In this podcast, Motley Fool host Mary Long caught up with Motley Fool analysts Tim Beyers and Kirsten Guerra to discuss:
- Super Micro Computer's real business.
- The original bull thesis for the data center service company.
- Supermicro's relationship with Nvidia.
- What's behind the stock's recent performance.
To catch full episodes of all The Motley Fool's free podcasts, check out our podcast center. To get started investing, check out our quick-start guide to investing in stocks. A full transcript follows the video.
This video was recorded on March 03, 2024.
Tim Beyers: I'm paraphrasing here, but as the late great Charlie Munger would often say that your results are tied almost inexorably to your ability to control your emotions in situations where you need to, and if that's not you, this is not a party you want to go to.
Mary Long: I'm Mary Long and that's Tim Beyers, an advisor on The Motley Fool's Interconnected Opportunities investing service. Tim and his colleague, Kirsten Guerra, have been following the Supermicro story for a while now. You may have seen the data center services stock recently went ballistic. I caught up with Tim and Kirsten to find out what this business actually does, the machinations behind the stock's performance, and what investors should do when they see a stock break the laws of gravity. Tim, I want to start with you because talking tech can get really technical really quickly, and before we talk about Supermicro the company, it's maybe important to get a picture of the landscape. This company operates within the data center landscape and so before we start talking about the ins and outs of the business, can you give us an idea of the different types of data centers that exist and why it's important to understand that landscape before talking about Supermicro?
Tim Beyers: Yeah, there's a lot of different ways to engage with the data center. Let's use a building analogy. The data centers that most people are familiar with are the ones that are owned by the big Cloud providers, and you can think of the big Cloud providers, so that's AWS, GCP: Google Cloud Platform, Azure, and when you are accessing those data centers, those were wholly owned by those big tech companies and they rent space inside of their data centers so when you are engaging in the Cloud, you're using somebody else's computer. If you're engaging in AWS, you're using AWS's computers, and those AWS computers live in a data center somewhere which is a really large building, it has lots of equipment in it, servers, networking gear, storage devices. They have tons of wires and hold just lots of industrial equipment to cool that facility. You could think of it as a giant factory floor that may as well be silent and frigid because you don't want computers to light on fire. That's the picture of a data center. All of them look like that, so then it becomes, OK, what do they do and what do they offer? If buildings are roughly the same but they have different purposes; that's a way to think about data centers, so the AWS one is like the furnished apartment you rent right out of college. They give you everything. You don't need to buy any furniture, you didn't need to do anything. You just go in and you've got a place and it's all set up for you and you pay the premium for that. That's AWS, that's GCP, that's Azure.
Then there is Equinix. It's the data center that has a lot of equipment in it already. Equinix, they don't always do this, but there are these hybrid environments, and generally they're called Bare Metal agreements. You have lots of equipment, you have lots of things that are inside the data center and you want to control that equipment, so this is the condo. This is the condo arrangement. I have a place, I own it, but it's still a managed community. There is some stuff that's managed. Its condo, it's a town-home. Then there is the co-located agreement, which is, you get an apartment and you are going to bring all of your stuff to your apartment, you're going to have to furnish this. You're going to rent it, but you're getting the space, but you're paying the utilities. You are paying really everything. Supermicro fits most specifically into that third part of it. They're the ones that furnish the apartments with the stuff that you want to buy so that you have the essentials inside your apartment. You are bringing equipment you own that you have brought from Supermicro. Supermicro's selling you the bed, selling you the couches, all that stuff, and you bring that into the apartment. So, in this data center arrangement, you own the equipment, and the co-locator gives you the space, power, things like that. But you really want to control the experience and so you use Supermicro equipment because you are very interested in not handing the keys to AWS. You want to control the actual hardware experience.
Mary Long: Let's zoom in on that equipment picture a little bit. A small company called NVIDIA has been in the news a lot lately.
Tim Beyers: Tiny company, never heard of it.
Mary Long: Tiny company, that's what I figured. We could do a deep dive on that at a later date but, Supermicro rides similar tailwinds to NVIDIA, but they're allies, not necessarily competitors. You've used a sports car analogy to visualize this partnership, really diving into the analogies today. But Kirsten, if data centers are sports cars rather than apartments, what part of that car is Supermicro building that NVIDIA isn't?
Kirsten Guerra: I was going to say I love all the analogies. If we're talking cars now, NVIDIA is the engine and the engine is exciting and everyone wants to talk about the latest and greatest specs, but you'd never roll up to a car dealership and just ask for an engine, it's bigger than that. You want a whole car. If that car maybe has the fancy engine you want but it pairs it with a restricted exhaust that builds back pressure or a poorly timed ignition system, ultimately, you've paid up for this exciting engine, but its effective power is squandered from an overall inefficient build. I hope that makes sense. I don't actually know anything about cars, but if you do I hope that helps.
