Motley Fool Hidden Gems Investing - The NVDA Show

Episode Date: February 22, 2024

What does Nvidia’s blockbuster quarter signify for the future? (00:21) Tim Beyers and Deidre Woollard discuss: - What is fueling Nvidia’s growth? - How far the demand for GPUs can stretch. - What concerns Tim about Nvidia’s capital allocation. Companies discussed: NVDA Host: Deidre Woollard Guest: Tim Beyers Producers: Ricky Mulvey Engineer: Dan Boyd Learn more about your ad choices. Visit megaphone.fm/adchoices

From morning hockey with a cup of coffee, to Timbits and road trips, Tim's and Canadian Tire have always gone together. Now it's official. You can now earn Canadian Tire money at Tim's. Link your Triangle Rewards and Tim's Rewards accounts to earn twice with every Tim's run. Terms and conditions apply. Visit timhordens.ca slash triangle for details. Now boarding the NVIDIA Rocket Ship. This is Motley Fool Money. Welcome to Motley Fool Money. I'm Deidre Willard here with Motley Fool analyst Tim Byers. Tim, how's your Thursday going? Fully caffeinated, ready to go,

and apparently I need to be because there are caffeinated stocks. There are caffeinated stocks, or rather, there's a caffeinated stock that seems to be supercharging the whole market. We got to talk NVIDIA today. We're going to talk about it for the whole show. It felt like the market was just like poised, ready, waiting for this moment. And, you know, I almost don't want to talk about the numbers because it's just, it's all these giant numbers, but we got to do it. So, you know, revenue up 22% quarter over quarter, but up over 265% from a year ago. For the full year, it came in with nearly $61 billion in revenue. So we know AI is the story here, but everyone seems to want to know,

is this sustainable? What do we make of this? Oh, no. These growth rates are not sustainable. No one should kid themselves and think that these growth rates are sustainable. It will slow down. Law of large numbers never fails. Just like, as my friend Carl Thiel likes to say, gravity is responsible for most falling deaths. Law of large numbers is always responsible for a slowdown in growth rates. That is inevitable. So it won't keep going like this, but it probably will keep going at very high rates of growth for longer than we expect. And that is what gets people excited about a company like NVIDIA, Deidre? Excited and also a little bit worried. Let's talk a little bit about what's driving this growth. On the earnings call, a lot of talk about

this transition from general to accelerated computing. You've still got a supply problem here. You've got just massive demand and the orders just seem to keep coming in. Yeah, that's true. You could see it in the financials. I would direct folks, if you really want to see how this plays out, maybe if you like learning a little bit more about financials and you want a quick one, whenever you see this divergence, take note of it. Here's where you'll find it. On the balance sheet, take a look at the line that shows accounts receivable net under current assets, and then right underneath it, the line that says inventories. What you're going to see, Deidre, is that the accounts receivable number goes from

about $3.8 billion in January 2023 to almost $10 billion in 2024. Inventories are up from $5.2 billion to about $5.3 billion. But basically, what that means, that massive divergence, that's a financial way of seeing what you just said, which is this. NVIDIA cannot keep enough product around to meet the demand that it has not yet turned into cash, because that's what a receivable is. A receivable is an order that hasn't yet turned into cash or it's turned into cash, but we can't yet recognize it because it's a sale over multiple periods or something like that. But when there's a receivable, there's a mismatch between the orders, the amount of orders, the amount of demand and what's actually come in, in terms of real revenue. And so, yeah,

there's way more demand for NVIDIA than they can possibly fulfill at this point. That's another thing that gets folks excited and why they think growth can persist for a longer than expected period of time. And I'll tell you, the revenue number was extraordinary, but even bigger than that is the data center business, which is where all of this AI is going. That was up 409% year over year. Unless I have my numbers wrong, Deidre, I believe that the data center business accounted for about, let's call it $61 billion of revenue during the quarter. I believe the data center business overall was responsible for pretty close to $47 billion of that. That is a huge, huge proportion here. No, I'm sorry. During the fourth quarter, data center was $18.4 billion.

