All right. Why don't we go ahead and get started? Good morning, Happy New Year, and welcome to JPMorgan's 21st Annual Technology and Automotive Investor Forum here at the Consumer Electronics Show. My name is Harlan Sur. I'm the semiconductor and semiconductor capital equipment analyst for the firm. Very pleased to introduce Colette Kress, Executive Vice President and Chief Financial Officer at NVIDIA. It's been a tradition, nine years to have the NVIDIA team, specifically Colette, be the first to present at our investor conference because the team is driving many of the trends that you're gonna hear about today, right? Artificial intelligence, compute acceleration, next-generation compute platforms in automotive and gaming and IoT. In addition to that, the NVIDIA team is driving an emerging software and services revenue stream.
I've asked Colette to start us off with an overview of what the team announced on Tuesday at its CES special address, and then we'll go ahead and kick off the Q&A. Colette, thank you for the nine years of support, and I'll turn it over to you.
Thank you. Really pleased to be here. We wouldn't miss it. Even with the great weather that we're experiencing, we wouldn't miss it and come here. I do need to make an opening reminder statement that the presentation and discussions that we will have contains forward-looking statements. Investors are advised to read our reports that are filed with the SEC for information that relates to our risks and uncertainties facing our business. Okay, now that I have that out of the way, I wanna first start with some of the key things that we have been working on, probably since we've last discussed after our Q3 earnings. Ada. Ada, which is our architecture for our gaming platforms, is off to a great start.
We began earlier launching our 4090, launching our 4080, which is also becoming one of the most popular GPUs in retail. We're here today to announce our RTX 4070 Ti version as well. We're really pleased in terms of what we're seeing. Of course, our Ada architecture is leading, and we are still working in terms of our inventory correction, which is going as expected right now. Secondly, we are focused on bringing to market a large lineup of new products, particularly in our data center. We have our Hopper architecture working on for many years and a really important improvement for many of the new types of workloads that are hitting the market. I think we'll talk a little bit more about that today.
We are continuing to watch China and its opening. China has certainly been through quite a bit in terms of with their zero COVID policy. It's still early for us, but we are watching them carefully. As you know, in the last year, in calendar 2022, really challenges in terms of both our gaming and our data center, based on the economy in China and the COVID lockdowns. We are here at CES. We're here for the consumer folks, both here for gaming as well as for auto. Let me kinda discuss some of the things that we discussed. Talked about the 4070 Ti coming to market, and we are also bringing our 40 series to notebooks.
We have more than 170 different notebooks, probably able to see throughout the floor of the convention center. It is also fueled with our DLSS, which really helps bring AI and ray tracing to so many of the gamers. Our DLSS is in its 3rd generation, very similar to our ray tracing architecture being in its 3rd generation. We also have more than 50 new games using DLSS 3, both here and available, but also coming in the next couple quarters. We're really excited to see what we can do with our 40 series. It's not just focused on on-premise gaming. Our GFN will now be available with our 4080.
GFN is our cloud opportunity for gaming and bringing a high-end 4080 is a really great experience for those cloud gamers. I love the fun fact also of what we're bringing is GFN to the automotive. We decided to bring those two things together. We have several of our automotive makers that are looking to provide GFN inside of the car. Gaming everywhere and always. We're really pleased. Moving to our automotive, probably one of the key highlights of our automotive announcements out here is Foxconn has decided to join and has adopted our Orin architecture for compute in the many NEVs. Not just the NEVs that we are already a part of, but any type of NEV, new energy types of cars across the world.
We're very excited about that partnership and their endorsement of the Orin architecture. There's more that we are seeing in terms of Omniverse. Omniverse has of course now been adopted by Mercedes in their next-generation factories, similar to some of the other automotive companies that are really seeing the need of the digital twin of the factories to find the perfect combination of how to run those factories very efficiently and using the AI. Those are some of the key highlights I wanted to start with. There's a couple other ones, but a really great show, and we're really pleased to be here in person.