Tim Beyers: That sounded amazing though.
Mary Long: I know, I believed it.
Kirsten Guerra: Thank you so much. In that analogy, NVIDIA provides the engine and Supermicro is in the business of that overall build, so Supermicro builds full rack scale server systems which require not just compute from these NVIDIA GPUs but also substantial memory storage, networking capabilities, and Supermicro isn't necessarily manufacturing a lot of what they assemble, what they sell is the expertise in putting it all together. Supermicro is building these hyper-efficient systems that take full advantage of everything offered by those fancy new NVIDIA GPUs that the data centers are already paying top dollar for, so of course, they want to get everything out of them that they can.
Tim Beyers: Yeah, and just to build on that super quickly, it's not like you're buying from a standard dealership. When you're buying from Supermicro, you're buying a tricked out vehicle, that is for you. You are Vin Diesel and you want your tricked out car.
Mary Long: That's a great point, Tim, and I know we're jumping around with the analogies here, but I do think for the lay person they're really helpful in visualizing what these highly technical companies do. Because, now that I have a grasp on what it is Supermicro makes and how it operates in this space, that's great, but there are also other companies that do the exact same thing. To help a lay person understand, why would a customer to Supermicro over a bigger name like Hewlett Packard, Cisco, Dell, etc.?
Tim Beyers: Yeah, those are the dealers. They make a car that is very easy to understand. They are not tricking out a vehicle, they are mass-producing a model and those models can be useful, and you may even put them in a data center and you might even go find a mechanic who could trick out these base models that you have bought from one of those other providers and build you an environment that suits whatever purpose you have. But very often when you are going with a co-located agreement and you're building out an infrastructure, you have a particular purpose in mind. There are probably some very specific workloads that you're trying to serve and so in the case of Supermicro, because they do something that's much more custom, they essentially custom build what you need, and then they give you an agreement that allows you to upgrade those systems over time. They're very upgradeable. They're also the Carson's point before about the vehicle being something. If it's inefficient in some way, like the power distribution inside the vehicle, you have a bad power train or something like that. I don't know much about cars either. Let's be honest. You just aren't getting a good ride. In the case of Supermicro, they have some base components that are very useful. One of their biggest is that the systems that they build our hyper-converged. They're not just giving you a server, they're giving you essentially the all-in-one meal.
They're giving you everything. Some of the vendors you mentioned there Mary, they will sell servers and they might sell some networking gear, and they might sell some storage equipment, and they may do it through third parties and then maybe they assemble it with the help of a systems integrator. That's not what Super Micro does. They build you a custom hyper-converged system that has all of the pieces that are built together, shipped together, and done so with some very specific technology they've created to make their systems highly power-efficient. Which if power is one of your main costs, as somebody who's renting out space inside a co-located facility, you would like your systems to be power-efficient. That's a real draw to Supermicro as well. It's the customization, that's one reason why people buy from Supermicro. Another reason is they have relationships with all the major suppliers so when they custom build something for you, you know you're going to get a premium product because they have all of the premium agreements with all of the premium providers, including NVIDIA and third, the way they build their systems is highly efficient. In fact, so power-efficient that they are essentially world-class in that area so you do have customers who are like, we would like to be climate aware and so in those instances, buying a highly efficient machine like what Supermicro is offering can be attractive on that basis as well. It's not the main reason I think people buy Supermicro. It's more about the customization, strong relationships across all the major chip and components suppliers, long history doing it, and then the power consumption which saves you big money.
Mary Long: If you look at a stock chart of Supermicro, there is a staggering to put it mildly, run-up in recent months. We're going to get to that. But if we hop on the time machine and go back a few months before the staggering run-up, what was the original bull thesis for Supermicro?
Kirsten Guerra: In addition to the fact that this is a solid business for many of the reasons Tim just outlined, the bull thesis was really twofold. This is a company with huge tailwinds from the exploding demands of AI. When Tim and I were looking at this last August, we felt that was underpriced because last August, CEO Charles Liang started suggesting that Supermicro could get to 20 billion in revenue and just a couple of years and that was a huge jump from the seven billion or so in the preceding 12 months around the time he said that. We had to ask, can this guy be trusted? Because that's a huge, big if true. Right? After looking at.
Tim Beyers: It was an even bigger question than that, either he's a genius or an idiot. [laughs] Which one is it? I mean honestly.
