NVIDIA Corporation (NVDA) Earnings Call Transcript & Summary

Okay. Thank you, everybody, for joining us. I will start off with a forward-looking statement, and then turn it over to John. As a reminder, this presentation contains forward-looking statements, and investors are advised to read our reports filed with the SEC for information related to risks and uncertainties facing our business. With that, I'll turn it over to John.

Thank you very much. It's my pleasure this morning to welcome Colette Kress, the Chief Financial Officer of NVIDIA Corporation. I know Colette and I would both prefer to be in Scottsdale right now. We will definitely be back there next year, knock on wood, but I really appreciate, Colette, you joining us virtually this year and supporting the conference and getting out in front of your owners.

I usually like to start the fireside chats with more of a longer-term strategic question, but we're sort of less than 2 weeks removed from your quarterly earnings, and I think there's still some unanswered questions that investors have on the October quarter that might be helpful to go through initially before going into some longer-term questions. The first is really on the gaming side where you're clearly in the midst of an extremely strong sort of product cycle with the RTX 30 Series, up almost 40% sequentially, up almost 40% year-over-year. A couple of things I'd like to ask. You talked about sort of there being supply constraints, and you mentioned it wasn't just your own silicon. I'd be curious if you could elaborate a little bit on the extent of those supply constraints, and importantly, when you think supply will catch up with demand?

Yes. Thanks, John, for the question. So in Q3, a very busy Q3 for us, really exciting for us as we continued our overall Ampere architecture. We took that opportunity to take our Ampere architecture to GeForce and produced our GeForce 30 Series cards for the market. Well, they came out, and we would characterize the launch and what we were seeing as overwhelming demand. That demand for GeForce for gamers is very strong for both laptops, for desktops, but the channel remains lean. We do have supply constraints, and our supply constraints do expand pass what we are seeing in terms of wafers or silicon. But yes, some constraints on substrates and components. We continue to work during the quarter on our supply, and we believe, though, that demand will probably exceed supply in Q4 for overall gaming. We do expect it probably to take a couple months for it to catch up to demand. But at this time, it's really difficult for us to quantify. So we stay focused on trying to get our cards to the market for this very important holiday season. And each day, things continue to improve. But towards the end of the quarter, we'll be able to provide some more information.

Colette, you guided the January quarter for gaming up sequentially. Now that includes an extra week. So despite some of these supply constraints, you are expecting a strong January. I know you only guide one quarter out. But as we look forward into the April quarter -- I'm just kind of curious, you've got the headwind of losing that week from January to April, but you also have supply coming online. And I think more importantly, you continue to introduce the RTX 30 Series down the stat. You'll have more notebook. You'll just have more general product availability. April tends to be a seasonally down quarter to begin with for gaming. How should we be thinking about that this year?

Yes. So I think you've highlighted a lot of the different factors that we take into consideration each year and the overall seasonality that is sometimes there for gaming. Our gaming business is broken down into 3 major components, which is that desktop, which is the notebooks and the overall consoles. Each of them having different characteristics as it goes through seasonality. We've also discussed that we have often seen our strongest overall demand focus around the holidays. But keep in mind, the holidays, in terms of what we're referring to, are holidays that start as early as October and move into the new year through to the overall Chinese New Year. So it's an important piece. But I think there's very important statement that you made there. That is correct, we are ramping RTX 30 Series, and we're just starting. So we're in the brand-new era of an amazing overall architecture and a performance leap that is larger than any other performance leap that we have done. And the overall reviews and reception of our launch has been quite strong. So we're going to watch in terms of how we see between Q4 and Q1 transition. It has normally been sequentially down. That extra week in Q4 should have, could have enhanced that overall decline. But we're just going to have to see where demand is once we get the supply more intact.

Colette, I almost hate to ask the next question, but I'm getting asked with more frequency than I would have thought. With bitcoin almost back to kind of peak levels in calendar year '18, I'm getting asked the question of whether or not we should be at all concerned that some of the strength in the gaming business was bitcoin crypto-related. Any comments you can make on that, especially in lieu of some of the channel issues you had back in calendar year '18?

