In this segment of Business Breakdowns, we're tackling Nvidia - the company on one of the greatest winning streaks all all time.
But can it continue? Where might the company go wrong? What are potential pitfalls with the business model?
Listen in for how our team would think about Nvidia in a case interview or as an advisor or operator at the company. We answer the following questions:
- How does Nvidia make money?
- What does Nvidia sell?
- What is Nvidia's business model?
- What is the Nvidia revenue breakdown?
Business Breakdowns drops on the 1st and 3rd Wednesday of each month. Loving it or have ideas to grow the segment? We want to hear from you! Reach out by sending us an email and connect with Jenny Rae and Namaan on LinkedIn.
Episode Links
- More Business Breakdowns
- More info on Palm Tree careers
Transcription
Nvidia, it's the company on the biggest winning streak in business right now. You probably already know that Nvidia has accrued more market cap in shorter time than any other company in history. But we're not here to talk investing today.
We're here to talk about the nuts and bolts of Nvidia's business. And I will be the first to admit that I am not a technical expert, and I'm not an expert on Nvidia's business. The goal for Jenny Ray and I today is to break down Nvidia's business in such a way, so that we can increase and expand our understanding of what the business model is, how Nvidia makes money, and what it should be focused on looking into the future.
Before we look into the future, let's just look at current state for a minute, and get aligned on some key high-level data for this incredible company. Let's start with annual revenue. Nvidia's annual revenue is around $96 billion, and it's Q2 2020.
That's it, right? No big deal, right?
Just up from $95.5 million? Or what happened in the past?
So Q2 2024 for Nvidia generated $30 billion on the top line. That was 122 percent year-over-year growth. More impressive to me, Jenny Ray, is the operating income.
So on that $96 billion run rate, we have operating income of almost $60 billion. That's 68 percent operating margin. It's unreal.
This is a business that's growing quickly and has expanded its margins to heights that I've not seen many other businesses reach. Now, what the heck does Nvidia sell? You might have heard a lot, Jenny Ray, about this company, what they do.
They're involved somehow in this AI revolution that's taking place. But what is it that they actually sell? So Nvidia at a high level has really three lines of business.
So they have a compute and networking business that includes the chips and the GPUs that enable a lot of the AI processes and large language models that are being built right now. So chips and GPUs, they have a data center business, and then they have a graphics card business as well. So chips and GPUs that enable AI functions, chips and GPUs that enable non-AI functions, like graphics card, automotive chips, et cetera, and then data center revenue.
That's really what Nvidia does and the value that they provide inside of this ecosystem. So now that we have a sense, January, of just the financial metrics around Nvidia, what they do, what would you be thinking about if you were on Jensen Wong's leadership team?
Namaan, I was asking some really basic questions when I was thinking about this business. I really, I think a lot of people can talk about Nvidia, now we can pronounce Nvidia, and everybody knows you should know something about Nvidia. And the baseline of my understanding was no more deep than most people's would be.
But I literally had to ask, like, can you put the chips into a computer on your own? Is this, are there their own computers? Like, do you have a data center behind it?
I could not visualize tactically what Nvidia does. Nvidia is a California company, super proud of the growth that they've had, but I really wanted to dig in to understand how they got to where they are, and if they could either replicate that success across other business lines or not. So my first and fundamental question is always, what kind of business is Nvidia?
Is it a fixed cost or is it a variable cost business? And up until really the late 2023, Nvidia was a variable cost business, and a pretty good one at that. They made some of the best graphics cards in the industry.
They were supporting robotics. Their chips are top of the line, specialty, niche chips that are premium in certain categories. And so they have done an amazing job of taking technology.
There's of course R&D in the technology at a baseline, but there is incremental construction cost and delivery cost. Each of these are done on a price per unit basis for each one of their items. And so historically, Nvidia has been a almost entirely variable cost business, and their focus has been maintain margins, focus on premium segments, make sure that you have high prices.
From a cost perspective, they definitely have thought about suppliers and driving down cost of suppliers. But for the most part, it's really been price management. Like, how can I drive up the margin on each of my items?
So they would be the luxury brand of chip makers. And that's when we didn't know anything about Nvidia, or when I didn't know anything about Nvidia. I think there were probably people like gamers and other folks that have known about them for a while.
That's where they've been. What I'm really interested in about Nvidia, and if I were on the leadership team, this is what I would think about, is that Nvidia is shifting their entire business model. They're moving from a variable cost business, a chip producer, to a fixed cost business, a service provider, from a leveraged service perspective.
And that's really where you see this explosive data center growth, where they're recognizing that the utilization of their chips in market is only one piece of the service that people are going to need, and they are really building out that capability. And so, I would think about how we could do that across multiple different segments, and even segments that don't use our chips. So, that would be one of my key focus areas.