Kirsten Guerra: Right. After looking into Liang's long history of issuing guidance at Supermicro, it was clear that this is someone who is quite accurate in his predictions, and if anything, he leans conservative when he does make statements like this. So we felt comfortable taking him at his word on that 20 billion projection, give or take. When you look at the math of that kind of medium-term growth, it looked very attractively priced at the time.
Mary Long: We talked about this staggering run-up. Let me put some numbers behind that. In September, shares of Supermicro traded at $231. February 15 of this year, $1,004 a share. It's bounced around since then and I'm not even going to attempt to name a number for today. We're recording this on Thursday morning. The price today could change in a few days time. The point is, things have happened and things are happening. My question, Kirsten, I'll kick this back to you. What in the world is going on here and has caused this just wild parabolic move?
Kirsten Guerra: Yes. Things are happening. [laughs] Anytime the market is valuing a company, there are elements at play of both the underlying business fundamentals and general market sentiment is a multiplier on that. On one hand, you have genuinely impressive business performance from Supermicro since the time we recommended it, and the year or so before, they are really seeing that revenue growth that Liang suggested and they're seeing margin expansion from operating leverage on that revenue. That's all business fundamentals. But it's that other component that maybe is getting a little out-of-hand here. Market sentiment around AI is generally very bullish right now. How could it not be? Valuation is all about predicting the future and projecting how much cash flow you think a company can generate into the future. When you have companies like NVIDIA in very related fields that are just blowing away earnings expectations every quarter recently with beats of 10% plus on earnings, yeah, people get a little excited. The market starts to feel like, wow, we don't even understand the potential ahead for this technology and the market gets a little frothy in that other side of the equation in market sentiment can go a little wild.
Tim Beyers: Can we also just talk quickly about the machines and how this works? What I mean by the machines is the machines of the market. When there are numbers that seem related to, in this case, Supermicro. Like the belief that NVIDIA has just an unbelievable long-term growth story and that Supermicro is going to profit handsomely from it. The machines all pile in at the same time and what has ended up happening, that market sentiment is essentially algorithms from institutional buyers putting way more money than has ever been put into Supermicro, into that stock at once. That drives it to crazy levels. That's what's happened. What you're seeing, if you can imagine the waves of money that are flowing into and out of Supermicro on a daily basis. But, the shoreline was relatively stable. I'm looking out over the horizon onto the ocean and I see waves but they're relatively minor, it's peaceful, good for a nice long walk on the beach. Now I'm seeing tidal waves and what the heck is going on and it's money that's moving.
Mary Long: But if I'm a surfer and I see tidal, to a certain person that can be appealing. Whether it's because of earnings, or machines, or tailwinds, or what have you, when a stock shoots up 300% in less than six months, why not just cash in on the gains and sell and walk away and enjoy it?
Tim Beyers: I mean, that's a great question. When we offered some guidance here, unfortunately with insider interconnected opportunities, which is a notional scorecard. Selling would've meant exiting the position entirely and I tend to not believe in that. We issued some guidance. It's a little bit in-between that hold on buying it and I still hold to that, hold on buying it. But if you have some, it's perfectly acceptable while we wait to see just how long the growth tailwinds for this AI trend are. It's perfectly acceptable to hold onto some, but if it had been a real-money portfolio and Kirsten can add to this here, but as we were talking about it, I think we tended to agree. This were a real-money position and let's say we had put $10,000 into a real-money portfolio and it had gone absolutely parabolic, we would've sold some. We may have even sold more than half, but we would have kept some as well. I mean, Kirsten you think that's fair?
Kirsten Guerra: Yeah. That's fair.
Mary Long: We're talking about tidal waves and wanting to surf. If I don't have a position in Supermicro, but I'm seeing this incredible run-up and I'm hearing about the tailwinds and the long-term growth story of Supermicro and other companies like NVIDIA that are benefiting from the same trends. It's tempting to see all that and want to join in the fun. Like to want to hop on the beach and start partying with everybody else. What's your advice to someone who sees that run-up and wants to join in?
Kirsten Guerra: Yeah, I would say just be cautious. Know exactly what game you're playing if you enter here. If you know that you're an incredible surfer, then that's one thing. If you know that you're not, that's another. I would say six months ago when we recommended the company, I would have described it as more of the balance of risk reward, they rested more solidly on business fundamentals than market sentiment. Today, maybe the opposite, it's a lot more sentiment-driven. That's not to say that this company can't continue to surprise, but I think investors today should have the expectations going forward of substantial volatility. If that makes you uncomfortable, where you don't know how to surf or you're just not someone who believes that the AI market ahead will continue to explode in ways that we can't predict, then this probably isn't the stock for you and there are plenty of other options out there.
