How AI Will Unfold and 8 Single Stock Takes: Daniel Pronk Conversation
Get smarter on investing, business, and personal finance in 5 minutes.
If you were forwarded this email, click here to subscribe for freeThis newsletter is adapted from a previous video Drew Cohen did with Daniel Pronk.
This issue of Five Minute Money is going to be a bit different than our previous issues.
Recently, I had a conversation with Daniel Pronk, who is an investing YouTuber with over 280k subscribers.
We talked about the AI investment cycle and gave our thoughts on 8 companies including; MercardoLibre, Sea Limited, Netflix, Meta, Google, and ServiceNow!
Below is an edited transcript of our conversation.
Hope you enjoy!
Is MercadoLibre Investing for Growth or Defending Against Shopee?
Daniel: So MercadoLibre reported its earnings recently.
The stock was down 13% following earnings.
Seems like just another quarter, where they're growing the top line very well.
I believe it was 45% again.
But margins are still down.
So they're continuing that reinvestment cycle.
And right now that is very clearly not what the market wants.
Drew: I feel like a lot of people equate it to Amazon, where they say Amazon was in this cycle of investing a lot for a long period of time, and they always kept their margin low on purpose.
Now, the difference I see with MercadoLibre is that they already were pretty profitable, and they took their margin down in response to competition, which is not something Amazon ever had to do because Amazon never really experienced competition in their home market.
So do you worry that this is actually an indication that they kind of had peak profits, and maybe all these investments are more about defending their remaining profitability?
Daniel: Yeah, I actually do agree with you that competition is a real risk to MercadoLibre's business, and I would also agree with you that their margins have peaked.
I mean, in Brazil, the largest market, I believe MercadoLibre is doing $20 billion in revenue there now in the trailing 12 months.
They're actually accelerating revenue and just grew 50% year over year, and that's their largest, most mature market still growing 50%.
But I think it's still a long ways out before these companies really have to get bloody with each other because there's still just so much to capture.
Drew: It's been hard to pinpoint the mature profitability of this platform because there's kind of a range I see.
There's Amazon in the US, which is, like, the highest of any e-commerce player.
Maybe it's 8 to 9%.
Then on the other end you have the Chinese players.
Pinduoduo kinda undercut a lot of people on price.
They're still able to make 1.5-2% margin, EBIT as a percent of GMV.
And I say it that way so we're agnostic of the 1P versus 3P accounting.
So my question is then where is MELI landing in here?
At one point, MELI had 14% margins.
Now they're kinda cut in half, closer to 7%.
Daniel: I do think, though, that as the fintech business continues to grow, it will actually bring up their operating margins as well.
I think e-commerce will forever be a low-margin business.
I think the margins will also come from advertising.
Someone said they have an 80% operating margin on that revenue, and that revenue is now $2.2 billion in the trailing 12 months.
So that advertising business is probably doing around $1 billion in cash flow profits today.
Drew: It's always tricky though, 'cause whenever a business is being pressured, you don't know to what extent that the investments are to defend their existing business versus to grow.
If I ran the business, I'd have to do the same thing too, but have to is kind of the operative word there.
This was not something they wanted to do, and I think would've been more profitable business without kind of the entrance of Shopee.
Daniel: I would agree that their business would be more profitable without Shopee, and they probably are going to have lower margins long term because of that.
I do think that they would still be putting the gas pedal down pretty hard without Shopee, though.
I actually think that Sea Limited is a very fierce competitor.
I don't like the markets they operate in as much as I like the Latin America market.
I think it's much more competitive over in Asia.
Drew: I think the flip side of that is they're outcompeting everyone.
Something I just find fascinating about Sea Limited is Forrest Li, who, for those that don't know, he's the founder of that business.
And it started out as gaming internet cafes.
And then they got into e-commerce in 2015 as a very late entrance, and was able to basically outcompete Alibaba, all of these other competitors that were there for years, and make very tiny margins work for it.
And then they were able to kind of wedge their way into Brazil to actually ship more packages than MercadoLibre does there currently.
We want these companies to invest for the long term.
We just want them to still be profitable while they do that and show growing profits, which, guys, it's just not how it works.
It's a trade-off.
How Risky Is Consumer Lending in Latin America?
Daniel: I think MercadoLibre is the stronger business than Nubank, for example.
