NVIDIA Corporation

Good afternoon. My name is Christina, and I will be your conference operator today. Welcome to NVIDIA's financial results conference call. All lines have been placed on mute. After the speaker's remarks, there will be a question and answer period. At this time, if you'd like to ask a question, please press start and the number one on your telephone keypad. To withdraw your question, press the pound key. Thank you. I'll now turn the call over to Simona Jankowski, Vice President of Investor Relations, to begin your conference.

Thank you. Good afternoon, everyone, and welcome to NVIDIA's conference call for the fourth quarter of fiscal 2019. With me on the call today from NVIDIA are Jensen Huang, President and Chief Executive Officer, and Colette Kress, Executive Vice President and Chief Financial Officer. I'd like to remind you that our call is being webcast live on NVIDIA's Investor Relations website. The webcast will be available for replay until the conference call to discuss our financial results for the first quarter of fiscal 2020. The content of today's call is indeed its property. It can't be reproduced or transcribed without our prior written consent. During this call, we may make forward-looking statements based on current expectations. These are subject to a number of significant risks and uncertainties, and our actual results may differ materially. For a discussion of factors that could affect our future financial results and business, please refer to the disclosure in today's earnings release, our most recent forms 10-K and 10-Q, and the reports that we may file on Form 8K with the Securities and Exchange Commission. All our statements are made as of today, February 14, 2019, based on information currently available to us. Except as required by law, we assume no obligation to update any such statements. During this call, we will discuss non-GAAP financial measures. You can find a reconciliation of these non-GAAP financial measures to GAAP financial measures in our CFO Commentary, which is posted on our website. With that, let me turn the call over to Colette.

