A $13 Billion Bet on Robots?

0:01Matt Frankel:We're going to talk about a$13 billion bet in robotics. Motley Fool Hidden Gems Investing starts now.

0:12Matt Frankel:Welcome to Motley Fool Hidden Gems Investing. My name is Jon Quast. I'm your host today. I'm joined by Foolish contributors Matt Frankel and Rachel Warren. We're going to get to a couple of questions from our mailbag today, talking about the bullish scenarios for buying a stock and also the bearish things that you need to consider. But first, I wanted to talk about this story. I think it's pretty big. And let's talk about the robot called the Micro Duck. Now, Micro Duck, if you ever imagined a Furby having a child with the Pixar lamp, that's what I think this robot duck looks like. It is a Cyclops duck that walks around your house.

0:54Matt Frankel:It quacks, it sings. And what is so interesting about this is that it is made in collaboration with Hugging Face. Now, Hugging Face is a company that NVIDIA is reportedly looking to acquire for about$13 billion. So perhaps let's start there. Rachel, we're going to go to you here. What is Hugging Face and why would the $5 trillion company known as NVIDIA want to buy it for$13 billion? Hugging Face is essentially one of those central town squares of the AI world. It's a cloud repository where you have millions of developers and researchers that collaborate, and they use this space to share, test, and host these open source, open way AI models, data sets, software libraries.

1:39And if you're not familiar with the term open source, maybe you've heard it, you're not sure what it means. It essentially just means that the underlying code is completely public. Anyone can modify it for free instead of being locked behind a proprietary wall. Now, you've got the closed source tech giants that are keeping their AI lock behind proprietary APIs, but Hugging Face, it's sort of one of those democratizing platforms in this space. It's where you're seeing a lot of the foundational innovations or the latest open source large language models live. Now, why would NVIDIA want this? I mean, some of this I've already explained.

2:13This obviously is a compelling element to add to NVIDIA's wheelhouse. And, you know,$13 billion is essentially a drop in the bucket for NVIDIA. But it comes at a time where the likes of OpenAI and Alphabet are aggressively developing their own custom AI chips to break their dependency on NVIDIA. By buying Hugging Face, NVIDIA would gain control over a really vital distribution layer where the entire global developer community meets. They would also take over the team behind an open source software library known as llama.cpp. That would allow AI models to run locally on consumer devices. So there's a lot of interesting things at play here.

2:52One final thing I'll note, the estimates vary, but I've seen analysts saying that NVIDIA would be paying anywhere from 86 to 129 times current revenue for hugging face should this deal go through. So they are willing to pay a premium to lock in what they say to view as a very captive customer base.

3:09Matt Frankel:certainly 13 billion dollars is a lot of money to you and i but to nvidia it can afford to pay 100 times its revenue and not really notice it at this point but it would seem like hugging face is a way that nvidia is trying to keep the ai model companies like open ai for example trying to keep them from becoming maybe too powerful because if they become too powerful then they start going into custom silicon and then perhaps become a stronger competitor what do you think about that read on it. Is that what they're trying to do here? Just kind of keep the competition there from the open source to keep the frontier more gated models from becoming too powerful.

3:47Yeah, I think that's right. I think that that's one element of this for sure. I think there's a lot of reasons why this would be a smart acquisition for NVIDIA. And because this is such a go-to platform and space where you have all of this open source work from developers and researchers around the world. It's a data goldmine, essentially. But I think it would also give them an edge in the open source space that they currently lack. It's interesting because obviously Hugging Face is not a household name. I think it might be easy for investors to overlook this news that we're talking about today.

4:22But I do think it's actually something that should this deal go through would be really, really important for NVIDIA long term. And also, you know, locking in that ecosystem into their broader network, I think NVIDIA would also have an entirely new captive customer base that can default to their hardware, their Kuda software ecosystem. There's a lot of wins that could come for NVIDIA out of this.

4:44Matt Frankel:Okay, so let's turn to MicroDuck here because I think this is the interesting part of this news. When you look at what Hugging Face has done so far is regarding open source, it's more in the large language models or the LLMs. Now this is more in the robotic area and And what MicroDuck is different about from other consumer robots that I've seen is it allows more personalization. It's users, the customers, the owners of these robots are going to be able to train it, to teach it things. And these things are going to kind of take on a little bit of personality over time. And so that really kind of creates this open source robotic operating system, if you will, from Hugging Face.

