Ladies and gentlemen, thank you for standing by. At this time, I would like to welcome everyone to the NVIDIA Financial Results conference call. All lines have been placed on mute to prevent any background noise. After the speaker's remarks, there will be a question and answer period. At that time, if you have a question, please press the one followed by the four on your telephone. If at any time during the conference you need to reach an operator for technical assistance, please press star zero. As a reminder, this call is being recorded Thursday, August sixth, 2015. I will now turn the call over to Mr. Arnab Chanda, Head of Investor Relations at NVIDIA. Mr. Chanda, you may begin your conference.
Thank you. Good afternoon, everyone, welcome to NVIDIA's conference call for the second quarter of fiscal 2016. 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 today's call is being webcast live on NVIDIA's investor relations website. It is also being recorded. You can hear a replay by telephone until thirteenth of August 2015. The webcast will be available for replay up until next quarter's conference call to discuss Q3 financial results. The content of today's call is NVIDIA's property. It cannot be reproduced or transcribed without our prior written consent. During the course of this call, we may make forward-looking statements based on current expectations. These forward-looking statements are subject to a number of significant risks and uncertainties, 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, the reports that we may file on Form 8-K with the Securities and Exchange Commission. All our statements are made as of today, the sixth of August 2015, 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, Arnab. Second quarter revenue was $1.15 billion, up 5% from a year earlier and well above our forecast of $1.01 billion. Growth was driven by our GTX gaming platform's continued momentum, the gaming market's underlying strength, and increased automotive sales. Viewed from a segment perspective, GPU revenue was $959 million, up 9% year-over-year. Tegra processor revenue was $128 million, down 19% from a year earlier. In the face of challenges on both the macro and semiconductor fronts, NVIDIA is executing well, helped by our sharp focus and strong market positions. Our strategy for creating platforms for gaming, enterprise graphics and virtualization, HPC and cloud, and automotive continues to serve us well as visual computing becomes increasingly more important in these industries. In Q2, these four platforms contributed 85% of our revenue, up from 68% a year earlier.
First, let's start with gaming, a market that's global and growing. Sales of PC gaming systems this year are estimated at $28 billion, with continued expansion expected. esports, professional online gaming competitions, is now a major entertainment category with a growing audience of 130 million. This week's championship tournament of Dota 2 offers a purse of $18 million, compared with $10 million for the PGA open golf event. There now are over 80 million people enjoying multiplayer online battle arena games, twice that of three years ago. Our gaming revenue rose 59% year-on-year to $661 million. This exceptional performance reflects the buoyant gaming market, as well as the continued strength of our Maxwell-based GeForce GPUs for the enthusiast and performance categories. The newly launched flagship GTX 980 Ti has received outstanding reviews.
AnandTech, for example, called it the card that all other high-end video cards are measured against. Across all regions, gamers are flocking to advanced GPUs to play enormously popular games with high production values like Grand Theft Auto V and Witcher 3. Demand is also being driven by excitement around new gaming technologies, including 4K, VR, and DirectX 12. During the quarter, we rolled out Shield Android TV, a revolutionary device that connects your TV to an infinite world of entertainment. The world's first 4K smart TV, Shield was created for a new age of television, where movie, music, games, and apps are delightfully simple to enjoy on one platform. Shield is the first of the next-generation smart TV devices. The first were streamers like Chromecast and Roku. The next will feature 4K, powerful processors, and rich app stores.
We are excited to take Shield globally as supply comes online and exciting applications become available. Moving to enterprise. Our enterprise graphics and virtualization revenue declined 14% year-on-year to $187 million. Our Quadro business was hampered by the weak refresh cycle and workstation market. We look forward to the launch in the year's second half of workstations based on Windows 10, Intel Skylake, and new Quadro GPUs. We continue to make good headway with our GRID virtualization platform, which enables companies to deliver graphic-rich applications, like design tools from Adobe and Autodesk, to employees on any device, anywhere. The number of GRID customers has more than tripled in the past year to over 300 from a broad range of industries. Among the more than 40 customers added this quarter are the University of Southern California, Textron, the maker of Bell Helicopter, and Gensler, the global architectural firm.
