues, to operating profits from when they started implementing AI. You've got to back out the family of app stuff and some of the super intelligent stuff in order to get there. Yeah, to that point, I think you might have talked about this on your video, but the historically, there's been sort of a bit of an inverse correlation between add volume and add pricing. As prices grow, there's a little bit of a lag on spending. And quick shout out to one of our sponsors, Fiscal AI, you can now see, I think it's double-digit growth, double-digit percentage growth on both average ad pricing and overall ad volume. Brad, I know you had one more question before we sign off here, but anything else. Yeah, this is the final question. We haven't talked about this much, which is the model companies trying to compete everything away.

another was a hyperbolic statement, at least released from the news. It was sourced. We don't know if Dario Bodhi actually said this, but he apparently said to someone that he believes Anthropic could be the only company left. Yeah. You know, Dario, you got to get control of your own thoughts there. But what is the risk here? Has any of this happened yet?

Has there been any sort of competitive threat from the model companies kind of, you know, entrenching or competing with everything. So that kind of belief stems from this idea that you're going to get basically perfect AGI that's able to do everything perfectly in the one shot. And there's no reason for humans or anything else whatsoever. Having said that, even in that scenario, it still doesn't make sense because you're not going to have all of like the physical assets in the world. You know, you're not going to own delivery trucks and all that.

So let's not spend too much time on that because it's a little too out there. But what I will say is that I think something that's going to happen with these frontier models is that over time, their rate of improvement relative to the best open source models or the other players, it's going to continue to shrink. I think it's going to get harder to continue to stay on the leading edge of intelligence and to do it in a way that actually matters and is meaningful, not just like, you know, hitting some marks on some test. But to continue to actually have frontier intelligence and have a leading edge, I think it's going to get harder and harder to do, become more expensive, and it's also going to that gap is going to narrow. Now, I know people can argue on the other end and say, well, what if you get recursive AI improvements, then it kind of leads to like a spin away effect and then you can't catch them. I don't know about that. I think, you know, the amount of time one model is still ahead of another. I think it's going to just continue to shrink even if, you know, that does mean that they're using a lot more AI reiteration to improve the models. So what happens in this scenario, the game theory suggests is that once Anthropic or OpenAI realizes that there's only so much more they could do in terms of improving their models.

The game theory says that at that point, they're going to need to solidify their competitive advantages because they're scared of getting commoditized. And so if they become commoditized, that basically means their models become more swappable. And that's happening explicitly with a lot of these SaaS providers. You know, use the Atlas Angelin with Salesforce. course, you could use Claude. You could use all sorts of different models. You know, same with Intuit's reasoning engine, all of them. They all swap out all these different models you could use. And the, the less differentiated they are in terms of intelligence, the more likely they are to be swapped out, not used, not be able to get their margin on inference. And so what I think is going to happen is they're

going to start to see the writing on the wall one day. And they're going to see that the way that they can actually survive this is by productizing more of their intelligence. And so what that means is they're moved down on the value chain. So they did this once with Claude Code for Anthropic, and that was very competitive to a lot of these coding startups, and that was a very big product. And they basically said they reserve the right to do something like that in the future. And I think what's going to happen is, as they can't differentiate as much on the model layer, they're going to have to leverage the fact that they have a big brand, they have a lot of users that currently use them, and they're going to get into more specific verticals because

that makes it a more specific, a more sticky product. So I quote, imagine them, for instance, saying, hey, we're going to roll out, you know, Claude accounting for small businesses, and now, you know, we're going to go after that market, basically. Or, you know, we're going to now go after vertical specific financing stuff or marketing. All of these different SaaS apps, I think eventually they're going to have to do that because I just don't see how they're going to be able to command a premium price if their model isn't always much, much further ahead of all the other ones. And it just, it doesn't seem like that's going to happen, at least in a way. You know, you could still be the smartest model, but if it doesn't matter for most work,

then people are just going to use stupid or models. And then, you know, the smartest models will be used for only very certain tasks and you have too small of a business, especially when you need, you know, hundreds of billions of dollars in order to support your valuation in the future. Now, the reason I think they haven't done this so far is because they don't want to alienate existing customers and freak them out, which was this whole rant that Palantir CEO, Alex Karp, went on, where he's like, you know, all of these enterprises are worried that Anthropics just going to steal their data and then eventually go into their own business, which they haven't said that they're not, though. And so I think that that remains the big risk. And I think they haven't

done it because all these profit pools are still kind of relatively small to what they're currently generating. You know, you could take all of into its revenues and it really doesn't move the needle that much for Anthropic. But eventually, I do think they're going to have to productize some of this if they want to have more of a lasting competitive advantage. And if the intelligence frontier differential starts to fade, which I do think will eventually happen. That could happen many years out. But I kind of see that. I just don't understand how it could be the case that you could continue to lead this lasting sort of improvement over everyone else indefinitely for so long without continually throwing, you know, more money at it too. Because I don't think that this AI funding

environment is going to last forever where you're able to raise hundreds of billions of dollars. Yeah, that is a good point to include. All right, Drew, thank you for taking the time to join us today. Before we get out of here, what is a 30-second pitch of what people can hear and read over at Speedwell Research or the other media that you do? Yeah, you can find out more about me at Drew Cohencapital.com. That's kind of my main website. If you want to buy the research report, you could go to Speedwell Research.com. otherwise check out my YouTube, any of my socials. I'm Drew Cohen, money on all of them.

All right. Beautiful. Thank you once again for joining. Thank you to the listeners for listening. As a disclosure, we are not financial advisors. Anything we say on this show is not formal advice or recommendation. Ryan and I are any podcast guests may hold securities discussed in this podcast. May have held them in the past and may buy seller hold them in the future. Thank you, everyone once again for tuning in. And we'll see you next time. Do you wish you could just hit skip on the worst parts of your life? You know, the same way you can skip an ad? I get it.

I'm Siyaya and I live in Ice Cove. I've made some questionable decisions that didn't end up the way I planned. And today I'm still figuring it out. Somehow things usually get worse before they get better. Apparently, that's how I roll. So bundle up and come along for the bumpy ride. Stream a new episode of North of North Tuesdays on CBC Gem. Advertising connects you with an audience that's engaged and ready to discover your brand.

Launch a campaign with Spotify Ads Manager.