MercadoLibre basically has a Nubank within it.
But Nubank, I think, is going to be received better by the market because it has those real profits coming in right now, and the GAAP earnings are growing every quarter.
Drew: The pushback I would have about that is I've never heard of the word comfortable together with Latin American financial institution or bank.
I think the biggest risk with both of these businesses, Sea Limited and MELI, is just the credit risk.
You're lending to the area of the market that all traditional financial institutions have overlooked, and that's the opportunity.
But they've also overlooked them because these are areas of the market that have very, very high default rates.
For some of them, it could be 80% plus, but that's because some default rates are 40%, and then you have inflation, that could be another 20%.
Daniel: I actually have some people in my community that live in Mexico and Brazil, and they say how aggressive Nubank and MercadoLibre are being with trying to get them to take loans from their businesses.
And they got no repercussions for not paying them back.
Drew: My understanding of what MELI does, though, is they'll restrict your access to the marketplace if you don't pay them back, which is a decent leverage tool because it's one of the most popular ones.
So then I guess you then go to Shopee next, and then you default on Shopee, and then you're out of options.
Daniel: And I think the market may actually price MercadoLibre lower long term because 50% of its business may come from banking one day.
AI Investment Cycle.
Drew: I'm looking at the Nasdaq down 5%, a lot of AI stocks, memory down 15%, SanDisk down 20%.
Do you have any thoughts about what's going on with the AI investment environment?
Daniel: I think what we're seeing in the market is really highlighting how short-term focused investors and the market are today.
For example, Amazon was up, I think, 7% yesterday because investors were thinking that the AI CapEx build-out was going to slow down, which is going to boost the hyperscalers' profits again in the short term.
But at the same time, I'm an Amazon shareholder, and I want them to be making these investments because I think it's going to create more long-term shareholder value.
I mean, the Democrats won the Senate and the House, and it sounds like they're not as pro-AI and pro-data center build-out.
And I think that's just causing a lot of people to rotate out of that sector due to the uncertainty.
Drew: The thing I keep going back to is just capital cycles.
What happens is there's some sort of new technological innovation, it sparks a lot of imagination, investment returns are thought to be really high, people continue to invest into it, demand is much higher than supply, and so more and more investment continues to enter the industry before inevitably supply overshoots demand.
The risk here is that, I call it like an air pocket, where basically supply overshoots demand.
When you cut prices, you see those returns that you thought you were going to get become illusory.
But that's not to say that long term, it can't be a good investment.
It just could be a year, a couple years where there's this air pocket.
I did incremental return on invested capital analysis for Google, and right now it's still 15%, even with all of this investment.
Daniel: I think it's even similar to what Amazon did during COVID when they were building out their e-commerce business.
And then it took them a couple of years to actually get into the demand that they built.
I do think that this one, I hate to say it's a little bit different, but it sounds
Drew: This time's different?
Daniel: Yeah, it sounds like it's a little bit different this time because they're getting so many commitments upfront.
And then also, just as a user of AI, this technology is real.
So my strategy has kind of been to invest in the hyperscalers that have the compute, and I'm not investing in the things that I think are going to turn out to be more cyclical long term.
Like the chips, for example.
Just look at memory stocks' cash flows throughout history.
It's a wave, and we're clearly in a big wave up.
Drew: Even Stanley Druckenmiller sold Nvidia too soon, right?
You're buying Micron, you're seeing an 80% margin, and it's maybe a mid to low single digit multiple, and you're like, "That's so cheap."
But knowing the history of these stocks, they have traded at low single digit multiples before, and that's because it doesn't take too much to push them into the red.
And I think if you were going to say there's one time where you could say this time is different, the only reason why I would think it could be AI is kinda twofold.
If you think of economic productivity, there's labor and capital.
Most of the time it's just the capital side of the equation, whereas with AI, now it's also influencing the labor side, which is kind of a difference because AI can become labor.
And we haven't even seen, I think, the scariest thing, which is going to be what happens when AI gets combined with robots, which is coming.
The second thing I was going to say is that it's almost been prophesized in a lot of sci-fi novels, in movies, for 100 years plus.
And so I feel like there's just such an inclination that humanity was going to eventually tilt in this direction, and once we hit it, it would be a very different paradigm shift than other technology before.