Nearly a quarter of the Fortune 100 are now in trials or in deploying GRID. In HPC and cloud, revenue declined 15% year-on-year to $62 million, reflecting lumpiness from major deep learning projects. We strengthened our offerings in the rapid growing field of deep learning, building on momentum from our March GPU Technology Conference. We launched DIGITS 2, cuDNN 3, GPU accelerated software that doubles deep learning performance for data scientists and researchers. Deep learning adoption continues to grow, with multiple developers for voice recognition platforms serving vast numbers of consumers. We are now engaged with more than 3,300 developers and companies interested in this area. Moreover, nearly 30,000 professionals on LinkedIn identified themselves as having proficiency in CUDA, our language for programming GPUs. In addition, GPU computing in the cloud continues to gather momentum, with multiple leading cloud services providers offering GPU computing resources.
We are also excited by the opportunities presented by President Obama's call last week for the U.S. to build an exascale supercomputer, one 30 times faster than today's most powerful systems. Just as GPU technologies are being used to build the pre-exascale computers coming online in the years ahead, we believe they will be important for exascale. Finally, automotive revenue rose 76% year-over-year to $71 million. Momentum here continues with new design at new and existing customers. Hitting the road this quarter in North America, for example, is Audi's TT sports car, which features a full digital dashboard powered by Tegra. In addition to our infotainment cockpit business, we are working with more than 50 companies interested in using NVIDIA DRIVE PX in their autonomous driving efforts. Last quarter, we announced our intention to sell or wind down our Icera modem operations.
In the absence of a viable buyer, we are winding it down. We have incurred this quarter $103 million in restructuring charges, net of tax. GAAP gross margin was 55.0% and incorporates our warranty provision for SHIELD tablet. Non-GAAP gross margins was 56.6%, in line with our outlook. Margins were slightly lower than the previous quarter as the strength of GeForce GTX revenue was offset by a lower mix of enterprise and accelerated computing GPUs. GAAP operating expenses for the second quarter were $558 million, inclusive of $89 million of restructuring and other expenses associated with the wind down of Icera modem operations. Non-GAAP operating expenses were $421 million, including litigation charges, slightly below our outlook. GAAP net income was $26 million, and GAAP earnings per diluted share was $0.05, including $0.19 charge for Icera and $0.02 charge for the recall.
Non-GAAP net income was $190 million, and non-GAAP EPS was $0.34. Earnings grew 13% year-over-year. Turning to some key balance sheet items. In Q2, our cash and marketable securities balance was $4.51 billion. During the second quarter, we paid $52 million in cash dividends and $400 million associated with an accelerated repurchase agreement. We returned an aggregate of $452 million to shareholders, following $99 million in the first quarter. Turning to the outlook for the third quarter of fiscal 2016. We expect revenue for the third quarter of 2016 to be $1.18 billion, ±2%. We remain excited about our business prospects. Gaming continues to accelerate, 4K, VR, Windows 10, and a pipeline of exciting games will lift it further. GPU accelerated data centers are expanding. Deep learning is a new, exciting application. The market for car computers is expanding.
We have excellent positions in each of these growth markets. Our GAAP and non-GAAP gross margins are expected to be 56.2% and 56.5%, respectively, ±50 basis points. This outlook is slightly below Q2 margins, reflecting our product mix. GAAP operating expenses are expected to be $484 million. Non-GAAP operating expenses are expected to be approximately $435 million.
We expect fiscal 2016 non-GAAP operating expenses to be approximately flat with fiscal 2015, excluding litigation costs, which are anticipated to be in the range of $70 million-$90 million as we defend our intellectual property. GAAP and non-GAAP tax rates for the third quarter of fiscal 2016 are expected to be 22% and 20%, respectively, ±1%. The above GAAP outlook amounts exclude restructuring charges, which are expected to be in the range of $15 million-$25 million in the second half of fiscal 2016. We will now open the call for questions. Operator, will you please poll for questions?
Certainly. Ladies and gentlemen, as a reminder, if you'd like to queue for your question, please press the one followed by the four on your telephone. You'll hear a three-tone prompt to acknowledge your request. If a question has already been answered and you want to withdraw your registration, please press the one followed by the three. One moment while we wait for our first question. Our first question comes from the line of Harlan Sur with JPMorgan. Your line is open. Please proceed.
Hi, good afternoon, solid job on the quarterly execution. If I look at the pipeline of blockbuster games coming down the pipe in October and November, Rainbow Six, Fallout 4, Black Ops III, and so on, this is shaping up to be a stellar year for new game releases. On top of that, GTX 980 Ti seems to be doing very well in the market. Why wouldn't we expect more seasonal growth in the business here for October? I guess the question is maybe offsetting that, are you guys still seeing headwinds in the enterprise, in cloud, or PC OEM segments in Q3?