That $47 billion, yes, that's for the full year. So I got that wrong. But still, you're talking about close to a third of the business is data center. That's up 27% from the previous quarter. And again, it's 5x year over year. Just outrageous. Well, you said something that I want to go back to longer than expected. Now, longer than expected doesn't mean infinity. It doesn't mean that this goes on and on. And I think that's one of the things that, that anyone who's an investor is wondering, like, what should we be looking for as a sign that maybe this is shifting? I mean, if you listen to, to the, to Jensen Wong talk about it, he's, he's, he's, it seems like there's other businesses that are coming up behind this, which should, which should keep growing NVIDIA.

but it's not going to be all GPUs all the time forever. No, not forever. Where you would start to see, I'm not going to call this a warning sign, but let's say where you would start to see things start to normalize is in the inventory line. Like I told you, Deidre, the inventory has been, you go year over year and looking on that balance sheet, it's flat. In other words, we just, you know, it's, it's not like the, everything, it's the opposite of everything must go in a clearance sale. It's like, we can't keep things on the shelves. You better get in here and buy now. When you start to see something that's a little more normal, like, you know, revenues up, like say 30% and inventory is let's say up 25% or also 30%. That feels a little bit

more normal. Then that would be like, okay, things are starting to normalize here. We should expect growth rates to maybe settle. Where you would get more worried is if you saw that inventory line growing way faster than revenue or even worse revenue and worse than gross profit. If gross profit is, say, growing at 60%, and revenue is growing at 55%, and inventory grew at 125% over the same period, you say, well, wait a minute. Maybe they don't have the math right here. They are still anticipating high demand and that demand is slowing. That's where you might see a little bit of room for concern and the stock maybe hasn't cratered yet. But more likely, if you start to see something normalized, then you should be like, okay, this is probably the

growth story. It's going to be growing for a very long period of time. This is a very resilient company and no way is it going away. I think we're going to need GPUs for just short of forever. but will we need them at this rate? No, we will not. Where it'll start to show up, maybe a little bit of cracks in the armor would be if the inventory starts really growing quickly, and the revenue and the gross profit growth rates are not really following. Those start to get disconnected, and then we'll see, okay, maybe we need to recheck things a little bit. it. I want to make sure that I understand this. The inventory is, at this point, mostly GPUs, and GPUs is what's driving the business. Is that correct? Absolutely. Yeah, absolutely. There's

no question. Let's be clear here. The way I think of NVIDIA's inventory, because they don't do all of their own manufacturing, and they don't make their own chips. Taiwan Semiconductor makes those chips. But they will do things like source components. They'll have third parties, presumably, that are doing assembly and then shipment. So there's a part of the inventory process that they'll take control of. They own part of the supply chain because they're a hardware company, but they don't own all of it. They don't build all of their own chips. They don't have the fabrication facilities, but they are still a hardware company that is responsible fundamentally for assembling things like systems that go out the door. So they are taking control

of supplies and making sure that equipment and orders are fulfilled. They are responsible for that. Okay. That makes a lot of sense. I want to talk a little bit about Jensen Wang as a CEO. And I know you have maybe some concerns about him as a capital allocator. Tell us a little bit about that well there's an infuriating part of this which so it's a good report that is also and i'll give you the hot take here there's a part of this report as great as it is that is just horrible and i'm not really trying to be facetious here i really do think it is i'm gonna say it's just poor judgment on Jensen Wang's part, which I do not like to say because I think he's a great CEO. But I think in this one, big swing and a miss here as far as I'm concerned,

Deidre. Here's what I want to explain. NVIDIA is buying back quite a lot of stock. NVIDIA also issues a lot of stock to employees. Now, according to the cash flow statement, in the trailing 12 months, the fiscal year ended in January of this year, their stock-based compensation expense was about $3.5 billion. They also, over the year, bought back about $9.5 billion in stock. That $9.5 billion, some of that went to common shareholders. A lot of it went to offsetting the dilution that would be created by NVIDIA giving a lot of stock to employees. That's not great. Now, to be fair, NVIDIA did retire some shares over the fiscal year. I think it's about $100 million, if I have it right,

in shares. You could argue that that $9.5 billion went to buying shares back at a lower overall price, but a good portion of it, probably a third of it at least, was to buy back shares that were issued to employees. You essentially took money from shareholders, gave it to employees, and then you're pretending that you're giving money back to shareholders, but you didn't. You just moved things around, moved the deck chairs around a little bit. I think that's That's a bad use of capital. I really, really dislike that. The thing that is frustrating to me is, I don't have a problem with Nvidia wanting to reward its employees for doing incredible work. In fact, I think they should do that. Here's what could have been great.