Yes. Our focus in terms of really understanding our channel over the last couple of years post-crypto has continued to be a focus of ours and enhancements to really understanding both levels in the channel and in terms of real-time or as close to real-time information that we can get in terms of how the overall channel is doing. On your crypto question, we have heard some interest from the channel, but nobody is aware of any real demand at this time for crypto.

That's helpful. The other near-term question from the October quarter was just on the data center business. I think you did an excellent job kind of explaining on the call and afterwards the influence of Mellanox in October and then going into January. I think one of the concerns that I've heard post the conference call is how should we think about the core growth rate of the core data center business? I mean, clearly, you're going through some product ramps there as well. The A100 is ramping up, the B100 is ramping down, but you've seen growth rates on a year-over-year basis decelerate from an unsustainably high 200% plus to now a very respectable 40%. But if you kind of model conservatively sequentially from here, you're very quickly getting down to sort of maybe that 20% year-over-year growth. How do we think about kind of normalized year-over-year growth in data center? And what impact might the macro be having on the data center business today?

Yes. So for the rest of the group, let me try and summarize a little bit more in terms of what we experienced in Q3 and how we overall guided Q4. Our overall Q3 results in data center, yet again, another record level, not only a record level in terms of NVIDIA compute, but an overall record level in terms of Mellanox. We had expected our overall sequential increase to stem primarily from our NVIDIA core compute, but we also saw strong growth in terms of Mellanox between Q2 and Q3, largely due to a China OEM and purchases that we didn't feel would repeat in overall Q4. So the record levels of Mellanox, we couldn't be more pleased with their performance in the first couple of weeks, first couple of quarters here at NVIDIA. They have done tremendous. And we've been even excited in terms of all of our integration efforts in terms of working on future products together. So our compute business, as we think about NVIDIA compute, continues to ramp with overall A100. So A100 started out several quarters ago in terms of launching to our hyperscales around the world, and many of our hyperscales sat at speed-of-light quickness in terms of creating overall cloud instances with the overall A100, more and more adoption worldwide across all of those clouds as they create these cloud instances. Q3 was the quarter for us to also create the opportunity to now expand to our server OEMs, our server OEMs that are building for the overall enterprises, building for industries and serving overall edge computing. That also came with the release of DGXs, DGX SuperPODs, that also support such an important part of these industries. We are pleased with both the adoption and the overall scale that we are getting with overall A100. It's been several years since we launched overall V100. Our breadth and depth of our customer base has expanded significantly, not only in terms of both their understanding of accelerated computing and AI, but our focus in terms of providing software capabilities for them to connect these platforms to their applications. We're able to serve so many more industries than what we had just a couple of years ago, and we will continue to build on that investment to do so. So as we move from Q3 to Q4, we still expect a sequential increase in NVIDIA compute. As we ramp the A100, Mellanox will be down, but they had such an amazing strong quarter in Q3. So as we move forward, we're still in these early days of A100. But most importantly, we're in the early days of AI and accelerated computing.

Well, Colette, to your guys' credit, you've kind of leaned forward on the product cycles that are driving your data center business, but it's got to -- if you look at the macro this year and the fact that enterprise went to a work-from-home mode, that had to have had some sort of impact on the business, especially in the second half. You look at server demand in the enterprise, and it's down over 40% year-over-year. Is there sort of a rebound that we should expect in core enterprise as we get into next year and the vaccine gets more widely distributed?