5:22Matt Frankel:And so I just want to get your thoughts here. My kids think that this robot looks pretty cool.

5:26Jon Quast:when i was younger we used to camp out at best buy for the latest playstation now i feel like john's going to be camping out for the micro duck with his kids times have changed for sure it's an impressive device i feel like nvidia is kind of making a bet making this acquisition nvidia has such financial flexibility they can make bets with 13 billion dollars this is like just to name a company that i follow very closely this is like if paypal spent like 70 million dollars on an acquisition. It's not a rounding error, but it's not a big chunk of money. So I'm not sure I'm going to get in line to spend$399 for it, but it's a neat advancement.

5:59Jon Quast:You can train behaviors and simulation. You can then deploy them to the physical robot. It's a cool teaching platform for physical AI. It's not really a household appliance or anything particularly useful at this point. This is not the Optimus robot, to be very, very clear. This is a data play more than anything, as thousands and thousands of these working in different households. It's going to really create an excellent data set for locomotion and manipulation that video will eventually own if this acquisition goes through. So I see this really as a data play, not that NVIDIA wants to own the micro duck, but they want to own the data behind it.

6:34Matt Frankel:All right, Rachel, walk us through. What is micro duck? What do we get out of the box? Micro duck? I would say it's a milestone for AI hardware. So it was actually developed by a startup called pollen robotics and hugging face acquired Pollen Robotics last year. So MicroDuck is this very tiny, you know, 25 centimeter, 800 gram bipedal duck. It's a$399 playground, essentially, for physical AI. It has over a dozen motors. There's a camera, miniature LiDAR. It has an articulated beak that acts as a gripper. You know, look up a photo of this. It's kind of interesting. But essentially, its purpose is to give everyday developers, as well as students, an affordable way to test real world AI software.

7:14So it really bridges that gap between digital code and reality. Users can train a digital twin in a simulation environment. They can then deploy custom behaviors like roller skating or playing soccer directly onto that machine. And because MicroDuck's code and simulation environments are completely open source, basically the consumer can become a creator. So this is incredible for professionals that are using this, but also just for individuals that are experimenting here and students, as I mentioned. So Hugging Face has said that MicroDuck has already blown past$2.6 million in pre-orders when it had its first 24 hours of launch.

7:52That's over 6 ,500 units in a single day. So a lot of excitement here. All right.

7:59Matt Frankel:Now it's time for my way-too-early, over-the-top take here. I want to get your opinions because if I look at NVIDIA, definitely there's an incentive to keep the open source large language models growing and adopted. Now I'm looking at this saying, are they really pushing into the robotics? And some of the companies that we follow strongly, Tesla comes to mind, a large part of the investment thesis is the growth and kind of domination of its future optimist robots. And then you also have companies such as Figure looking perhaps for an IPO in the future. This is also a robotics company. These would be more locked ecosystems that they're trying to create.

8:40Matt Frankel:Is NVIDIA trying to keep that open source? And is that a threat at all to the theses of these more robotic companies? Yeah, I mean, I think that the future of robotics is definitely going to feature an open source faction. And I don't think this has to be a winner takes all environment. it, I think that there will probably be a healthy mixture of open source and closed source as we look into the robotics landscape over the next decade and beyond. The reason I think the open source piece is so important is solving physical AI. It's too complex for a single company to do alone. And I don't think you need to have a single company do it alone to have a really interesting space full of wonderful opportunities, both for customers and for investors.

9:17And, you know, proprietary ecosystems, a lot of these closed source systems, there's also some bottlenecks that that can create in terms of general purpose learning, data aggregation, And having that open source foundation, obviously it democratizes the AI stack, but it also pulls real world data from thousands of diverse environments. It could really accelerate how quickly the machines learn when it comes to navigating what is, you know, in fact, a very messy and unpredictable real world. So I do think if NVIDIA is able to successfully acquire this company and push forward that open source robotics framework, I think we could potentially see some kind of shift in the landscape.