Thanks, Ramona. As you know, we lowered our fourth quarter guidance on January 28th, and our results are in line with our preannouncement. Q4 revenue was $2.21 billion, down 24% from a year earlier, driven primarily by a 45% year-on-year decline in gaming. Full-year revenue was $11.72 billion, up 21% from a previous year. Starting with our gaming business, revenue of $954 million was down 45% year on year and down 46% sequentially, weaker than our expectations heading into the quarter. Full year revenue was up 13% to $6.25 billion. Three factors contributed to the Q4 gaming revenue decline. First, Post-crypto inventory of GPUs in a channel caused us to reduce shipments in order to allow excess channel inventory to sell through. We expect channel inventory to normalize in Q1 in line with 1 to 2 quarter timeline we had outlined on our previous earnings call. Second, deteriorating macroeconomic conditions, particularly in China, impacted consumer demand for our GPUs. And third, Sales of certain high-end GPUs using our new Turing architecture, including the GeForce RTX 2080 and 2070, were lower than we expected for the launch of a new architecture. These products deliver a revolutionary leap in performance and innovation with real-time ray tracing and AI, but some customers may have delayed their purchase while waiting for lower price points or further demonstrations of the RTX technology in actual games. The significant volatility in our gaming business over the last few quarters has been challenging to model. Crypto mining demand and its after effects have distorted the quarter-to-quarter trends in the gaming business and obscured its underlying trend line. Let me try to give you some visibility into what we believe the long-rise business looks like. As you know, our gaming business consists of desktop gaming, notebook gaming, and gaming console products. To get a sense of the underlying run rate in our gaming business last year, it is helpful to look at desktop gaming revenue across a period that doesn't include crypto demand. Let's look at the four quarters starting from Q2 of last year to the current quarter or Q1 of this year. In Q2 and Q3 of last year, to the benefit of hindsight, we shipped a higher amount of desktop gaming products relative to where end demand turned out to be. To allow the channel to work down that excess channel inventory, we shipped a lower amount relative to end demand in Q4, and we'll do so again in Q1. Therefore, exiting Q1, we expect channel inventories to be at normal levels. On average, our desktop gaming revenue across these four quarters is about $900 million. We believe this represents the normalized level of desktop gaming for this period. Notebook gaming and gaming consoles have averaged close to $500 million per quarter over these same four quarters. Thus, in total, we believe the normalized quarterly gaming business revenue run rate is approximately $1.4 billion. As we look past Q1, we expect the channel inventory correction to be behind us and our business to have bottomed. On a full year basis, we expect our gaming business to be down slightly given the tough first half compares with growth in Turing and notebook gaming. At CES last month, we launched into the recovery of our gaming business. We announced the GeForce RTX 2060 at the mid-range price point of $349. The 2060 delivers a 60% performance improvement over the GTX 1060, while also bringing Turing's real-time ray tracing capabilities and AI features to the mass market for the first time. The 2060 has received rave reviews and is off to a great start. In addition, we announced a record of 40-plus new Turing-based gaming laptops, which became available on January 29th. This is more than double the number of GeForce-powered notebooks in the market last year. Featuring the energy efficiency of the Turing architecture, Ben and Mike laptops are able to deliver the performance of desktop gaming VCs. We expect GeForce laptops to continue to be the fastest-growing segment of gaming. We are also pleased to see growing momentum in the RTX ecosystem, as more game developers are creating content to take advantage of the Turing architecture's amazing capabilities. Just this week, DLSS technology is becoming available in two blockbuster games, Battlefield V and Metro Exodius. and Anthem coming soon. In addition, at CES, Justice and Atomic Heart showed demos featuring ray tracing and DLSS, and a large pipeline of games planned to integrate RTX technology. Pairing DLSS with ray tracing can provide comparable frame rates to traditional rasterization, but also much more beautiful cinematic visuals. The best of both worlds. This is the next generation of gaming. While this was a challenging quarter in our gaming business, we look forward to putting the channel inventory correction behind us and building on the solid foundation of our Turing architecture. Moving to data center. Revenue was $679 million, up 12% year-on-year and down 14% sequentially. Full-year data center revenue was $2.93 billion, up a strong 52%. The Q4 sales decline was broad-based across vertical and markets and geographies. As the quarter progressed, customers around the world became increasingly cautious due to rising economic uncertainty, and the number of deals did not close in January. In addition, hyperscale and cloud purchases declined both sequentially and year-on-year, as several customers paused at the end of the year. We believe the pause is temporary. The strength of NVIDIA's accelerated computing platform remains intact. We continue to lead the industry in performance for scientific computing and deep learning. And with CUDA's programmability, we can continue to expand the value of our platform. For example, we recently announced RAPIDS, our CUDA acceleration stack, for data analytics and machine learning. In December, the first objective third-party AI benchmark called MLPerf became available, and NVIDIA captured the top spots in the six test categories for training deep learning models that we competed in. And in January, Google Cloud announced that NVIDIA T4 Tensor Core GPUs are now available in beta in its data centers in the U.S., Europe, Brazil, India, Singapore, and Tokyo. The T4 is a universal cloud GPU that accelerates a variety of workloads, including high-performance computing, deep learning training, and inference, broader machine learning, data analytics, and graphics. Our visibility remains low in the current cautious spending environment, and we don't forecast a meaningful recovery in the data center segment until the end of the year. However, we are working closely with hyperscales around the world to integrate NVIDIA TensorRT software and Tensor Core GPUs into their inference production flow. Inference currently drives less than 10% of our data center business, but represents a significant expansion of our addressable market opportunity going forward. We have also strengthened our product portfolio and go-to-market capabilities to address vertical industries that have an enormous data and analytics requirements, such as automotive, financial services, retail, healthcare, and consumer internet services. With our rapid software stack, NVIDIA can accelerate data analytics and machine learning as we have done in deep learning. And we made it easier for customers to adopt our technology by partnering with Cisco, IBM, NetApp, and Pure Storage to create pre-integrated systems that can be sold through their global IT channels. Moving to pro-visualization. Revenue reached $293 million, up 15% from the prior year and down 4% sequentially. Full year revenue was $1.13 billion, up 21% year on year. New applications like data science, AI, and VR, as well as the need for thin and light mobile workstations remain key ghost drivers for the business. Good key wins in the quarter, including Boeing, Google, LinkedIn, and Toyota for applications including AI and robotics. This past week, with our partners, HP, Dell, Lenovo, we announced the availability of Quadro RTX workstations. Quadro RTX is the most significant workstation GPU upgrade in 10 years. It will enable millions of designers and creative artists for the first time to work interactively with super high resolution media and photorealistic 3D rendering enabling them to be creative with dramatically improved productivity. Finally, turning to automotive. Q4 revenue was $163 million, up 23% from a year ago and down 5% sequentially. Full-year revenue reached $641 million, up 15%. The sequential decline was largely seasonal. The year-on-year growth was driven by the increasingly adoption of next-generation AI cockpit solutions and autonomous vehicle development deals, partially offset by declines in legacy infotainment. Last month at CES, we announced Drive Autopilot, the world's first commercially available level 2 plus self-driving car computer. This system offers sophisticated automated driving features that far surpass today's ADAS offerings, increasingly increasing the vehicle's performance, functionality, and road safety while the driver remains in control. To deliver these capabilities, DRIVE Autopilot uses multiple deep neural networks, surround camera perception both in and outside of the car, and significant AI processing capability. Systems from our Tier 1 partners, including Bosch, Continental, Vionier, and Zeta, were all on display at our booth. as announced back in October, was our first Level 2 Plus design win with cars slated for production in the early 2020s. Mercedes-Benz has also chosen NVIDIA for its next-generation autonomous vehicle and cockpit computer. This centralized AI computing system replaces dozens of smaller processors inside current cars. Drive Autopilot is a major milestone for NVIDIA and takes our high-functioning self-driving capability into the mass market. This will be an important year for robo-taxi pilots in initial Level 2 design wins. Moving to the rest of the P&L and balance sheet. Q4 gap gross margins was 54.7%, and non-gap was 56.0%, down sequentially and year-on-year, primarily due to a $128 million charge for GRAM boards, and other components associated with our lower-than-expected Q4 revenue and current market conditions. GAAP operating expenses were $913 million, and non-GAAP operating expenses were $755 million, up 25% and 24% year-on-year, respectively. GAAP EPS was 92 cents, down 48% from a year earlier. Full-year GAAP EPS was $6.63, up 38% from the prior year. Non-GAAP EPS was $0.80, down 53% from a year ago. Full-year non-GAAP EPS was $6.64, up 35% from the prior year. We returned $1.95 billion to shareholders in the fiscal year through a combination of quarterly dividends and share repurchases. As we announced last quarter, we plan to return $3 billion to shareholders through the end of fiscal 2020 in the form of dividends and buybacks. We repurchased $700 million during the fourth quarter of fiscal 2019. With that, let me turn to the outlook for the first quarter of fiscal 2020. We expect revenue to be $2.2 billion, plus or minus 2%. Gap and non-gap gross margins are expected to be 58.8% and 59%, respectively, plus or minus 50 basis points. GAAP and non-GAAP operating expenses are expected to be approximately $930 million and $755 million, respectively. GAAP and non-GAAP OINE are both expected to be an income of $20 million. GAAP and non-GAAP tax rates are both expected to be 10%, also minus 1%, excluding discrete items. Capital expenditures are expected to be approximately $150 million to $170 million. For fiscal 2020, we expect Q1 to mark the bottom as we pass the inventory correction and gain. We expect total revenue for the year to be flat to down slightly with growth in our 4M markets, compensating for the absence of crypto revenue and the excess salary from last year. We plan to grow OPEX in the high single digits this year, and we continue to invest in our focus growth areas of graphics, AI, and self-driving cars. Further financial details are included in the CFO commentary, and other information is available on our IR website. In closing, I'd like to highlight upcoming events for the financial community. We'll be presenting at the Morgan Stanley Technology, Media, and Telecom Conference on February 26th, and our next earnings call to discuss our financial results for the quarter of fiscal 2020 take place on May 15th. We will now open the call for questions. Operator, would you please take a question?

