NVIDIA Corporation (NVDA) Earnings Call Transcript & Summary

All right. Welcome back, everybody. Colette, pleasure to have you join us, as always. Thank you for making time for us. Just a quick housekeeping note. My moderator Q&A thing is finally working, so please feel free to pop in. And as always, you could just keep shooting me e-mails as well. So a great pleasure to have NVIDIA join us. And Colette Kress needs no introduction, CFO of the company. So before we get started, Colette, you reported earnings last week. Maybe just would be helpful to get a quick recap of what -- how you highlighted the business conditions near term. And then we'll take it from there into Q&A.

Great. Thanks so much for hosting us here at your conference. I'm going to have to start with just a reminder that our discussion and 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. So let me first start with some of the key highlights and takeaways from our earnings that we just had in terms of last week and our thoughts in terms of where we think moving into Q3 will go. First, our data center demand for NVIDIA compute continues to accelerate. We saw exceptional growth in this last quarter, particularly in AI inferencing, more than doubled year-over-year. This is an area where customers are taking deep learning AI workloads to production. They're shifting away from CPUs to use GPU for inferencing. And our A30, ramping quite well, and it includes about 4x the performance improvement to T4, our prior inferencing-specific card. Additionally, we're seeing a leadership in our AI training to continue. We have both swept the ML performance benchmarks that were out there, but also strong growth with natural language processing, deep recommendators, which are very important new workloads that are occurring. This is likely going to be a phase of industrial AI, high-performance computing revolution. It is underway as we see more and more large supercomputers, supercomputing out there taking advantage of this opportunity to introduce acceleration to a good part of the industry, and we'll talk about that more. Moving to our overall NVIDIA RTX. NVIDIA RTX is available as a great opportunity to advance overall graphics for both our consumer gaming but also for our enterprise graphics. Entering into the holiday season in the next couple of months, we are entering with the strongest of our product cycle, expanding both the set of RTX titles for gaming and also in lean channel inventories. We also have the biggest wave of NVIDIA laptops for both back-to-school, studio as well as that holiday season as well. Now when we move to further talk about the opportunities of high-end graphics, we can't start that without talking about Omniverse, the power and the future of an immersive overall entertainment and industrial collaboration approach as we move forward. And we'll talk about that in addition. We have been highlighting our NVIDIA AI software and subscriptions as a key opportunity for us as we move forward, probably a multibillion-dollar opportunity as we go forward. So we've talked about our guidance, and what we provided as our guidance, we expect another strong quarter. We expect sequential growth driven largely by the accelerating demand in data center. In addition, we also expect sequential growth in each of our other 3 market platforms. That means our gaming business, that means our pro visualization as well as our automotive. We expect the contribution for CMP, our crypto mining processors, to be minimal as we think about the outlook going forward. So our revenue guidance for next quarter is $6.8 billion and a large portion of that growth is related to our data center business.

Great. Just in terms of visibility -- and especially given how tight things are, which is actually quite remarkable that given lead times stressed out, you guys have delivered the kind of growth you have. But in terms of visibility, can you please comment on your visibility given the stressed lead times? I think earlier in the year, you had characterized that your visibility to the data center business was very strong because a lot of the more project-based product -- projects that you are working on. So how does the visibility look like versus what it would look like normally at this time of the year?

So when we think about our data center business, understand we have a wide set of different types of customers. Yes, we have conversations both with our hyperscales, hyperscales for internal use of our infrastructure but also providing it for cloud instances. We also have a very meaningful amount of vertical industries that make up a portion of probably close to 50% of our data center business. That can be achieved from purchasing from our certified OEMs as they create many different server offerings for this industry, but they can also be very specific sales such as superPODs and/or our DGXs. Our work in terms of visibility stems all the way from helping many of the companies, hyperscales and vertical industries look for infrastructure solutions to solidify what they're trying to do for AI, what they're trying to do for high-performance computing, inferencing. And much of that is engineering engagement from the overall architectural level. Our overall visibility as we move into Q3 is quite solid. This has been an area of continued growth on our part of helping companies end to end on their platform and system choices as they build that.