9:51But also this means that a lot of the smaller startups, academic research labs, independent hardware builders could potentially gain access to a lot of these state-of-the-art models. And that could really drive the industry forward as a whole. And then there's a lot of downstream benefits for NVIDIA. So I think this can all maybe seem a bit confusing to non-creators, but I personally think this is a fascinating acquisition to watch if you have NVIDIA in your portfolio, on your stock watch list. Keep an eye on what happens with this.

10:18Jon Quast:I agree with Rachel for the most part, even though I feel like I'm the only one of the three of us who's not going to go buy the micro duck. What she's describing is known as cross embodiment learning. That's the official term for this. It's already happening at scale at Hugging Face. Viddy is not paying$13 billion for a model library or for the physical hardware. They're paying to own the place where the world's robotic data is stored. The open source has a lot to do with it. I want to point out some important context about the maturity of what's going on here. I mean, the own limitation of crossbody is the majority of things are like pick and place.

10:49Jon Quast:Like John mentioned, the beak that can pick things up and move it to another place. This is not complex multi-step manipulation yet. Right now, it's a million demonstrations of a duck picking a block up off the table. There's still a lot of research advancement that would need to take place before this gets to the point where NVIDIA is hoping it to get to. But it's definitely a really interesting step in the right direction. And I look forward to seeing what both of you do with your micro ducks.

11:10Matt Frankel:Yeah, we'll see about that in the future. After the break, we're going to dip into our mailbag and talk about the work that goes into buying a stock. You're listening to Motley Fool Hidden Gems Investing.

11:25Jon Quast:As a podcaster, my voice is heard by thousands of people. And now with Vanguard Investor Choice, I can be heard by the companies I invest in too. Vanguard Investor Choice makes it easy for eligible Vanguard fund investors to have a say in how their funds vote at company shareholder meetings. With just a few clicks, you can set your proxy voting preference and make your voice heard on topics like executive pay, board director elections, and more. Investor participation is the heartbeat of a healthy corporate governance ecosystem. You have a voice. Let it be heard. Visit Vanguard.com slash Investor Choice to learn more.

12:04Jon Quast:Vanguard investors own shares of Vanguard index funds and those funds own shares of the companies they invest in. Available for Vanguard index funds that participate in investor choice. Vanguard Marketing Corporation Distributor. Trading at Schwab is now powered by Ameritrade. Unlocking the power of Thinkorswim, the award-winning trading platforms loaded with features that let you dive deeper into the market. Visualize your trades in a new light on Thinkorswim Desktop with robust charting and analysis tools. All while you uncover new opportunities with up-to-the-minute market news and insights.

12:36Jon Quast:Binkerswim is available on desktop, web, and mobile to meet you where you are. It's built by the trading obsessed to help you trade brilliantly. Learn more at schwab.com slash trading.

12:47Matt Frankel:Welcome back to Motley Fool Hidden Gems Investing. We're going to double dip into the mailbag today. And up first, we got a question and it really revolves around this concept of an investment thesis. Before we read the question, I don't want to take for granted that everyone knows what an investment thesis is. So, Matt, give us the quick rundown on this term.

13:08Jon Quast:Yeah, a lot of people don't know what the term means or they have the wrong definition of what it means. So it's definitely worth clarifying. So in simple terms, an investment thesis is your reasoning behind why a particular business or stock might be worth more at a later date than it is today. That's the simple one sentence description of it. it needs to be specific enough that you would know pretty quickly if you were wrong. So an investment thesis should answer three questions. One, what needs to happen for me to be right? Number two, when do I expect that to happen? And number three, what would prove me wrong?

13:42Jon Quast:So that's kind of my general overview of what an investment thesis is, but there's a whole lot more to it than that.

13:48Matt Frankel:All right, well, let's get into this question here from Ravi. I'm not going to read it word for word, but I am going to just kind of run down. Ravi says that they have a stock portfolio of roughly 40 to 50 stocks, and they have an investment thesis before buying. And they have five key questions that they ask themselves here. The first one is, why buy this business? Why now? Reasons for future growth? What are the major risks slash bear case scenarios? What disconfirming evidence should I watch for? Now, Ravi, we're going to get to the first three things there. That's more of like, what for the stock that I'm buying?