I appreciate that, Colette. Thank you for that. I'll start off with the first few questions, and I'll certainly turn it over to the audience to see if you guys have any questions. If you hit your guidance this quarter, you know, revenue for calendar 2022 is expected to remain flattish. You know, that's actually much better than prior down cycles and inventory correction cycles for the company, and I think it's a strong reflection of the product cycles, the diversification. As we look into this year, fiscal 2024, consensus has your revenues growing. What's the team's confidence on driving growth this year in an environment where global demand trends will be soft? You know, help us understand the trends and product cycles that will drive the business this year.
Maybe longer term, how do we think about the overall growth profile for the different businesses, right? Gaming, data center, Professional Visualization, and automotive.
Yeah. A great way to start out in talking about it's a new year, what do we wanna see and focus on in the new year. One of the things that, of course, we can't control is the macro environment around the world, and the macro environment around the world is just not the same in every single country, each one of them facing quite uniqueness. We can spend our time focusing on our opportunities in front of us, as you're correct. We have a large portfolio of new products coming to market. Those even start with our data center business, one of our more important architectures, our Hopper architecture, supporting the massive growth that we've seen in AI over the last several years.
Not only is it about the Hopper architecture, but remember, we're also bringing our Grace CPU to market. We are bringing in our networking business our latest DPU to market with BlueField, and we will also be looking at a large part of our portfolio in networking at the higher speeds that are necessary to support the environment of data center computing architecture as a whole. We're not just focused on a singular GPU type of platform. We're really focused on the data center computing as a whole. The environment couldn't be more set up right now as folks have really understood that the ending of Moore's Law is now really here. When you're in a situation of having to make choices, the movement to accelerated computing and the use of AI couldn't be more important for them.
That's only one part of it. As you know, in gaming with our 40 series, with the kickoff of the 4090, we really believe this is a great architecture, the excitement of gaming and what a third generation of ray tracing with DLSS can provide, not only to just gamers, but the creators out there. There are so many different use case opportunities of that ray tracing that really takes us into the workstations on Professional Visualization, when we talk about what they could also do with it. The creatives thinking about the use of ray tracing and all of the work that they are doing. Lastly, we are continuing a growth trajectory on our automotive business.
Orin has been a whole run, both working in terms of that compute platform of many of the NEVs. We also have an $11 billion pipeline moving forward, and we are just at that turning point that we could be getting to grow to a billion-dollar business in our automotive, realizing that potential of that $11 billion pipeline. But keep in mind, there is still work to do. You will see the work coming with our key agreements that we have with JLR, what we have with Mercedes-Benz and Daimler. All of those are still in the future. So the growth is still here to go forward. Those are the things that we are focused on as we go forward, even in the current macro environment
You know, near term, I think as you mentioned, the team has been working through excess channel inventories in gaming as well as your Professional Visualization business. On gaming in particular, you know, Q2 and Q3 combined, you're shipping about 30% below end demand consumption of around $5 billion. When we went back and looked historically after two quarters of undershipping demand, the team typically does see an inflection in the business. You had anticipated channel inventories back to more normalized levels exiting this quarter. Is the team still tracking to this milestone? Is end consumption still trending at roughly that sort of $2.5 billion-ish run rate per quarter plus or minus even in, even in this sort of tough macro environment?
Yes. The macro environment and what we're seeing in terms of demand for gaming in North America and in Europe, as we've discussed, we're seeing solid-.
Yeah.
solid demand. China has gone through a lot in calendar 2022. The end of zero COVID may improve, but it's still too early for us to say. It has been a little bit choppy in terms of China, and we had discussed that in our Q3. Still, gaming as a whole is a very important area for us to focus on, and the demand is definitely there and solid. Moving to our 40 series, bringing that to market, that's one piece of it. In terms of the demand, but also again, getting the correction-
Yeah.
-of the channel. Yes, we are on track as we leave this Q4, our Q4 at the end of January, to be near normal channel levels in the market. We believe that this is just through a combination of the holiday seasons.
Yes.
A really direct focus on making sure working with our AIC partners and our OEMs to move that channel inventory. You're correct, it does sometimes takes a couple quarters to do that. We moved quickly, and we think it's working quite well.