Thank you. And at this time, if you'd like to ask a question, please press star then the number one on your telephone keypad. And our first question comes from the line of Toshia Hari with Goldman Sachs.

Great. Thank you so much for taking the question. I had two questions. First, Collette, you talked about the weakness you saw in the 2070 and the 2080 and the quarter. I guess this question is more for Jensen. Are you concerned at all about your ability to convince and incentivize gamers? As Collette pointed out, it's more of a timing thing. And the second question is, inventory was up on the balance sheet. Collette, if you can just provide some color there and expectations going forward. Thank you.

Yeah, Toshia. When we launched the 2070 and 2080 games, It was the first time we've ever launched a new generation where the only available SKUs were very high-end. And in addition to that, the early boards that came out into the marketplace were the special edition and the overclocked versions. And the MSRP versions didn't show up for some short time after, you know, a couple of months after. And so the conditions weren't ideal, if you will. Now, we weren't able to launch into the mainstream segment with 2060 for all the reasons that I think everybody understands now. And so I think that this issue wasn't ideal. When you take a look at our situation now, every single graphics card had the best performance at its price point. And it remains so today. And And I think that right out of the box, it delivered excellent performance. It is true that everybody was hoping to see more games with RTX on day one. But it's such a new technology with ray tracing and AI for image processing that it's only really possible to make available with new games, which is tied to the schedules of new games. And now they're starting to come out. Battlefield V, Metro Exodus, I think the reviews from this week are just spectacular. People are finally realizing what it is that we were talking about and that it's possible with RTX technology, the combination of applying ray tracing and AI for us to deliver much more beautiful images without sacrificing performance.

To understand now the benefits of

Both of the game engines are going to incorporate RTX and ray tracing technology in the engine itself. And so all future games in the future will be able to take advantage of that. So that's a really big news, and I'm excited about that.

So, Toshia, to answer your second question regarding our inventory balance, our inventory balance at the end of Q4 rose just due to the weaker-than-expected finish to Q4. Inventory right now is primarily related to Turing Volta and DGX, and we don't expect any further write-downs as we have incorporated approximately 128 million of write-downs within the current Q4.

Your next question is from CJ Muse with Evercore.

Yeah, good afternoon. Thanks for taking my question. I guess on the commentary regarding a pause in spending in data center and a handful of deals that got delayed, can you give a little bit more color in terms of, I guess, what you're seeing across enterprise, cloud, high-performance compute, and I guess within that, you know, how you're seeing the ramp of T4? And I guess if you can kind of then speak to, I'm sure, embedded in the fiscal 20 guide is a pretty nice ramp into the second half. What are the key drivers, key milestones that you're looking for to see that business reaccelerate higher as we go through the year?