Very Few Businesses Are Free of AI Risk.
Drew: There are very, very few businesses that do not have an AI risk.
And at the same time, then you're also wondering, are we getting too far out ahead of ourselves?
Because we know technological developments don't usually happen as quickly as people predict they are.
A lot of people are freaking out about Muse being able to book travel.
If it takes 5 to 10 years for that to become a mainstream habit, then they have a lot of time to adjust their business.
Daniel: I mean, even in the construction industry, as you said, I think we're eventually going to get robots that are powered by AI and can do this stuff.
I would not be surprised if universal basic income of some sort is implemented in the next decade or so.
I just continue telling people I think it's going to get weird.
How Do You Invest When Nobody Knows How AI Unfolds?
Drew: I had a podcast with Chris Mayer, and he talked about how the more decisions you're making, the more possibility to make wrong decisions.
I don't necessarily want to jump to AI is going to take all the jobs, everything's going to be done through agents, all of these businesses are going to be disrupted, 'cause I still know that adoption takes a long time.
And so it's kind of a balance between what do I think this business looks like in the long term, and what do I actually expect to happen?
Daniel: I would also say that a lot of the discussion that we just had was honestly pure speculation, right?
So how I have been investing in my portfolio is, as I said, I've been investing in mainly Amazon and Meta, because I view Amazon as kind of like the utility for AI.
But a lot of the stocks that I actually own are ones that are like regular boring businesses.
For example, one I've been talking about on my channel that got destroyed today is Construction Partners with the ticker symbol ROAD.
They just build and maintain roads.
Drew: We're not going to need roads with AI, don't you know that?
Daniel: That's what the market's telling me.
But the thesis is very simple.
As long as we're building more roads, and roads are made with asphalt, and they need to be repaired, this company's going to have business and produce cash flow.
And personally, I think a lot of digital businesses and even some software businesses are much more disruptable.
Drew: Do you want to give names?
Is Intuit Vulnerable to AI Tax Preparation?
Daniel: I think the one that comes to the top of my head is Intuit.
I think online tax is going to be done with AI pretty soon.
Drew: I'll push back on that one.
I think that you can get AI to do your taxes, but you can't get it to sign off on them and get the guarantee that they're done correctly.
And now more and more you're paying to talk to an expert on very complex tax issues, which is where more of their business is coming from.
And my general thought process on all of the AI model companies is that they're going to eventually have to verticalize and productize their AI because AI over time, the difference between leading edge and second tier, third tier, it's going to become smaller.
I think OpenAI's fully going to have a small business suite.
But when a big conglomerate tries to go after all these opportunities they usually fail with most of them.
Daniel: I've kind of been operating under the assumption that they're going to become commodities long term and not really have that much pricing power.
So it is interesting that you went a step farther and said, "Well, then they're going to have to verticalize."
Who Will Control the Outer Harness, and Can Model Makers Profit?
Drew: Right now, there's this debate about who owns the outer harness.
You're going to want something kind of wrapped around all of it that has the permission layer, the identity layer, the governance layer, the audit layer, and all of that together is going to be the outer harness.
ServiceNow wants to be the outer harness with their AI control tower.
My suspicion is that enterprises might want to build that themselves, and they could use something like Amazon Bedrock to do that.
All of this pressure in models becoming more commoditized becomes very deflationary, and that means a lot more AI usage on the one hand, but also a lot lower gross margins for Anthropic, which then doesn't have the profits to continue to support a build-out.
Daniel: I 100% agree because I think that Anthropic and OpenAI's margins and profits are at risk long term.
But I do think that at the end of the day, what it really comes down to for me is if the demand is real and if AI is real, then that air pocket from Anthropic will be sucked up by someone else because someone else is going to want that compute.
Drew: You don't just have OpenAI and Anthropic, you have Meta also trying to be a frontier model.
I don't know why people count out Google, but they're still there, even if maybe they did fall a little bit behind.
SpaceX is up there too as well.
So that's 5 very well-funded competitors, all trying to get to the leading edge of intelligence, and it's just hard to believe they're all going to be good businesses.
Why Meta Looks Like the Clearest AI Beneficiary.
Daniel: I do like that Meta is building their own near frontier model, if not at the frontier in some areas already.
But what they're doing mainly is using it for their own business internally.