Perfect. Before I move on to some of the product segments, let me see if anybody has any questions before I move forward. Let's start off with your data center business. You know, it's more than 60% of your overall revenues. You know, the business is tracking towards 40% year-over-year growth in fiscal 2023, which is more than 1.5x the cloud CapEx growth last year, and the team has driven roughly about a 70% CAGR over the past three years, which is about 2x the annualized growth rate relative to cloud CapEx spending growth. Our hardware team is forecasting cloud CapEx spending still up high single digits this year, but more of the spending is focused on strategic compute initiatives like accelerated compute, right?
You combine this with the next generation Hopper ramp, your H100 ramp, continued strong enterprise adoption, strong networking demand pull. You know, history would suggest that the team should grow the data center segment by double-digit % growth rate this year. Help us understand the puts and takes around data center business in this challenging macro environment this year and confidence on driving growth for fiscal 2024?
Yeah. There's a focus within the company to really get these great products to market that we have. The Hopper architecture that we're bringing to market it feeds so many great things. The AI market has expanded exceptionally since our last architecture, and not only just the advancement of the market as a whole, just the complexity, the size of the models have continued to grow. We're seeing three very important areas in AI that were probably in their infancy or did not exist in the last three years or more, and those areas are these large language models-
Yeah.
fueled by the natural language processing. The recommender systems are yet another one that are still very, very important to monetization. Monetization and marketing, a very key area. This last one that I think you've probably all seen during the holidays is the focus on generative AI. Really a great opportunity of both now creating these models, creating these trained models, and seeing what is now possible to be produced out of generative AI. We're very excited for this. You also focused on key areas of the end of Moore's Law and what accelerated computing needs to do. We're in the right position with our architecture to both improve performance of nearly 5x or more from the last generation, but also the energy efficiency, because the concept of the end of Moore's Law really goes into the sustainability.
What are we going to do with our data centers going forward if we're not focusing on the efficiency in there? We're all set up for all of these right things as the market continues to grow. It's not just Hopper, it is a full lineup across the board. We expect our Grace CPU to help influence the connectivity of working with our GPU, really working in terms of the speeds of all the work in terms of the data center. Again, not being able to control the macro environment, we have a great lineup of products to help fuel growth if it's there.
The team started shipping Hopper production and initial shipments in Q-- your fiscal Q3. Server OEM availability was this quarter, right, Q4. Cloud instances available next quarter. The team noted, I think, in the last earnings call that Hopper is expected to ship in large volumes in the April quarter. Is that still the case? Can you just talk about the overall sort of cloud hyperscale customer traction, design win pipeline so far with the H100 and how that compares to the A100 cycle?
Yeah. Each generation now of bringing our data center architecture to market has had more advancements and more connectivity with our customers. When we think about the work that we've done to bring this to market, it's not alone. It's with our partners, it is with our customers. So much of the work and the anticipation of Hopper may have not been an excitement of, "Here it is," because we've been working on it and working with our customers for quite some time. Those relationships are very important. They have been helping us both qualify it, working with it and being ready to bring this to market. We will bring production level H100 to market this quarter. We started some initial ones in Q3.
Yeah.
That is starting here in Q4. Yes, we expect a ramp into fiscal year 2024, the first quarter as well as we continue to ramp H100. This is a big undertaking. It is a fabulous product, but the relationships and our work will really come through.
You know, as we talk with, a lot of the users, the power users of accelerated compute, right? One of the things that they're anxiously awaiting is a CPU, GPU architecture, which it's called cache coherency. Basically, the thought process is memory sharing, like efficient memory sharing between the CPU and GPU. That's not currently available. I think Intel, AMD are two years away with the CXL initiative. Your Grace CPU, which is expected to sample, I believe, in the first half of the year, that you guys have already pre-integrated, cache coherency, in your Grace CPU. Give us a sense of like I said, you team expected to sample in the first half of the year, ramp into the market in the second half of the year.
Is the team tracking to its timelines on Grace, and how do we think about the revenue profile and the opportunity for Grace?