Yeah, hi, CJ. The slowdown is broad-based. We saw it across every vertical, every geography. There was just a level of cautiousness across all of the enterprise customers and the cloud service providers that we've not experienced in a while. And I think that it has to be temporary. The computing needs of Earth has not certainly been satisfied with what was shipped last quarter. And so I think that the demand will return and customers will return. Our situation in data centers is dramatically better year over year. And if you take a look at where we are – Our deep learning solution is unquestionably the best in the world. We introduced T4 with inference capability. It's the world's first universal cloud GPU, and it does everything that NVIDIA does all in one GPU and 75 watts. And so it fits into every hyperscale data center. We're engaged with Internet service providers all around the world, optimizing and porting their high-production data models, networks, so that we could deploy it into production. So we now have four different new growth drivers for our data center, in addition to deep learning and scientific computing. We have inference that we're actively working on. We have data analytics that's called RAPIDS. Some people call it big data, but data analytics and machine learning. The third is rendering. And because of the partnerships that we've developed and the excitement that people see around enterprises around the world, we've developed partnerships with large IT companies to pre-configure systems that make it easier for enterprises to be able to adopt our technology. So we have four new ways for us to grow our enterprise business. And so we're looking forward to when the pause releases and we'll get back to growing.

And your next question comes from the line of a buyback area with Bank of America, Merrill Lynch.

Thanks for taking my question. I just had a clarification and a question on the clarification. Gross margins, Colette, what is the normalized run rate for gross margins as you get your sales back to normalized levels? How should we think about the trajectory of gross margins and will there be any impact from the balance sheet inventory? And then on the question, Jensen, can you give us more reassurance that gaming is still a growth business? I understand that over the last year there's been a lot of confusion, there's been macro issues, but if you look at the number of gamers and the mix of products that they're buying, so essentially to sell through to gamers, has that been on an upward trajectory? And as part of that, When do you think we could see Turing exceed the demand you saw for Pascal? Thank you.

So I'll go first. Thanks for the question. On gross margin, our gross margin, the largest contributor to our absolute gross margin is really just the mix of our products. The mix of our products based on our market platforms, but also the mix of our products even within data center as well as within gaming. We provided guidance for Q1, which has a good level of confidence from us, and we'll see how it goes from there.

Hi, Vivek. The fundamentals of gaming has not changed. There are more gamers than ever. Games are better than ever. There's been a recent shift in the popularity of multiplayer competitive games. esports-like games. That's good for hardware. It lowers the barrier to entry because it's free to play with the exception of downloadable content. And so the barriers to entry is lower, but you can see that the excitement around Fortnite and recently with Apex Legend, PUBG is still popular, League of Legends is still popular. And so this genre of games is both both competitive, requires great hardware. It attracts a lot more players because it's social and you want to play with your friends. And it's much stickier because it happens to be social. It happens to be a game where you have to play with a whole bunch of other people. And so I think that gaming is vibrant as ever before. If you take the methodology that Colette described earlier and you averaged out our... our underlying gaming business, and you compare that to a year before, surely it grew. If you compare that, if you compare also the rate at which our gaming notebook is growing, I think that's pretty exciting. I think last year, we had mentioned it before, that our gaming notebook business grew 50% year over year, and just at CES, the number of new notebook designs that came out with Turing because of an invention that we created called MaxQ and because of the energy efficiency of the Turing architecture, you can now make notebooks that are really wonderful and also high performing at the same time. And so I think the dynamics are the same and gaming is going to continue to be a growth business.

And your next question comes from the line of John Pitzer with Credit Suisse.

Yeah, good afternoon, guys. Thanks for letting me ask the question. Colette, I appreciate all the data you gave us on trying to size normalized demand for gaming. What I have to ask, though, is if you're still going through channel inventory worked out in fiscal first quarter, it seems like to hit your four-year guide, the expectation is for gaming revenue, to accelerate well above that normalized level you talked about? What am I doing the math right? And if I am, kind of what gives you confidence throughout the year that you can see that kind of gaming growth off of these bottoms?

I'll start and I'll let Jensen finish that question. So, along the lines of Jensen's response in terms of what we do believe are the key drivers of gaming and everything's still intact in terms of gaming. both with our Turing architecture, the growth expected with our Turing architecture, as well as the growth from the notebook. We do believe we'll be great drivers as we head into the rest of the year. We'll have to wait and see in terms of how that plays out, but that is really the underlying reason why the growth will continue.

Yeah, I think the... Your math isn't wrong. The part that you probably didn't consider is notebook. Our GeForce notebook business is quite large.

That's helpful, Jeff. And then maybe as a follow-up, just on the data center side, clearly you've talked about new applications that should help grow your team inside a data center. I'm just kind of curious. The calendar fourth quarter of last year, I think, marked the first time that a competitor – had some meaningful volumes of GPU in the data center. You know, there's always talk about the Hyperscale guys wanting to do their own ASICs. What kind of anecdotal evidence can you give us to help us get more comfortable that what's going on here is more macro and not share loss either to competition and or architectural differences between GPUs and ASICs?