And I think that actually is the more sustainable way to build an LLM because you're not really relying on selling that LLM to other people.
Drew: I've been saying this for a few quarters that I thought Meta was the clearest beneficiary to AI just because it just works that ad algorithm and makes everything better with that.
And it wasn't trading at such a high multiple that you actually needed all these other AI initiatives to work out.
You could say that now they're going too far with the couple hundred billion in compute they're spending, all the R&D spend too.
Daniel: Their balance sheet overall is still pretty pristine, and this company's doing over $100 billion a year in operating cash flow.
And I agree with you, there was a margin of safety recently, so that's why I made Meta a pretty meaningful position at the lows there.
Can Google Search Withstand Agentic AI?
Drew: I think though that maybe Google's in a different position, and this is one I've been thinking more about because Meta's core business is not at risk.
The distinction I see here is that Google has never been a company that monetizes time spent on the platform.
They've been a company that monetizes commercial queries, and that's a very different way to monetize a user.
I think why it hasn't been disruptive to Google Search yet, or I should say Google Search's ability to monetize yet, is I don't think LLMs are cannibalistic to commercial search queries.
But my worry would be that agentic AI is.
I just think if you think of there being 10 commercial queries that used to happen on Google, even if just a couple of those go to ChatGPT, to Muse, to all these other options, that's still a big headwind to their business.
And they don't make that up by more engagement because they lost that commercial intent.
But is it going to be a better business in 5 years?
Daniel: I am going to Muse already being like, "Find me a plumber."
That is a sponsored search that Google would use to make money off of, but now I'm just going to my agent.
I've also talked to a lot of business owners that I know who run subscriptions and websites, and they say that their search traffic has just like plummeted.
So yeah, Google is one that I don't own anymore.
And I don't know if you use ChatGPT's free version or not, but I do sometimes, and they're showing me ads when I'm in the conversation, and it's a terrible experience.
I mean, they are accelerating the business with Cloud, but I don't know if Cloud is a better business than their search business was.
Drew: That cloud business is growing incredibly fast, but search is one heck of a business.
Is Netflix a Business Worth Owning?
Drew: Are you in Netflix?
Daniel: Netflix is a company that my audience loves for me to talk about, but it's one that I have never been interested in owning.
Drew: Why aren't you interested in it?
I think it's around 18-19x forward earnings, something like that.
Daniel: I don't like that they're getting more opaque with their reporting over time.
In my opinion, that suggests that their metrics are no longer sexy, if you want to put it that way.
I believe it was the metric their watch time per viewer is actually declining, which I don't love because I think that suggests that the perceived value of the service per user is going down.
So to put that simply, I think that business has a cap on it in terms of how much revenue it can charge.
Drew: I'll take the other side to that.
Yes, you're right, a lot of growth is going to come from price increases, but that's a formula that actually has worked for a very long time for cable companies.
And even most recently they said churn was even less than it was historically on this last price increase.
I think eventually it's very possible they have a free tier, and then it's just loaded with ads.
The fact that they moved to advertising after years of saying they'd never do it, that was a big thing.
Daniel: It seems like Netflix's business success now almost relies on their advertising business working.
If Netflix can even continue to grow by 10 to 12% per year, then I see the value in the stock today.
But for me, it always comes down to, is this a business that I feel comfortable holding for the long term?
And Netflix is one of those ones where I just don't.
Does Execution Matter More Than a Moat?
Drew: I think I've been humbled in my investing in thinking that companies have moats.
I remember this exchange between Warren Buffett and Elon Musk where Warren Buffett said moats were really important, and Elon Musk said, like, "They're useless, it doesn't matter," or something like that.
And I think the thing I've kind of come to appreciate is that execution matters so much more than historical moats.
And there's almost no company that is so good that a manager can't ruin it.
Daniel: I am a big Warren Buffett guy.
But over time, I have come to kind of your conclusion where execution actually is what matters a lot more.
There's asphalt paving companies for sale in my community that are selling for 3x cash flows.
There's no moat there.
You’re just an asphalt paver.
But if you can run your business well, you can make a lot of money.
Drew: I don't want to overfit that though, 'cause there's still a lot of businesses that can be very well run, and it's just the wrong industry.
Any lesson you could overlearn almost.
But the reason why I bring up execution is because I was thinking about AppLovin.