Okay. I have a few more questions on the -- what seems to be an endless supply-demand imbalance. I certainly haven't seen anything extend out over such a long time. In that context, could you talk about lead times? And especially in the gaming business, it sounds like we should just throw out seasonality out the window, A, because the business is different because of how big laptops are, but then also how constrained supply has been.

Sure. Let me -- let's talk about our overall gaming business and what may be referred to as lead times. There's multiple parts of our overall gaming business. Our gaming business is structured to support desktops, to support notebooks, to support overall cloud gaming. So each of those require a different level of supply and overall working towards that. We have articulated now for a couple of quarters that we are supply constrained on our gaming business. But the way to think about that is demand is quite high. Demand is very strong. And we continue to produce strong results in terms of revenue growth year-over-year, but we continue to work with all of our different partners on supply. So it's not just looking at the supply that we have but it's also understanding our OEMs, our AICs and helping them create the systems, the boards that need to go to market. So our lead times, as we think about this, have already been determined because this is something that you are usually procuring for the long term when you think about your long-term growth. Those procurement of things have been made. But now it is for us to work with our partners to make sure that they can get the boards and systems to market but they also have all of the systems, parts and components that are necessary to put these into market. We feel very good about our guidance that we provided for Q3. We believe that we have enough supply to serve what we refer to as the company's growth aspirations for both the second half as well as moving into next year.

Okay. And many of the chip companies, larger ones especially, have -- I don't want to use the word resorted to but they have resorted to noncancelable orders, signing up many different ways to -- really to ensure their own planning and help them plan capacity. So Colette, what has NVIDIA done to ensure that you're planning your capacity better in order to meet the demand given all the challenges on the supply side?

We can break down your question regarding planning in 2 different types of areas: what we're doing for overall consumer planning -- consumer gaming planning and then what we are doing for the overall enterprises particularly as we deal with data center and data center compute. From a planning perspective of gaming, we do take the role of making sure we have a good understanding of the growth that could happen in terms of consumer gaming. And we will plan from both a consolidated level how to think about that levels of supply that would need to be available in the channel at any point in time. As we leave the end of Q2 and go into Q3, our overall supply levels at the channel are still quite lean. So we have a goal to continue to improve those levels to allow GeForce cards to be available to our gamers as we move forward. Now that is working with each of our channel partners to understand what is available in the channel, how we can continue to serve those different markets, those different regions in terms of the types of parts that they want. That is where our long-term planning comes into mind, that says how do we restore those channel levels. Now when we look at overall data center, purchasing overall data center compute is a long process, both planning in terms of what the customer would like, also planning on our side because many of our products that we ship for the data center take quite a bit of time to produce from end to end. So you could think that most of our supply planning has been over a long period of time for us to serve this market. And now we are trying to find those best solutions for our customers. There will continue to be areas where we have to focus with our OEMs, with our partners to assure that everything that is needed for the data center is available both at the OEMs as well as through NVIDIA to serve our overall customers.

Okay. That's very helpful. Let me switch gears to longer term. And this was a question that Jensen did address to a certain extent during the earnings call, your software strategy. It seems -- as you mentioned, it's a multibillion opportunity. And the way I think about it obviously has very positive implications for the operating model. So the question here is just recap for us, please, what are the 3 pillars for the software strategy as you have laid out. And then I had a couple of follow-ups on that.

Absolutely. Great question. We're not new to software, we're not new to software. A very large portion of our systems that we provide throughout all our market platforms incorporate software. And we have 3 different ways to deliver software value to our customers. First, software and hardware could be offered separately. This is a case where we're talking now that we may enable a software offering to the enterprise channel that they purchase through our OEM partners. Number two, software and hardware could be delivered together as a platform. This is very common what we do with our CSPs and/or hyperscales. And number three, for the community as a whole, whether you think of research, education, start-ups, our software is free. This is the underpinnings for example of CUDA over the last decade in terms of how we provide it, providing them the capability to learn the overall development platform and think of new opportunities on how to use it. The functionality and support level with each of these 3 is different for each of them. Enterprises, for example, may prefer to pay for what they have in mission-critical software and related support.

And then what are the milestones that we should be -- investors should be looking at to see the -- if the approach is being adopted? And I recognize revenues are not that immediate, but how will you inform the community to be aware of the practice on this front?