We don't see them in high-performance computing. And so I haven't found where we haven't met them in high-performance computing and deep learning and in the areas that we serve. And so competitively, I don't really, we don't see it. But the bigger picture, I think, is this, that the market segments that we serve, whether it's in deep learning, machine learning, data analytics, those segments are really quite large. And I think that it is unquestionably the future of high-performance computing is going to be highly data-driven, both computational methods, algorithmic methods, as well as data-driven methods. And so I think the fundamental trend has not changed. We have four new growth drivers, four new ways to grow in the data center. The first one, of course, is inference. We're making a lot of progress there. T4 is doing great. I think we're going to be quite successful with T-Force. You know, you've just got to keep saying that it has second-generation tensor cores, 75 watts, and you can use it for training. You can use it for inference. You could use it for remote graphics. You could use it for high-performance computing. And it fits literally into any hyperscale data center. The second way is data analytics. This is a brand-new thing for us. You must know that big data and using data to predict dynamics in the marketplace is really important in retail and e-tail and healthcare and financial services. And there's never been an accelerated approach to solve this problem for people. And because of the flexibility of CUDA and because of the performance of our architecture, Over the course of last year, we re-engineered the entire data analytics stack so that we can accelerate it. It's called Rapids. That work is really, really important, and I hope to give you guys updates on that on a regular basis. Rendering is a brand-new market for us because of Turing. Finally, we can render photorealistic images and accelerate away. There are millions of servers in the world that are driving render farms, and they're getting upgraded on a regular basis. And then lastly... We've been successful with CSPs because they're easy for us to reach. But the world's enterprises are far and many, and they're giant industries. And our company's sales coverage doesn't allow us to reach every single health care company and every single insurance company and retail company. And that's where our network of partners really come in. You know, we have great partnerships with HP. We have great partnerships with Dell. and Cisco and IBM, and now we've developed relationships with the storage vendors so that – and the reason for that is because most of these big data problems require a great deal of storage. And they both – they all see – they see the opportunities that we've created, and we came together to create pre-configured systems that are optimized and tuned and, you know, these high-performance systems that you can just bring into the company – prop up and install, and we're seeing a lot of great success with that. And so we have four different ways to grow our data center business, and we're enthusiastic about it. I'm optimistic about it.

And your next question comes from the line of Timothy Arcuri with UBS.

Thanks a lot. First I have a clarification, Colette. I just wanted to clarify what the mix is assumed for the fiscal Q1 guidance. Are you kind of assuming that data center and gaming are both kind of flat sequentially? And then I guess my question was, can you help us sort of shape the revenue through the year? You know, to get to your four-year guidance, you have to add roughly maybe $1.3 billion, $1.5 billion from, you know, where you are in fiscal Q1. How does that shape through the year? Thanks.

Yeah. Our guidance for the next quarter is a makeup of many different types of options across our market segment. We feel confident in terms of that roll-up. as we provide the guidance today, and we'll just have to see how that ends out. With the expectation that we will be flat or slightly down for the full year, you are correct in some case that we're going to have to build up to that over the course of the several next quarters. Likely, the second half of the year will definitely be stronger than the first half of the year, and that is our expectation at this time.

Yeah, Tim, one of the things to keep in mind is that we have four growth drivers. We have four growth businesses. Our data center business is growing. It's unquestionably that our footprint is larger than ever. Our provis business is growing. Our workstation business now has three ways to grow. One is rendering. The second is data scientists are now a workstation customer. That has never happened before. And our our software stack with Turing turns a workstation into an ideal data science workstation. And the third is finally we're able to make workstations into notebooks. And they're delightfully thin, using all the same technologies that I talked about for gaming notebooks. And so workstations is a growth business. And then lastly, our automotive is going to be a growth business. We've been investing, as you know, in self-driving cars and And this year we announced entry into Level 2+, our first foot into the mainstream marketplace of autonomous vehicles. And the first design one is Volvo, and we have others to announce. And so I think this is going to be a good year for self-driving cars as well. So we have four growth businesses. Our four core businesses are all growing.

And your next question comes from Stacey Rasgown with Bernstein Research.

Hi, guys. Thanks for taking my questions. First, I wanted to get again at the mix. This is a question for Colette. You had said that mix was going to be the primary driver of your gross margins. And I know that sequentially they're up, but if I correct for the inventory right down in Q4, the normalized gross margin this quarter was 61.7. And she's guiding it to 59, so it's down 270 basis points sequentially on flat revenues. So do I take from that guidance that that's an indicator of the mix between the businesses that's the primary driver of that gross margin degradation? Is there something else going on that we should be aware of?