It was a business I understood well, I researched it very well, I wrote even a 100-page research report on it, and I was able to refute every bear case on it.
But I ultimately made kind of the wrong decision on it.
Especially in ad tech, that thing is a graveyard of businesses that have blown up.
And on the other hand, though, I don't know that I've ever seen a business execute as well as AppLovin had in the past 4 years in specific post-ATT.
Daniel: I have a little bit of AppLovin in my portfolio.
I considered it speculative for, it sounds like, pretty much the same reasons.
So it's basically a bet that they're going to be able to run faster than their competitors and maintain that lead.
Drew: Very, very hard to sustain 77% EBIT margins they had the quarter before last one.
Usually competition comes and takes that lunch.
And then the flip side is I knew from all the advertisers I talked to, their e-commerce return on ad spend was very, very strong.
But AppLovin still remains dependent on out-executing their competition in a tough industry like ad tech.
So the difficult question that needed to be answer was if they can keep out-executing their competition and maintain their take-rate.
Which Software Companies Are Hardest for AI to Disrupt?
Drew: In terms of software companies that, at least to me, are the least likely to be disrupted by AI, maybe Palantir is up there too, but ServiceNow is definitely up there too.
I think that existing enterprise software companies, I think they're going to be fine.
I just think they're going to struggle to get new business and to cross-sell new products because there's just going to be less of a need to rely on your existing software provider.
If you're looking at ServiceNow, that's not something you want to vibe code.
I think if we talk about someone like Salesforce, a little more likely, especially in terms of replacement, in my opinion.
Daniel: With ServiceNow, though, the things that have kept me away, at least recently, is the company's margins are coming down.
The revenue is still growing pretty strong at 20%.
And also, the operating margin has been coming down because their stock-based compensation is exploding.
Also, the CEO. He's a little too promotional.
And then they're kind of guiding for a little bit of a deceleration of revenue, so it's like, okay, you're saying you're this massive AI beneficiary, so why are your numbers not necessarily lining up?
Drew: When I ran my math, I think you need at least 20% growth to get a strong return on that.
It's hard to know whether or not a company's going to grow 20% for 20 quarters in the future.
Daniel: What I've learned, especially in this market, is if you show any sign of deceleration, the market is going to re-rate you down immediately.
What Are Your Biggest Investing Mistakes?
Drew: What do you think are your biggest investing mistakes you've made?
Daniel: I've had 2 really big blunders in my time on YouTube.
The first one was back in 2020 to 2021.
I was invested heavily in the cannabis industry in the US, and at the time, the businesses were growing 100%, 80% year over year.
But then once all of their inventories got built up, and then there became more supply than demand,
Drew: Sounds familiar.
Daniel: Exactly, profit margins just tanked, and the stocks all tanked with them.
So I learned from that, in general, that investing in new industries is extremely challenging and risky and speculative because you don't know what the margins of the business or the entire sector are going to look like long term.
Drew: I want to highlight that is exactly the capital cycle argument that I was bringing up earlier in Capital Returns.
Or is the real mistake in the cannabis investment the fact that these are pretty commoditized products that people don't care about?
Daniel: I guess that would be the second layer, is that it really was just a commodity.
And then the other one was I invested in a subprime lender, which is also kind of funny since I own MercadoLibre.
I invested in a subprime lender called GoEasy, and they did an acquisition of an auto loan company, and then quite literally they were reporting false information to the public.
So they were making their earnings look really, really good when in reality they were losing $50 million a quarter.
And then it came out, a new CFO came in, saw this, released the data, and the stock dropped 70% overnight.
I did talk to another investor up here in Canada.
The reason he spotted it was he was looking at similar businesses in the US that were reporting much, much worse numbers.
So maybe my lesson there should be to pay more attention to the competition.
Thank you for reading! I really enjoyed this conversation with Daniel, and I hope you did too!
Also, Daniel wanted to give you a special 25% discount on his investing community. Use the code DREWCOHEN25OFF at checkout using the link below.
Daniel's Investing CommunityFor More on my conversation with Daniel, check out the video below.
YouTube FAQ
Nothing in this newsletter is investment advice nor should be construed as such. Contributors to the newsletter may own securities discussed. Furthermore, accounts contributors advise on may also have positions in companies discussed. Please see our full disclaimers here.