Really awesome. Just take that $9.5 billion. If you are insistent upon offsetting dilution, just offset the amount that is, or even honestly, don't even do that. But if you're going to insist on it, just match it to the stock-based compensation expense and then use the rest of that money, $5, $6 billion. You know what you could do? You could do two things that are amazing, one for employees and one for shareholders. The first thing, pay a bunch of bonuses in cash to the people that have really done the work to get you where you are. And even better, Nat, if you've got that kind of capital and you are trying to, I know we're going to talk about software, but you're trying to build out a software advantage, go find the best software

engineers you can find that are going to help you and pay them a ridiculous amount of money and steal them from your competitors. How's that for an idea? Instead of putting this into buying back stock. I just dislike that. Now, here's the thing you could do for shareholders. Right now, Nvidia pays a very meager dividend on an ongoing basis. It's tiny. It's almost insignificant, really. In fact, I would say it is insignificant. But you know what? You have an option, Nvidia. You could have a special dividend that ties you down to nothing. It ties you to nothing. And you could put like $2 to $3 billion into a special one-time dividend because we are amazing right now. We're generating way more cash than we need, and we could take $2 billion to $3 billion

and just reward common shareholders and not pretend that buying back stock, when you're also really offsetting dilution, is rewarding shareholders, because it isn't. It really isn't. But if you gave me a one-time dividend as a shareholder, phenomenal. Love it. And you don't even have to repeat it. You can just do it one time. I think you can tell that really bothered me about this report because Nvidia has a lot of cash, tremendous amount of, I mean, it has more than enough capital to do the work that it needs to do. So I have no problem with them putting additional capital to work in other ways to benefit shareholders. I think the way they chose to do it with the buyback, I'm probably hammering them more than I should because they

probably bought back stock at attractive prices given where the stock is today. But I think there were way better things to do. There were so many better things to do and you still could have bought back some stock deeper. You just found out that your sales team is at risk of missing quota. Don't panic. Just ask Rippling AI. Since it's built on your real-time people and business data, Rippling AI can pull metrics from Rippling and Salesforce into a meeting-ready dashboard showing quota attainment, headcount plan, and monthly revenue to quota by region. In seconds, you'll see exactly what's behind your quota risk and fix it before it's missed. Question answered, action taken crisis averted when you have critical business questions that need answers don't just

file a ticket and wait weeks for an outdated report describe what you need and have rippling ai build it instantly from your live people and business data whether it's a dashboard with detailed charts or automated workflows with the right triggers conditions and approvals ready to rule your business head to rippling.ai slash fool to get the only ai built to give you full visibility and take complex actions across your entire organization. That's R-I-P-P-L-I-N-G dot A-I slash F-O-O-L. Sign up for exclusive access today, rippling.ai slash fool. Welcome back to Motley Fool Money. Now let's get back into my conversation with Tim Byers about Nvidia's earnings. Let's switch gears a little bit and talk about some of the future

for Nvidia, because as you talked about with the inventory, the GPUs, eventually everybody has, if not enough GPUs, the demand isn't going to be what it is right now. They're going into some other areas that I want to make sure that I understand as an investor. One of them is they talk a lot about Ethernet networking for AI. And they also talked a lot about software. And I want to make sure I understand the software component here, because Jensen Wong was saying the NVIDIA AI enterprise becomes this kind of operating system for AI. Now, if that's the potential, that sounds really huge. But is that what this is? I think operating system is a little bit of a misnomer here. I know why he's saying it. He's talking about orchestrating all of the things that

go into making AI functional in an environment, like providing all the software tooling, providing management, providing the hardware, providing some of the backbone technology. So you mentioned the Ethernet here. It's easy to forget that NVIDIA doesn't just make chips. They also have systems that are not fully functional primary compute systems. They are still what NVIDIA calls, you know, they use this term accelerated computing. Accelerated computing means, you know, when you have, using the car analogy here, you know, your engine is your general purpose compute. That's the CPU.

Every car needs an engine. But if you're going to turbocharge it, you know, you have other components. maybe you have like nitrous oxide in the back and that's to accelerate give more speed more power things like that nvidia makes those turbocharged components not the engine that's not what they do so they make the accelerated i'm using air quotes here a listener the accelerated parts of the compute that exists inside a a large-scale environment like this that's what they do And their argument and Jensen's argument is that those accelerated components are for the general purpose task of AI are way more important. They are the things that you must have in order to do AI well.