Harlan, there are several, I would say four major drivers in the gaming industry. The first one is blockbuster titles, as you mentioned. The install base of game platforms is growing so greatly that developers can now invest an enormous amount in the production value of great games. The second half, you're going to see some pretty amazing games. You didn't mention Star Wars. Myself along with everybody else are practically out of our minds waiting for it. I think this is going to be a huge game. Call of Duty, of course, Assassin's Creed, of course, Metal Gear Solid is going to be huge. The second half of the year, we're going to see some really huge games. The second dynamic is just esports. Esports is probably the most social gaming platform that we have.
The reason for that is because you want to play with your friends. It's also a form of gaming where milliseconds could make the difference between winning or losing, so performance of your platform matters a great deal. It's a solely PC phenomenon, and it's a global phenomenon. In just a few years' time, esports has doubled to almost 100 million gamers around the world. This is, as I mentioned, it's social. Therefore, the network effect is quite important. That's why you can imagine it doubling in just a few short years. The larger the numbers, it just seems like the larger the numbers become. The third driver for gaming this holiday season is the combination of several platforms coming together finally. 4K is here. The pricing is fantastic now. Windows 10 is great.
The Skylake platform from Intel is going into production now. The fourth driver is a driver that we all have heard a great deal about, and surely the experience is as wonderful as the promise, is VR. VR will go into production the second half of the year. Each one of these drivers are pretty large scale. So I would agree with you that my expectation would be that gaming is going to continue to grow, and high-end gaming particularly is going to grow nicely. The guidance that we provided is guidance that we think makes sense at this time. We'll report on how it turns out at the end of the quarter.
Okay, thanks for that, Jensen. The team has got this great pipeline of auto design wins, and with the new model changeover that's happening now that you guys mentioned in your prepared remarks, how is this expanding the number of new models that are rolling out here in the second half of the year that are using the Tegra platform and that you will also be recognizing revenue on a go-forward basis? I guess what I'm trying to figure out is auto contributing to the growth in Q3 and second half of this year?
Let's see. Well, auto is growing over 70% a year. We have eight million cars on the road now. We have 30 million more cars coming with our Tegra platform in it. Our recent initiative of utilizing deep learning in our CUDA processors, the Tegra processors, for autonomous driving is really gaining a lot of traction. We're now engaged with about 50 companies around the world developing autonomous vehicles. The advantage that we have is the ability to capture camera inputs from all around the car, in addition to other sensors, from LIDAR to radar to sonar. Sensor fusion is just such an important part of autonomous driving that this Tegra platform, we call it the Drive PX platform, is really incredibly valuable to the car companies.
We're developing autonomous driving vehicles with many of them at the moment. I expect the car business to continue to grow, and hard to exactly to say how it's going to do the second half. It depends on how many cars get sold. There's no question at all that year-over-year, the growth is going to be steady, and it's going to keep going for several years.
Thanks a lot, Jensen.
Yeah. Thanks a lot, Harlan.
Our next question comes from the line of Stephen Chin with UBS. Your line is open. Please proceed.
Hi. Thanks for taking my questions. Jensen, first one for you, if I could. In terms of VR, as you mentioned, some of the new products will go into production later this year. As far as a bigger demand inflection for that technology goes, approximately when would you think that would be really taking off in terms of unit volumes and also helping to uplift your GPU business further? Related to that, just given that you have a lot of development on the software side as well as on the hardware products in terms of your Shield products, does it make sense for NVIDIA to go more vertically integrated for VR products?
The first question is, it's hard to say, but we know that several hundred thousand development kits were sold of the Oculus glasses. If we just take a couple 200,000-300,000, they tend to require higher-end GPUs because you're driving stereo, you need to have extremely low latency, and you need to have 90 frames a second, so that you don't see flicker. Those factors combined requires you to have a relatively high-end GPU. Several hundred thousand new high-end GPUs at $several hundred a piece adds up a bit. That would be for the first year. It's hard to say exactly how big the VR market becomes ultimately. If you've had a chance to try it's really a whole new experience. Being immersive, it just doesn't begin to explain how wonderful it is.
I think that the buzz around VR, from all the people that have experienced it and the content developers that have spoken so highly of it, is consistent with the reality at this point. It's taken many years, well, many decades, frankly, several decades to have brought VR to this point. I think that it surely looks like it's ready. As to the question of what can we do more in VR, there's quite a bit that we can do in VR, we're working with Oculus, and we're working with Valve to really integrate the entire rendering pipeline, the entire experience into a seamless one. That's an extraordinary amount of work to do already. In the other industries, we're going to see a lot of adoption in VR.