Thanks, Stacy. We'll start with that first question on gross margin. You are correct. Mix is still the primary driver of our gross margin every single quarter. You have correctly reduced or, excuse me, changed our Q4 numbers to remove the overall inventory write-down. So when you look at Q1, there is a mix around our products that we plan on shipping by platforms, but also within our gaming business and within our data center business. We also have different gross margins that we're going to influence. This is our best estimate that we have at this time, and we'll see at the end of the program.

So what do you think the bigger – between those two, whether it's intra-business mix or inter-business mix? between the businesses or within the businesses, which one of those is the biggest driver of the gross margin degradation sequentially into Q1?

I think it's more on the inter pieces. Now, keep in mind our Q4 has a very low percentage of gaining as a total in terms of there, and then a different mix within there as we move to the next quarter as well.

So you think it's the mix between the businesses within, like you said, inter, is that mix between the businesses that you think is the bigger driver then?

The mix of the intra, both within the segment as well as between the segments.

Okay, thank you. For my follow-up, you know, Jensen, you and Collette, I guess you mentioned data center withdrawing, but if I've got four-year revenues sort of flat to down slightly and I've got gaming revenues down a bit, like you said, And I have ProViz and AutoGrowing. It's kind of really hard for me within the envelope of that guidance to get data center growing much, if at all. I mean, it could even be down within that. I mean, how are you thinking about the idea of data center growing within the context of the full year guidance that you've given?

Well, in the short term and in the near term, we have relatively limited visibility. We don't think it's going to remain this way. And with a little bit of tailwind, I think we could have a fairly good year. And so we'll just see how it turns out. You know, this is our year guidance for now, and we'll update you as we go. The fundamental dynamics doesn't change. The fact of the matter is the world needs more computing, and a lot of that computing is related to machine learning, data analytics, deep learning. It's related to the things that we're working on. And we have four new ways to grow our data center business. I think our deep learning position is as good as ever. Our scientific computing position is as good as ever. And we have four new ways to grow. We have inference. We have data analytics and machine learning. We have rendering. And now we're taking that entire stack to the enterprise. And so I think we have the right strategy. We have the best platform. And the utilization, the utility of it is really fantastic. And so with a little bit of tailwind, I think we could have a fairly good year, and we'll just report it as we go.

And your next question comes from the line of Joe Moore with Morgan Stanley.

Great. Thank you. I wanted to ask about, again, competition in data center. You know, AMD on their call had talked about graphics in their data center business being its biggest server, which is sort of north of $100 million a quarter. which surprised me. My sense is they're doing quite a bit different applications than you guys are, but maybe you could just give us some context around what they're doing and how you see the competition coming from within other graphics vendors.

Our data center business is really focused on computing, and we just don't see anybody. Our primary competitor is CPUs. That's really... the starting of it and the ending of it. And it's very clear. The vast majority of the world's data center only runs on CPUs today. But the advance of technology has slowed, and it's creeping along at a few percent a year, and unfortunately that's just not good enough. And so either data centers are going to continue to increase in CapEx, or they're going to have to find a new approach. And I think people are fairly enthusiastic about university, about accelerated computing. And I think our position is really quite good. And so I would say that those are largely the positions. If you think about competitively comparing our GPU to a competitor's latest GPU, I think the expectation was really high. And and didn't turn out quite that way. I think we've established that the Turing energy efficiency is much better. I think we've established that NVIDIA's Tensor Core architecture, as a result, allows our Volta to be four times the performance of the highest end of the alternative. And the T4 is one-fourth the power at the same performance. And so... So the benefit of having great architectural advantage, a really rich software stack, and an engineering that resulted in the energy efficiency that we've achieved generation in, generation out, I think those are really great advantages. And then lastly, because of the broad reach of our architecture, an OEM or a cloud service provider can adopt our architecture And the utility of it is going to be greater because there's just a lot more applications. And the best way to reduce costs for any utility is to increase its utility. And I think that our position there is really strong, as you could imagine.

And your next question comes from the line of Matt Ramsey with Cohen.

Thank you very much. Good afternoon. I have just a couple of questions. The first one, Jensen, it seemed like I mean, the Turing platform is delivering some amazing results, but as you talked about, relying on some new software features to do it. I wonder if you might talk about any steps you're taking in the roadmap to really upgrade performance for the installed base of games, and given the time that it might take for some of those software features to roll out. And then secondly... You noted in the pre-announcement something about the write-down having to do with DRAM, and obviously that commodity pricing has been volatile. Colette, is there anything you can talk about, about how big of an effect that might have on the business and on pricing overall? Thank you.