But the shift right now that we're starting to see, it's still not by a percentage of revenue over the tipping point, but we're getting really close, is that movement from the variable cost business to the fixed cost business. So, back to you. If you were on the leadership team, what are some of the metrics you would care about, knowing that we've been a variable cost business, we will be likely a fixed cost business going forward?
What would you care about? What would you manage to? How would you think about where Nvidia is?
Yeah. Jenny Ray, great question. The way that I thought about this is, how would I mitigate risk?
So, Nvidia is a fantastic business. There's a lot of good things happening. We have incredible margins across our different product lines and segments.
But this isn't necessarily a fairytale. There are still sharks in the water. And so, the first thing, if I was on Nvidia's leadership team, that would potentially keep me up at night a little bit, is our customer concentration.
Did you know that 40% of Nvidia's revenue comes from just four clients? Microsoft, Meta, Alphabet, and Amazon. So, there is concentration risk that I would be concerned about if I was at Nvidia.
Now, those four companies that I just listed are fantastic companies. They are high growth companies. They're not disappearing tomorrow, but they're also facing their own headwinds.
And when so much of my top line and so much of my growth projections are reliant on the capex of four companies, then there's naturally concentration risk that I would be concerned with. So as I'm looking at these premium chips, these luxury brand chips that I'm selling, one of the things that I'm at Nvidia that I'm intuitively understanding is that there's a limited market for my highest end chips. So not everybody is going to be investing in AI capability the way that those four tech companies are.
And so what I'm going to be focused on is maximizing my margin and my profit per chip, but also thinking about how I can mitigate that concentration risk so that I can ensure a more sustainable future for my business. And to me, the way that I think about mitigating that concentration risk is by expanding my data center footprint. Because not everybody in the market is going to need to spend the billions in CapEx that Microsoft is spending on NVIDIA's chips to power ChatGPT.
But there are a lot of use cases for a lot of different businesses around data center management and services, et cetera. And so I'd be thinking about and focusing on growing my data center business, which I would expect is already the highest margin business for me, to expand my customer footprint.
Third, I would also be concerned about supplier concentration. This is another concentration risk that Nvidia faces. So, TSMC, Taiwan Semiconductor Manufacturing Company, is the only manufacturer with the expertise to mass produce Nvidia's graphics processing units.
So these GPUs are used for both AI and video game use. There is one-
Can I stop you there? You're saying that there's one company that can produce this? Like we don't even have two or three, there's one in the world that can do this?
There is one company that can mass produce these at the scale that Nvidia needs to continue to fulfill its growth trends and expectations. So there's incredible concentration risk by having just one manufacturer who can mass produce these chips and these GPUs. And it doesn't help, and it wouldn't necessarily help my sleep to know that this supplier, this manufacturer is based in Taiwan.
And there's geopolitical risk that's also associated with having my biggest manufacturer be in that part of the world. And so we're not going to get into the ins and outs of geopolitics today, but I would really be concerned about customer concentration risk, about growing my data center revenue as a result of that, and then also seeing if I can diversify my manufacturing concentration as well, and not be beholden just to TSMC.
That's so good, Namaan, I really didn't know a lot of that. I didn't know that when you're thinking forward, that there are so many ways that this could go wrong for Nvidia.
I really don't feel like that's priced into this storybook stock either. So let me just talk a little bit about how I would think about managing it. On the first level, I think about managing to profit per chip.
And one of the first questions I asked, I think we were looking at some of the numbers. It was, I mean, extraordinary. It's almost infinite profitability per chip that they have.
And so, you know, thousands and thousands of dollars, like on the orders of tens of thousands of dollars. And then many of their customers, like you mentioned, I didn't know the concentration, but I knew that they were purchasing a large volume of these chips from them. But I still asked the question, and maybe this is an a-hole question, I'm not sure, but I asked the question of, like, why can't they price more?
Why can't they raise their prices by 25 percent or 30 percent more? If someone can literally go nowhere else and this is a critical pathway issue for their business, like, what is the problem with extracting more value from their customers? And when I'm thinking about this as a legacy and a future variable cost business, I think like that.
There's some customer goodwill, there's some limit to what you do. There's some tipping point for that, but I'm just not sure that they've reached it. And I would potentially recommend that they would think about their prices.
I think that on a cost per chip, it's already so low that it's really just a pricing question. And then I think that you've already probably identified a lot of the categories of demand for these current areas, but you're certainly looking all over the world for opportunities of demand that you haven't thought of before. So profit per chip is my number one.