This is an area where our workstation business, our Quadro business, is going to get a really great lift. The reason for that, of course, is being able to see and experience a product in virtual reality before you manufacture it. Whether you're designing a building or designing a car or creating a showroom, the ability to be able to put on VR and experience it before you build it is quite an extraordinary benefit. We're going to see some exciting adoption in the design community as well. There's plenty of work to do, plenty of innovation at the rendering layer for us to do. I think that's plenty for now.
Great. Thank you for that, Jensen. As my follow-up for Colette, on the OpEx side for the October quarter, can you talk about some of the puts and takes for the OpEx spend and sort of what the main driver for the sequential increase for non-GAAP OpExes? Thank you.
Sure, Stephen. When we look at Q3 has a couple of key things in it. We get ready for the holiday season, revenue-related OpEx as we think of marketing and others, is also in our Q3. It is also the time that we take the opportunity to look at the salaries of our employees. Our overall salary increases for worldwide do take place in Q3.
Great. Thank you very much.
Thank you. Our next question comes from the line of Vivek Arya, calling from Bank of America Merrill Lynch. Your line is open. Please proceed.
Thank you for taking my question. Very good result, very good growth in gaming. Jensen, I just wanted to go back to the quarter you reported. I'm wondering what changed versus your original expectations. Were you just too conservative before, or what changed in the quarter to really create this kind of positive surprise? How are you making sure there is no buildup of excess inventory given just broad macro conditions seem so volatile right now?
Well, answering the second part first, we monitor sell-out in the channel literally every day. That's how we manage inventory. We don't manage inventory on sell in; we manage inventory on sell out. Having our own specialized platforms instead of selling into OEMs gives us a lot more visibility, frankly, than we've ever had before. As you know, we really changed our business model from a company that sold components to PC OEMs and mobile OEMs, we're now focused on specialized platforms for specialized applications, and we have identified four applications where we could add a lot of value. In each one of these four applications, we're so much closer to the marketplace, that the visibility is just surely better.
If you go back to the first part of your question, I think 90 days ago, just about everybody was rather uncertain about what was happening. There were a lot of changes going on around the world. Surely, there's still changes around the world. Vivek, I think your plane is here.
Sorry. Sorry, I just-
It's all right.
Maybe I just.
It's all right. Anyways, that was the best we could do at the time.
Please go ahead.
We provided the best judgment we had. It turned out to have been much better than that.
Got it. Jensen, as my follow-up, could you talk about your average content in the car? Where is it now? Where can it get to? I think you mentioned the 30 million cars, I think you mentioned 25 million before, if I take the 30 million cars, say you've 10 years, it means it's roughly 10% share of all the cars that are sold in a given year. You're penetrating just the high-end of the car. Please talk about what the direction is. If you could quantify it, that'll be very useful. Then in terms of penetration, is it just the high-end, or are you able to get into the mid-range and on the mass market as well? Thank you.
Well, we're trying to win as much as we can while delivering value to the customers that we select. Our value proposition, as you could imagine, tends to be where visual computing is really important. It could be because of large surround and dynamic infotainment systems, virtual cockpits where you've got a digital dashboard and it's rich and it's moving and infotainment's integrated into it, and maps integrated into it. It might have several virtual machines that are running at the same time so that infotainment and digital clusters are not integrated into one computer, and they provide safety among itself. It could be in the future, several computers inside the car, some in the front, some in the back, as displays become more affordable and people's expectation for rich graphics continue to grow.
You could expect us to have infotainment systems, virtual clusters, as well as drive computers that are used for driver assistance or autonomous driving. There's a lot of different ways that a computer is going to be included in cars. I think largely, the trend is very clear that in the future, the best way to add value to products is to connect them to the internet and make them computerized, make it a software-defined product. I think your car is going to become a software-defined car. The amount of processing inside will continue to grow at an exponential level. As a visual computing company, we think we could add a lot of value here. I think our software content is going to continue to grow. This isn't like our parents' car industry anymore.
This is really a computerized car industry, and cars are going to be supercomputers in the future.
Got it. One last one for Colette, if I may. Cash is almost 40% of the market cap now. A lot of buybacks. I'm wondering, Colette, is still that the best use of the cash, or are there other options for the large cash that you've built up? Thank you.