14:23Matt Frankel:And then the second two seem to be more like, what not do I like about this stock or what is it going against this stock? So let's focus on those bearish ones first here. What are the major risks, bear case scenarios, and what disconfirming evidence should I watch for? The reason that I like this is that is something that's often overlooked. A Charlie Munger quote comes to mind. He said, I'm not entitled to have an opinion on this subject unless I can state the arguments against my position better than the people do who are supporting it. So in other words, you need to understand the bear argument better than the bears do before you can have a bullish thought on a stock.

15:01Matt Frankel:So I really like that. And I feel like that goes right along with the spirit of Charlie Munger. And so Matt, you're up first here. What do you think about Ravi's process here? And what do you have to say about how we incorporate that bear thinking into our theses?

15:16Jon Quast:Yeah, I love that Munger quote. And I've said before that if you haven't tried to talk yourself out of an investment, you shouldn't buy a stock. So that's kind of similar idea here. So first, just to give some credit where it's due, that five question process is better than how some investment professionals I know structure their investment thesis. So pat on the back right there. I'd reframe that question five about disconfirming evidence just a little bit. You're not looking for just bad news. You're not looking for a bad quarterly earnings report, for example. You want to identify specific observable things that would make you sell.

15:46Jon Quast:For example, if Google Cloud Alphabet's a stock that I have in my portfolio, if their cloud revenue were to unexpectedly decelerate, specific and measurable, or if it started to lose market share to AWS and Azure, specific and measurable, it could prompt me to sell the stock as cloud growth is a very big part of my investment thesis. So as far as a sustainable cadence is concerned, like use your question three reasons for future growth when you're doing this. So for each reason for future growth, identify a specific trackable metric, kind of like your specific reasons for selling. If part of my thesis is that a company's operating leverage will improve as it scales, meaning that it's going to get more efficient over time, I might track operating expense growth versus revenue growth.

16:29Jon Quast:And my tripwire, as I would call it, would be if operating expenses grew faster than revenue for more than two consecutive quarters. So that allows for a natural quarterly cadence. You don't have to obsess about your stocks every day, every week, anything like that. And it's a very specific observable point to track how your thesis is playing out. Yeah, I think that those are all really good points. I mean, Charlie Munger felt so strongly about this. And I love that quote you referenced, John, because I think he knew that our brains as humans, as investors, were naturally itching to be right. And that can make us vulnerable to confirmation bias.

17:02You know, when you fall in love with a company's story, you can sort of subconsciously filter out bad news and maybe actually only focus on the positive metrics that make you feel good about holding the stock. It's a natural inclination. But I think Munger's rule, there's some psychological inversion that comes out of that because it stops you from acting like a cheerleader for your own portfolio. It forces you to think like an inspector looking for damage. And this is something that I incorporate into my own investment approach. When I am approaching an investment thesis before buying a stock or even when reviewing holdings in my portfolio, for me personally, I don't consider that thesis complete until I can either physically or at least in my mind write a convincing one-page short report, if you will, on my own position.

17:44So in my mind, I'm thinking if I can't state a short seller's argument with total clarity, maybe I don't fully understand the risks I'm taking or the business I'm buying. And that kind of really helps me take an honest assessment of the risks that a particular business presents. And, you know, there's a lot of ways to practice this without being a professional analyst, you know, without getting overwhelmed. I personally like to take a look at the risk factor section of a company's annual 10K filing. It's not particularly exciting. And, you know, there's ways to use AI to help kind of filter through some of those and summarize that information.

18:15But it's really informative and maybe a seemingly dry part of a company's financial reports. But that's what tells you what keeps management up at night. And I also like to go and seek any bearish research I can find from independent analysts. It doesn't necessarily change my decision to buy a stock in the end, but it helps me have a much more holistic picture of the business I'm buying and one that I hope to hold for a long time. So I think that treating your thesis as a living document and really taking an honest look at both the bear and the bullish sides, that can help you succeed in the long run, but it can also help you catch maybe a failing business before the rest of the market does.

18:52Yeah.

18:52Matt Frankel:Ravi, we love this question. We love your frame of mind. And also, I just want to point out 40 to 50 stocks in the portfolio, that does a great job of mitigating a lot of risk in and of itself right there, just that diversification. So good for you on that. And we're going to answer the rest of your questions in the next segment as we answer another question from the mailbag. You're listening to Motley Fool Hidden Gems Investing.