Yes. We're happy to share in the next couple of years in terms of how we're doing in our specific software approach if it is a separate purchase from the overall hardware. We're just entering in a space of an important era of industrial growth, industrial growth that will use AI and use that software. So you will likely see first something in terms of the overall bookings of what we're seeing in software and then in the future, what we'll see in terms of on the income statement in terms of revenue. As we go from time to time, we'll be happy to share what we're seeing in that area.

Great. Can you share with us some examples of some early wins that you've had since you officially rolled it out, which I'm sure was before you told us about it?

So we have been working on many different types of subscriptions with key providers out there. Many of our OEMs are now fully trained to sell our overall software applications, whether that be Base Command, Fleet Command, enterprise AI software. So we'll, in the future, talk about those key lighthouse wins and maybe the size of what we're seeing in terms of the number of pickup from that.

Got it, got it. Okay. Let's switch to the demand drivers for each of the segments, and we will start with gaming first. You're 4 quarters into the Ampere launch, and I think you mentioned about 20% of the installed base is upgraded. So just help us understand typically how long does it take. And to what extent does asymptote out in terms of the upgrade cycle?

It's an important area to focus on regarding our Ampere architecture and the performance that is being delivered with our Ampere architecture for gaming but also on the surge of gaming and the use of overall real-time ray tracing. When we think about the Ampere architecture, it's in its fairly early part of the architecture as a whole. We've probably been in market for about a year, and we are continuing to provide new cards into the market at this time. Now when we talk about that 20%, the 20% relates to the percentage of folks that have upgraded in our installed base to either Ampere or are able to take part in terms of ray tracing. So that will incorporate some of our prior architecture that incorporated ray tracing. This is now looking at 80% of our installed base that still has the opportunity to upgrade with the Ampere architecture, not just looking at our prior architecture of Turing, but you can even look at our Pascal architecture and understand the great performance enhancement now that you have available to you with Ampere going forward. We still have lean supply out in the market. And we do believe we'll continue to serve the gaming market with more cards in the quarters to come. So as we move into this back-to-school season as well as the holiday, we will look to upgrade the installed base but also add new gamers who are interested in that Ampere architecture and the availability of overall supply. When we move from architecture to architecture, it's not uncommon that we will reach a portion of our installed base on the most current architecture, but it may not be the majority of the installed base on that specific architecture. We may take that opportunity to move to a new architecture to entice even more excitement of the performance and the games that are coming out. But right now, getting ready for the overall holiday season, we'll be ready with Ampere to serve the market.

And Colette, what is the right way to think about the installed base? A few years ago, it was relatively simple. I say relatively simple. You could look at desktops sold. Now it's laptops, it's cloud gaming, it's e-gaming. How do you -- when you think about the installed base, what's the right way investors should be thinking about it?

The way to think about the installed base is there are hundreds of millions of overall gamers that are out there. And you're correct, there's many different ways that you can now achieve overall gaming. You can have multiple devices. You can be leveraging the cloud for your experience because now we're enabling not just on a standard desktop or notebook but supporting many different infrastructures to support the overall cloud as well. So this is an ability for us to look at many of the online gaming sites to really see how well NVIDIA's GPUs are serving in there. We continue to be one of the fastest growing GPUs on the list of many of these sites, such as Steam. So this is an area where we focus on, but we have continued to provide more and more opportunities and options for our gamers to take advantage of high-end gaming.

On the mix of laptop and desktop, could you remind us what have you said publicly about how big is the laptop business now as a percent of overall -- or as a percent versus desktops?

So far, year-to-date, we've talked quite a bit about new laptops -- high-end laptops that leverage our Max-Q technology available for overall gaming. Both the thin and light has been an important area where gamers would like to see, and it has been very, very successful. Our notebooks are growing quite well over the last couple of years, not just this year-to-date. It now represents probably 30% of our overall gaming business with our notebooks.

Okay. Question on crypto. I thought you guys did a great job not only in addressing it with the CMP product line but also in communicating with us much earlier than the prior time. So just remind us regarding your confidence on the comments you made, which I thought was pretty meaningful, that you said you expect CMP to be pretty almost -- I think that may not have been the word you used, immaterial to the business going forward.