Yes. It's a really good question that says COVID and the pandemic had pluses and minuses, and there are companies that have benefited, there's companies that have really struggled. But I think it's important to understand some of those pieces and how that influenced such as the data center and such as the enterprise. So if you look in terms of our Pro Viz business, there's an example that says the overall Pro Viz business was purchasing from our overall OEMs. At the initial onset of the pandemic, a real struggle to both get overall components and building through some of the challenges of the overall supply ecosystem to build that. But additionally, how do you support from the work from home? What we saw in Q3, which was really interesting with pro viz as we started to see that sequential increase and then returning back. Super strong results in terms of our workstations and our mobile workstations as the OEMs and enterprise found a way to both procure and build for this very important market as they are working from home. And so we do believe that's coming back up. Now when you think about in the data center and you think about the enterprises, we're not general purpose overall enterprise purchasing. We are focused on accelerated in AI, and we are in those initial years as they build out those workloads to do. But it has been an important area for many companies that have to rethink about where their investments are. They continue to invest in that overall AI. The cloud has been a very important part of them as they have been working from home to use the overall cloud to continue their work in expanding to accelerated and AI, and you see that in our results. But I do believe the overall business is continuing to work through some of the challenges with the pandemic and find ways to reach the overall enterprises, and we hope that this will get better in the quarters to come.

Colette, that's a great summary of some of the near-term issues. Maybe I can step back now and ask a longer-term, more open-ended question. I mean, NVIDIA has always had an extremely rich history in parallel compute. You've initially leveraged that into the gaming market. There were a lot of hard work. You brought that to HPC and then broader data center. But you've never just been a silicon company. There's been a strong software element to your offering. Now with the acquisition of Mellanox and the proposed acquisition of ARM, you're broadening out even further. Maybe you can spend a few minutes and just step back and reset kind of what you think the value proposition NVIDIA offers to the investment community? What's kind of the overarching strategy here?

Yes. I think you've watched so much of our transformation over the last decade or even decade and longer. And that transformation has really been about moving away from just a chip company focused on gaming. We have moved to become an overall computing company, a computing company that is not focused on the history of computing, but the future of computing and the future of accelerated computing, which will be with us for many decades going forward. In order to scale to that accelerated computing and the use of overall AI, it required a full stack approach to how we approach this overall market, meaning we -- yes, are still thinking about the silicon -- thinking about the best performance that you can get for the overall hardware, but we are also focused on systems and software because that full stack really differentiates us and is what is going to be necessary to achieve this transformation to accelerated computing as we go forward. The overall work of just focusing on just our hardware performance but not addressing the need for stitching together the software, and in many cases, easing the adoption by providing an end-to-end system that all they have to do is plug in for these overall markets. We've been able, with our full entire staff, to rapidly expand the ecosystem faster than we've ever done in the past. We have 2.3 million developers. We have more than 1,800 GPU-accelerated applications at this time. And we have received more than 6 million CUDA downloads alone in 2020. So a feat of a lot of work over a full generation, but we still have a lot of new things to address for these markets. Our platforms touch broad markets. Our platforms are not built specifically for one overall application, but built often for markets that are in their early days or things that we think will be strong and growing markets into the future. We've created a platform for RTX. That RTX platform is like no other ray-tracing platform that is out there. It is multi-years in its making, and it also includes AI. So to think about our overall real-time ray tracing but also putting AI into the hands of ray tracing in forward is something that none others have done. But we take that platform approach also to our workstations, to our data center and to automotive, which are -- with our HDX platforms, with our EGX edge platforms as well as our AGX platforms or autonomous platforms. Each and every single one of these platforms begins with a unified architecture and addressing the overall end market with the most important piece, which is that software. We have written so much software specific for these industries. You've seen different system softwares that help people bring AI. We can talk about Clara. We can talk about Jarvis, Merlin, the many other things that we do for the Metropolis, for overall DRIVE for cars. We have, at this point, as equal amount of software as we do hardware engineering, focusing on these markets. So when you step back and look at what our future entails, now with Mellanox, it is focused on end-to-end computing, computing that doesn't start with when the overall data reaches the overall GPU, but starts when you want to compute the data. So we're thinking about everything from the interconnects to the GPU to the overall GPU performance as well. All of those will matter long term when we think about accelerated computing.