Yeah, thanks a lot, Matt. First of all, the Turing architecture is the highest-performing architecture at every single price point, and it's a big jump. from our last generation in every single way. Without ray tracing, the Turing architecture is the first GPU to do concurrent floating point and integer operations at the same time. The instruction per clock of the Turing processor is so much better than our last generation, so much better than what's available in the marketplace. The cache architecture is a big improvement. And you can just see it in all of the existing games. If you just measure the existing games without touching anything, Turing gives you a big boost. And that's before we talked about ray tracing. And we've already spoken about ray tracing earlier, Matt, and we know that every single game that are coming out, we're working with the developers to incorporate RTX technology and a very, very big deal. both Epic with Unreal Engine and Unity Engine are going to incorporate ray tracing. It is very, very clear that the next generation of computer graphics is ray tracing, and all of the work that we've done with RTX to move the industry forward is well worth it. But remember, that's just the graphics part of Turing. Turing comes with it several new opportunities for growth for us. The first is, of course, advancing games. advancing games for notebooks, advancing computer graphics, photorealistic rendering for film, all the work that we've done with TensorCore that we just talked about. It's our second generation TensorCore making it great for training as well as inference, a big leap for us for inference. And then lastly, all of the work that we're doing for data analytics and machine learning will take advantage of all of the capabilities of Turing. And so Turing is is a big deal for us. And that's one of the reasons why last year was so busy for us as we put Turing into workstations, into data centers, into clouds, into rendering, into video games. And so Turing is really a gigantic leap for us architecturally. We're really excited by it. I think the turbulent Q4 kind of overshadowed all of it, but in the final analysis, I think Turing was a home run for us.

And to also answer your question regarding DRAM prices, yes, they have definitely been volatile over the historical period. It is great to see them coming down in price. Over the long term, yes, that is beneficial to us from a gross margin perspective. So if we look out to the whole region later, we will probably be able to incorporate that into our gross margins.

And your next question comes from the line of Aaron Rakers with Wells Fargo.

Yeah, thanks for taking the question. It's kind of building on the discussion around the Turing platform and particularly to the gaming market. I'm curious, you mentioned in your prepared comments, you know, pricing of these new solutions was a bit of an inhibitor. Has the company invoked any changes in their pricing strategy around Turing? Then also I'd be interested in, you know, can you help us frame how important China is to the gaming segment and whether or not you're assuming that the China market rebounds in your annual assumption?

First of all, on the pricing part, the biggest inhibitor was that we couldn't launch our mid-range segment. The inability to launch 2060 was a big inhibitor for us, but we did so at CES. The launch is a great success. The reviews are fantastic. People love 2060. The price point is great. And so now we have a great stack from the mid-range all the way to enthusiasts. The other part, which could have turned out better, is at the time of the launch, there were so many special editions and there were so many overclocked versions that the price point appeared high. But now we have MSRP here. pricing for all of our segments. And so that's terrific. China is an important market. China is an important market, and it's an important gaming market. And I have every confidence it's going to rebound.

And your next question comes from the line of Mark LaPasas with Jefferies.

Hi. Thanks for taking my question. Could you just, a clarification, could you repeat what you said you thought the OPEX would grow in this fiscal year? And where's the focus of that investment? To what extent is it R&D versus SG&A and then maybe one layer deeper? Where's the focus of the higher OPEX? Thank you.

Sure. I'll start off to repeat what we indicated in our transcripts. We plan to increase OpEx in the high single digits over where we finished in terms of fiscal year 19. That is related to our opportunities that we see in front of us, gaming, AI, as well as self-driving cars. Our focus in terms of investment, we are a very R&D heavy significant company. but there are investments across the board, both in R&D as well as what we need in terms of go-to-market strategies to obtain these high markets in front of us.

And your next question comes to the line of Mitch Steeves with RBC Capital Markets.

Hey, guys. Thanks for taking my question. So I don't want to poke too many holes on the memory side and the downturn in gaming due to the crypto inventory, but if I think about kind of the gross margin profile, you guys almost reached 65%. So if I look out, let's call it a year or even let's call it 18 months and make it more of a kind of a long-term target, is there any reason why you guys can't get back up to kind of the mid-60s level?

Yeah, on your gross margin question, yes, we still have drivers within the mix of our products that allow us to grow our gross margin over the long term. Absolutely, and there is definitely a goal for us to continue doing that. There is a focus in terms of both the cost components. of what we do, but also moving the entire portfolio into the higher value-added platforms that we sell. So over the long term, absolutely, all of those things are still in place and intact that we can do. And we'll look quarter to quarter to give the best guidance that we can to help you see that.

And your next question comes to the line of Blaine Curtis with Barclays.

Yes, thanks for that question. Obviously, applause in data center, particularly with the hyperscale, is pretty well documented. I'm just kind of curious your perspective, particularly being AI still a growth area versus more run rate, big memory and CPUs. Can you give me a perspective as to how widespread it is, maybe number of customers or geographic perspective? I'm just kind of curious how AI is affected with this greater slowdown. Thanks.