Sure. Let's talk about why Grace and what our overall goal was there. When you think about the work that is being done for AI or the work that's being in the data center, you can't ever accomplish that with just one GPU. We have been spending years focusing on linking together multiple GPUs, essentially coming up with supercomputers. Supercomputers to help both the size of the models as well as the efficiencies and speeds of the work that they are accomplishing. We've done that both with our own interconnects and then came together with Mellanox as well, working on that.
Yeah.
Now here's another important piece of why we brought Grace here. Grace gives us the opportunity to focus on the speeds between that CPU and the GPUs, but also focusing, as you had mentioned, in terms of the memory.
Yeah.
On how can we make sure that is the most efficient processes that we can do. Our Grace CPU is focused on very large models. Large models focusing on AI, focusing on the supercomputing, focusing on high-performance computing workloads, and many of our new things that we see in the future, such as digital twins, and the amount of work that will take into account. Grace and Hopper go together. Yes, we're on track for sampling in the first half, and we'll bring this to market in the second half of the year. This is an area where the design wins of working with many of our customers as they see this connectivity of putting the two products together to be great for their opportunities, and we'll continue to work on that.
Let's shift the focus over to gaming. You know, despite a challenging consumer market demand for the new Ada 40 series gaming products, the initial ones, the 4090, 4080 seems quite good, right? We checked some of the online retail sites. 4090 is tight availability in most places. 4080 GPU card pricing is marked up about 10%-15% relative to MSRP. This all suggests really good demand pull-through. You mentioned at the special address, or Jeff mentioned at the special address, 40% of your installed base has adopted RTX GPUs, and we're now in your third generation of the RTX architecture. With the new product launch, how should we think about the additional penetration rate into your installed base? Typically, like, how long are these GPU replacement cycles?
Great set of questions, agree wholeheartedly in terms of our 4090, 4080 and what we're seeing out in the market. 4090 sold out quite quickly. We've already started the restocking of that working with our AICs to do so. The 4080 is doing very well in terms of in retail as well in terms of selling. We have RTX now in our installed base of nearly 40%. Interesting because we were probably on stage here several years ago, talking about what is ray tracing, and are there going to be games with that? Now we have in our installed base more than 40% focused on ray tracing. Ampere architecture probably was about a third of that.
Now here we come with the third generation, and there's again great opportunity because it's not just about that platform, it has been about our work with the ecosystem in creating the games that can now use ray tracing and use DLSS. We're on a great trajectory there to see that take off more. As we go forward in the new year, I think you're going to see more of our 40 Series.
Mm-hmm.
come to market. How do we think about how much of that will be sold to those that are refreshing versus those that are coming in as new gamers? Always tough to look at, what we have seen is we bring a new generation of the product to market 2+ years, every 2+ years we'll bring a new generation. What we see is probably refreshes some of the most important gamers out there to be a little bit over three years. This is a great opportunity for not only upgrading to Ampere, but thinking about those that are in prior generations that have not upgraded yet. That is a material performance increase, and a no-brainer for them in terms of purchasing. I think we'll see both new gamers and that refresh.
Let's switch over and talk about software. You know, in your software solutions, you're enabling customers to come to market rapidly by developing full stack. You know, we've always talked about the NVIDIA strategy as focused on ecosystem enablement, right? A part of that is full stack development, turnkey solutions, managed services. You're monetizing your software platforms, and you've got monetization opportunities across all of your end markets, right? The business is running. I believe the software services and ratable type of revenue streams is running low $100 million annualized run rate today, and you're now selling software separately, right? You've got NVIDIA AI Enterprise, Omniverse-AV, like how do we think about the size of these opportunities and what's the monetization timeline of these software and services-based opportunities?
Yeah. Software is a very important piece in terms of our success almost in each one of our businesses. For many years, we have put software in both the ecosystem as well as in our products that enables customers to take out of a box and begin their work that they need, because we've enabled a lot of that software. There now is that opportunity to sell it separately. Why? It creates more of that relationship and the help that is needed for many of our enterprises. Our enterprises need that support as they are beginning their work on AI projects or otherwise. We took this opportunity to break out software and sell it separately with our enterprises. You're gonna see it in three major areas and probably more going forward. The first is NVIDIA AI Enterprise.