And to do other versions of large scale cloud compute where you're putting a lot of compute to bear on some very maybe like a very big workflow across a lot of regions around the world. And you're doing that in a cloud environment. So accelerated computing really matters. And accelerated computing equals GPUs. But it doesn't just equal GPUs. So again, let's get back to Ethernet, what this means. In order for an accelerated environment to really function, it's not just that you compute a lot of data. You also have to transmit it. You have to network it. you have to bring data into a large-scale environment. How you network systems, GPUs, all this together really does matter. A few years ago, NVIDIA invested in a technology called Mellanox. They invested in, essentially, a backbone technology that was for connecting

lots of GPU compute together in a networked environment. What he's talking about now The classic technology that NVIDIA has used for a long time is called InfiniBand, which is very precise. I want to connect together in a networked environment, a lot of compute systems, and they're going to be super precise. They're going to be really efficient. It's like if I whisper into your ear a particular thing and I want to get a message across, that's a very direct way to do it. And you're going to get that message, because I've been really clear here. I've gotten as close as I need to get, and there's very little chance that you're going to mishear me. I got close, and now I've communicated what I needed to communicate very precise, infiniband.

Ethernet, Deidre, is really messy. The way that somebody, when I was first learning networking, the way a friend described it to me, it's like shouting across a crowded room where it's noisy. That's Ethernet. Ethernet is designed as blunt instrument networking. I'm just going to send a bunch of packets, and I'm just going to keep sending them. And if it doesn't go through, I'm going to send it again, send it again. I'm just going to flood the pipes with data, lots of packets. And I've done enough of that, that even if there are losses, the message is still going to get through. In other words, I'm going to shout loud enough that you can hear me. Does that make sense? So, what NVIDIA is saying is we need to

have a more general purpose. Ethernet is everywhere. And if we want to network and we want to network AI, we probably should have a good way to use Ethernet. Let's make Ethernet better. In other words, to use the analogy again, let me give you a bullhorn so that if you have to shout across the room and you shout with a bullhorn, I'm going to hear you or you're going to hear me. Does that make sense? That does make sense. So I think they are putting a lot of these tools together to orchestrate and make AI environments more fruitful, easier to build for, easier to manage, easier to expand and plug into other environments, easier to network. All of these things matter for expanding NVIDIA environments for AI. You can't just have the chips. You got the

steak and you got the trimmings. A steak is good, a steak and trimmings is better. This is what he's saying. We got really good steak with those GPUs, and here's a bunch of trimmings. We're going to give you a whole meal. Does the trimmings become more of the meal over time? Or is it still going to be the steak, and this is always going to be the trimmings? I don't know. That's a really good question. You're probably going to see much more of a buffet. That's because environments will change because they always do in tech. But it's a really good question. I don't know the answer to that. But I think one of the encouraging things about Nvidia is that they've

been thinking about this at this broad level of we don't just make the chips. We need to make something that's bigger, that makes the entire environment better, starting with the CUDA software that they had all those years ago and just gives them such an advantage because there's a lot of developers that understand the CUDA toolkit. So in order to make AI real, they use these accompanying nvidia tools so nvidia has known for a very long time that it isn't just about chips it's not just about the steak it is about the trimmings and they've been doing that for years and that puts them in such a good position right now especially because ai really isn't about like chips generally it really is about gpus and making gpus available

in your environment. And that is, I mean, you could not be positioned better for that kind of movement than NVIDIA is. They're right at the center of it. And probably will be for some time to come. You would think so. I mean, it's hard to say, it's hard to value this company because it really is difficult to say how long these growth rates persist, but they're in a very, very good position. And so, if it's me, and I have some control over a portfolio, a real money portfolio in a Motley Fool service connected with interconnected opportunities called Cloud Disruptors. It's a real money portfolio, has a fairly substantial position in NVIDIA. I can tell you, Deidre, I am not in a hurry to sell NVIDIA shares here. I'm going to watch it

closely. I will promise you that. I am going to watch it closely, but I'm not in a rush to sell a business that is compounding and has good advantages to continue compounding. Yeah, really good point. Thanks for your time today, Tim. Thanks, Deidre. As always, people on the program may have interests 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 Deidre Wollard. Thanks for listening. We'll see you tomorrow. Thank you.