Yeah, Blaine, the hyperscalers, their pause is probably the most dramatic. We still see a lot of activity in enterprise. It's just a much smaller base for us. But we expect it to be a much larger base in the future. And the reason for that is because most of the enterprises today don't use deep learning. They use an approach called machine learning. They might use things like decision trees or graph analytics or regressions or clustering or, you know, things like that, algorithms like that. And they run data analytics applications for business intelligence on a large amount of data. And they might be running it on top of a Spark stack that was created out of Berkeley and open source from Databricks. And And so if you recognize some of these things, that's what health care companies do and financial services companies do and retail companies do. They use it for fraud detection, predicting inventory, trying to make the best matches between riders and drivers, and trying to predict which route to take to deliver food, dinner to you. And so those kind of applications, most of the developers today use machine learning and big data analytics. And so we invested in this stack called Rapids and built our architecture called T4. And we're in the process of partnering with large IT companies to take the stack and the solution out to the world's enterprise. And so... I expect enterprise to be a fairly exciting growth opportunity for us. Meanwhile, the CSPs, their pause will end. The amount of computation they do is increasing. More and more of them are using deep learning, and we have inference opportunity with the work that we did with T4 and TensorRT. And so we've got a lot of exciting opportunities to go.

And your next question comes line of Pierre Ferrego with New Street Research.

Hi. Thanks for taking my question. A question for Colette. I'm trying to figure out how your position with cloud data centers evolved last year in financial year 2019. So if I look at the cloud capex, they were up about 70%. If I look at the CPU going into the cloud, according to Intel, was up about 60%. So I was wondering how much you guys have been increasing revenues within your cloud business. Did you grow faster than CPUs? Did you grow faster than overall CapEx? And then I have a follow-up on 2020. Okay.

Sure, let me start on that piece. As we talked about earlier on the call, our hyperscales and our hyperscales, many of them are also cloud providers, did start to slow down in terms of their purchasing in the latter half of the year. But the overall growth rate, as you can see from our data center business, grew more than 50% for the full year. Now, even with that strong growth, we are still a very small percentage of overall count that we see in the cloud providers or the overall hyperscales. We are likely one of their top priorities of areas of where they need to grow in terms of in their data center as they focus on AI, as they focus on the cloud instance importance of that compute and necessary. But we are still a very, very small percentage of it. So slowdown in the second half of the year, full year growth, phenomenal growth of 50%. and we track with their properties.

Thanks, Kelly. This 50% growth, that's for your overall data center business. Is that what, like, specifically your cloud hypercell business grew as well, or did that grow even faster than that?

Yeah, depending on the quarter, we will have a mix of what is hyperscale growth or what is cloud. And, again, in the fourth quarter, That was not a growth opportunity for us, but earlier in the year, definitely it was.

Great. And then quickly, maybe Jensen on 2020. So you seem to be very cautious on the data center outlook. When we listen to players like Google and Facebook and others, they all seem to be still very keen to grow their overall spending this year. And when we listen to other providers around you, they all seem to be – I mean, they all demonstrate a bit of confidence that in the second half, spending should resume. Is that just because they are being more optimistic than you? Do they see something you don't see? What do you hear from your clients about the second half?

Well, we're enthusiastic about the second half. We're enthusiastic about our position, and we're enthusiastic about the solution we offer. And, Pierre, as you know, we've also expanded our application and our market reach. We guided – I think probably the biggest takeaway is we guided Flight to Slightly Down for the whole year, for the whole number. We do have four growth drivers. And maybe the best way to think about it is we should just wait and see how it goes. You know, I think considering where we are, I feel pretty good about our guidance. But I feel even better about our strategic position. And so I look forward to working through the year with you guys.

And our last question comes from the line of Harlan Sir with JP Morgan.

Good afternoon. Thanks for taking my question. On the China gaming weakness, is it the slower economic environment or is it sort of government policy related? Because we know that the China government has had a freeze on new gaming approvals, although they recently started to approve new games. This ban has been in place, I think, since the first half of last year. So Given what you know of the business, how much of the China weakness is, you know, coming from the China gaming bands versus just the overall slower economic environment?

I don't know that we could tear that apart, tease that apart, Harlan. We just know that China, the consumer market, is relatively slow towards the end of the year. And... But, you know, the China economy is in the final analysis of growth economy. And so we're looking forward to it recovering. And gaming is one of the most important pastimes of that culture. And so I'm excited about our prospects there. You know, all the things that we're seeing in the near term, Colette has done a really good job describing. And as we... leave the bottom and leave this inventory issue behind us, we're super well positioned. We have a full stack of RTX. The Turing architecture is fantastic. It is unquestionably the best in the world. We have the best performance at every single price point. And we have great notebooks that the market can now buy. And so I'm looking forward to to reporting our status with you guys as the year goes on. It should be a good year.

Thank you. I will now turn the call back over to Jensen for any closing remarks.

2018 was a record year, but it was a disappointing finish. This quarter, we expect to put the channel inventory issue behind us and get back on track. As the pioneer of accelerated computing, our position is unique and strong, and the opportunities ahead in graphics, high-performance computing, AI and autonomous machines remain enormous. We are as enthusiastic about these growth opportunities as ever. Thanks, everyone, for joining us today.

This concludes today's conference call. You may now disconnect.