Revenue per data center is my number two. So similar to what you talked about with the data center, but because the data center is a fixed cost business and I probably know how much it costs me to set up and how much it costs me to operate it, I either switch it on or I switch it off. When I switch it on, how much throughput am I using?
How many customers am I gaining? And how much revenue per customer am I utilizing? So on those data centers, I'm absolutely interested in that business.
And when I look at their cash flow, 60 something billion dollars of profit in the last year, that's where my money is going. I'm building data centers and I'm working on customer acquisition into those data centers. I'm building long contracts.
I'm developing a business that will be sticky way, way beyond my chips. And the number three are the customers serve for data centers. So I love that you mentioned concentration currently in the business, because I would manage to a lack of future concentration.
I'm concerned if I run a data center for Alphabet. I'm concerned if I run a data center for Google. I would like to have a data center that has 10 customers that are all working into those data centers.
And I would love to have some kind of leverage with my customers, and I'd like to have some operating safety, knowing that we are not necessarily basically just like a wholly owned subsidiary of one of these operators. And so I would think about the concentration per data center as I'm thinking about managing to these going forward. So overall, I really think that there's a lot of opportunity from a pricing perspective, potentially more than Nvidia has even captured.
I think they're in a moment where they're figuring this out, and I'm really looking forward to see what happens. So as we conclude, Naman, what's one of your hot takes for Nvidia? What is nobody talking about that they should be talking about?
What's something that they should be managing to? What are some of the issues that you think they're going to face that nobody sees right now?
Yeah, I've got two hot takes for Nvidia. Number one is I think by the end of the decade, Nvidia could be the largest or one of the largest energy consumers on the planet. These data centers that they're rapidly building out are incredibly energy intensive and energy hungry.
And so you might see Nvidia start to try to manage energy and power costs in a different way and start to manage to that metric in a way they're not managing now, if the data centers are as power hungry as I think that they're going to be. So if we were doing this episode five or ten years from now, that might be one of the metrics we would have mentioned that we are not mentioning today.
And to be clear, that's still a fixed cost, right? We are still going to have a baseline of energy consumption, and you're going to have incremental ones as you open up new data centers. But per each one, it's just you turn the lights on, it's energy hungry, you turn them off, it's not.
But there is a baseline of power that they're going to have to be able to supply themselves.
And that baseline is going to go up and to the right very, very quickly. The second hot take that I have for Nvidia is really not a hot take as much as it is a prediction, which is that this incredible free cash flow that Nvidia is realizing, they are going to reinvest in staying at least a generation ahead of the competition when it comes to these chips. So Nvidia's Blackwell chip that they're selling now is really compared to what the rest of the market has, next-generation technology.
And if I was at Nvidia, one of the things I'd be thinking about is how I'd reinvest some of that capital instead of returning all of it to shareholders into continued R&D to continue to build and expand the moat that I have around my chip business as I'm building out these datacenters for the future.
Namaan, those are so good. They're better than mine, to be honest. I feel like the only hot take that's truly hot on Nvidia is Nvidia is overblown.
It's not going to be that great. There's no reason that you're going to see Nvidia's name in three years anywhere on any charts. And I just fundamentally don't believe that, so it's a little bit of a rough hot take to take on, but I feel like that's the only hot take.
The only hot take that's contrarian is Nvidia actually has these kind of major risks. So my hot take is a little bit more nuanced. What I think is that overall with Nvidia, what you'll see going forward in the future, if they are successful, is a complete services business that is married to their chip business.
I think right now it is they're dating. Like the chip business and the data center are there together because they have to be, because there are no other providers in the market. But I would like to see Nvidia capture mandatory backend services tied to their chip services.
And that chip deployment, I think would create a tremendous amount of stickiness and opportunity. Right now, they're being flexible with it. I think they have an opportunity to lock that down.
So I would love to see, and from a hot take perspective, I don't think they're thinking about this. I do think there are risks to doing that, but I think that the opportunity outweighs the risk, and I think their ability to gain such significant share of the market in this entire managed services space really is contingent on creating that marriage. So that's it.
I hope you enjoyed learning a little bit about what Nvidia does, about this business that's powering things that you may not have even known, like a computer that you use, or a game console that you're working with, or a robot that comes and delivers you food. All of these things are powered by Nvidia today, and if you can imagine that AI might do more tomorrow, Nvidia is probably going to be right there along with it. So if you've got thoughts about Nvidia, if you think we've got something wrong, which we probably did, we would love to hear about them.
But overall, what we want for you from these episodes is to be able to gather business information, insights, and key metrics to understand simply what a complicated business actually does. And if you got that from here, then that was an episode of success. We hope that you join us for more.