Tim Beyers: Yeah, I could go further taking your analogy there Mary, since we're having fun with analogies. If you're on the beach and going to the party. Two things, recognize there's always going to be another party. You can always go to another one, call a friend, and find a different party. Or if you're going to go not to use the old man advertising line here, but please drink responsibly. I think would be the way to think about this. Don't get overindulgent here because it's very easy. Your emotions get away from you when you are getting into these types of situations. I think Charlie Munger, I'm paraphrasing here, but the late great Charlie Munger would often say that your results are tied almost inexorably to your ability to control your emotions in situations where you need to and if that's not you, this is not a party you want to go to.
Mary Long: There is not insignificant short interest in Supermicro. It's not quite Beyond Meat levels. About 12% of the float is short interest. In the grand scheme of things, is 12% a lot from your perspective? But beyond that, what are the shorts thinking? What's their case?
Tim Beyers: I mean, I have an answer for this. I'll tee us up here and Kirsten you going to add what you want. My answer to that is the short-sellers here, they have a very basic argument, Mary, they will tell you that this run and the growth that is expected as a consequence of the run-up is so enormous over the short-term that the opportunity for disappointment is very high and they expect to pull back here. I think it's primarily driven by valuation concerns and that does happen. There are short-sellers that will do that. It's very risky because valuations can reset incredibly fast. I tend not to be at all interested in valuation-based shorts, but there are many institutional money firms that do that. Another potential is that there's a real concern that hardware saturation is going to come much faster and the cliff will be much steeper than the market realizes. In other words, you reach a tipping point where a lot of companies say, OK, I think I've bought all the hardware I need for my AI workloads. Then you see what was extraordinary demand. If you can imagine it, you can imagine a line and then just an immediate 90-degree drop. I think that's the other thing that some of the short sellers are betting on. That is also a risk because you don't know when that's coming. It may come but there are no signs of that coming right now.
Mary Long: Tim, when NVIDIA reported earnings last week, you came onto the show and you raised some concerns about Jensen Huang's abilities as a Capital Allocator. A hot take I've heard you say before, that you think Super Micro CEO is a more skilled Allocator than Jensen Huang. What is it about Liang's strategy that you think is so impressive?
Tim Beyers: This is what I think and Kirsten correct me if I'm wrong here, but Supermicro is far less interested in buying back stock. They'll do it, but they do it very selectively. When they've done it, it has been at much lower stock prices. Which is exactly how you're supposed to do it. That's what you're supposed to do. You're supposed to buy back stock when it's cheap and you're supposed to sell your stock when it's expensive. I think that is roughly where his head is at. He doesn't seem to be under the same pressure as most of the big tech companies to be buying back stock, to do things like offset dilution. When he's buying back stock, he's buying back stock because he believes the stock is cheap and he's generally done pretty well with that.
Mary Long: We've talked about the wild ups and less wild downs of the stock over the past few months. But moving forward, if you already have shares at this Supermicro and you're intending to hold for the long term, as you two have outlined throughout this. Do you have any parting words of wisdom or mindset writing advice for people as we continue to watch this story unfold and anticipate whatever might come in the near future?
Tim Beyers: I mean, Kirsten this is perfectly teed up for you here.
Kirsten Guerra: [laughs] I would just remind people that this is why we take a portfolio approach. Supermicro is one company among many, it maybe promises a lot of opportunity, other companies do as well and you should balance that risk across a portfolio of ideally maybe 25 or so ideas, but yeah, diversify across a lot of your best ideas, not just this one.
Mary Long: How fantastically Foolish of you.
Tim Beyers: Don't be afraid to have small positions, that diversity is the number one thing you should do. What Kirsten said is spot on, but you can always have small positions. If it's a highly speculative bet, which is what Supermicro is right now, just keep it small. Don't spend a lot, don't put a lot of capital into it. If it goes bananas, you're going to be rewarded anyway because this is going to go bananas.
Mary Long: Kirsten, Tim, thanks so much for walking through this with me today. I appreciate your time and all the sage wisdom and awesome contexts that you have provided us with today.
Tim Beyers: Thanks, Mary.
Kirsten Guerra: Thank you, Mary.
Mary Long: As always, people on the program may have interest in the stocks they talk about and the Motley Fool may have formal recommendations for or against, so don't buy or sell stocks based solely on what you hear. I'm Mary Long. Thanks for listening. We'll see you tomorrow.