This is a core you should think about as the operating system of AI available for our enterprises. Not only being sold with us and our partners, but also being sold with several of our CSPs, putting it in their marketplace. They really believe that it's not just about setting up cloud instances, they need to support that software that they are all using with NVIDIA to do that as well. We're really pleased with that piece of work that's coming together. NVIDIA AIE is an important piece. The second one we've talked about is Omniverse. Omniverse really fuels what we will see as the future of 3D internet and folks really working not only on those three dimensions, but what you may see in terms of digital twins and work in that piece.
We'll sell that, whether that be with groups inside the enterprise or even to an individual that may be a creative working on this as well. The third one is our focus on software for automotive. Our focus is not just on NEVs and creating a computing platform and the hardware infrastructure. We have a significant amount of work and effort and design wins as we focus on AV. We are now able to do both of selling them that hardware, but also being a part of the software, revenue stream that we'll enable within the OEMs. We'll be able to share that with them, with Daimler and both with JLR. That is, more of the medium and long term, with what we're seeing in terms of NVIDIA AIE and Omniverse is here right now.
In automotive, as you mentioned, the team has an $11 billion design win pipeline that is expected to unfold over the next few years. That business started to inflect pretty meaningfully in Q2 of last year. In fact, it's driving like 100% year-over-year growth, I think, this quarter driven by your Orin-based platforms which are starting to ramp. You also still have in front of you the revenue-sharing opportunities with Daimler, Jaguar Land Rover in front of you. How should we think about the ramp into this $11 billion automotive pipeline over the next, you know, three to four years?
Yeah. You're gonna see different stages, but we've already probably passed an inflection point that we're seeing the growth of Orin with our NEVs. Why are the NEVs focused on Orin? One, both the performance level and when you're thinking about building an electric car, you're really starting from the ground up and redesigning and having that central computing platform has been very important to them. Both our work now with Foxconn as well as many of the top NEVs around the world focused on this. It is a great opportunity to continue to ramp. Additionally, the next phase will be coming with our design wins both on robotaxis and bringing some of the early AV to market. In this case, this can be both in infrastructure and may have some of our software incorporated in them well.
Furthermore, in the long term, probably when we get to calendar 2025, is when you are going to see production vehicles for AV start to ramp and be a more meaningful position in our automotive. At this point, we're gonna be now seeing that software really take off. Right now, we're still in the hardware infrastructure that's driving our revenue, but again, growing into that pipeline that we have going forward.
On the financials, how do we think about the OpEx profile relative to the revenue growth given, you know, you talked about a very aggressive cadence, right? Across your GPU, CPU, and your networking products, the team is driving a 2-year cadence. That's an extremely aggressive cadence. On top of that, you've got ecosystem enablement, you've got software development, and so on. How should we think about the R&D or the OpEx intensity and that relative to revenue growth going forward?
When the first challenges of the economy hit, it was a focus of ours to really slow down the hiring.
Yeah.
that we can focus on our teams that we had, that we could focus on our employees that we had, and that was the very first thing that we did, is slowing down that hiring. We are continuing to work to try and keep on a sequential basis about as flat as we can in this environment. You're right. We've got a very.
Mm-hmm.
large portfolio of bringing new products to market, and they're quite complex in that work. Our business model and our structure as a company allows quite a bit of efficiency. When I mean that is, remember, we're at essentially one architecture across everything that we do. All of our software is both forward compatibility and backwards compatibility. Having that consistent unified architecture has really allowed the efficiencies in how we structure the organization as a whole. In this environment, that is helping us fuel bring this stuff together. We'll continue to focus on making the appropriate investments to bring these products to market, but we're always looking for new areas to find efficiencies.
Well, thank you. We'll look forward to monitoring the progress and the execution of the team this year. Colette, thank you very much for your support.
Great. Thank you.
Thank you.
Thank you.