Thanks for the question. We tend to look at our cash balance net of our debt as well, as we have about $1.5 billion outstanding debt. Then with our synthetic lease that we also just completed, we'll add on top of that. Our net is probably a pretty good, better barometer of what percentage of our value is associated. We've looked at our capital return program heavily. We are excited at the amount that we are returning this year, close to $800 million, which is nearly in the high percentage of our free cash flow. We'll continue to look of opportunities for our cash, whether that be investment in the business, more capital return, and/or M&A. Thanks for the question.
Thank you.
Thank you. Our next question comes to the line of Deepon Nag calling from Macquarie Research. Your line is open. Please proceed.
Yeah. Thanks a lot for taking the question. Looks like the GTX 980 Ti had a very strong quarter. Based on third-party benchmarking, it looks like it's extremely competitive with High Bandwidth Memory parts from your main competition. I think it's something we all underestimate is just the value of your software and the advantage it gives provides you despite maybe some technological changes on the silicon. Maybe you could talk about your GPU roadmap for the next 12 to 18 months, and whether you think your software advantage can actually allow you to maintain share even before or without an architectural change.
Yeah, Deepon, first of all, I appreciate the comment about software. We are really, really proud of our software. In the final analysis, a computer is not just a bunch of silicon. The computer is silicon and software and algorithms and system software and all kinds of stuff in the middle. You can't just be a good transistor slinger, if you will, and you have to create ultimately a great experience. In the final analysis, NVIDIA's a visual computing company, which is everything that includes architecture and silicon and system software and algorithms and deep understanding of applications and everything in the middle, and being able to optimize across that entire stack. First of all, I appreciate that comment.
However, it turns out that it's not solely because of our software that the Maxwell architecture delivers so much performance, even though the memory that we use is relatively available. The reason for that is because Maxwell includes a groundbreaking piece of memory technology inside the chip that compresses, reorders, and does amazing things so that we effectively received one and a half times our bandwidth, through amplification of compression and others. It's a brand-new technique that has taken us several years to create. As a result, we got a lot of effective bandwidth out of Maxwell because of that. The benefit of doing this approach, of course, is that we can scale it across the entire family of Maxwell GPUs, all the way from the Titan X, to the GTX 980 Ti, to the 980, to the 970, to the 960.
We could take it all the way to the mainstream because of this technique. I think your question about long-term roadmap, I like to reserve the opportunity to surprise you with some new products in the future. However, one thing that I can tell you is this, and I think maybe you're alluding to it. People have observed us increasing performance 2X every year and a half for quite a long time, two times, maybe even a year's timeframe, and we defied Moore's Law. The reason for that is because Moore's Law isn't a transistor law necessarily. Moore's Law is an innovation law, and we can innovate across multiple layers, from architecture to the silicon, to the silicon design, to all of the system components around it, to the algorithm on top, to the system software that you mentioned just now, great software.
We can innovate across so many layers independently and together that we can bring value to the marketplace irrespective. For example, you're probably alluding to the fact that we stayed on 28 nanometer for quite a long time, and we did. During that time, we increased performance by a factor of four to five in one node. I expect that we'll be able to continue to do that.
Great. Thanks a lot for that. Maybe if I can talk about the workstation, the Tesla business. It seems like it slowed down a little bit in Q2, and obviously Tesla a lot of high-performance computing and cloud customers are very lumpy in their purchases. How should we think about the growth for the entire year and if you expect some kind of acceleration either in Q3 or Q4?
We saw some slowdown in enterprise buying as other people have. I would say that in the case of our workstation platform, it probably was also postponed because the new platform, architecture from Intel, was delayed by a little bit. Skylake is a really excellent platform. It has higher speed connectivity, it has the ability to support more memory and include DDR4 memory. In the space of workstations where the dataset is very, very large and moving data in and out of the computer is so important, having that connectivity and IO capability is really important. I think that, along with everybody else, we're all delighted that Windows 10 is now out and Skylake is now ramping in production. Hopefully the OEMs can launch a new line of great workstations.
Long-term, the way to think about it is this. The way I look at it is Quadro is all about design and creativity. There's no question that more and more people are using digital approaches to design things and create products. So long as people design and create in digital, it's going to be a growth opportunity for Quadro. As for Tesla, it is a growing business, but it's still a lumpy business. The reason for that is because every so often the data center comes in and/or a supercomputer comes in and buys tens of thousands of chips and to outfit a data center or outfit a supercomputer. We're going to expect to see lumpiness. The trend is clearly visible, that accelerated computing is the way to the future.
In fact, I don't think that we've ever experienced something like this, but President Obama, this last week, signed an executive order that multiple agencies in the government will collaborate with education and industry to create an exascale computing platform. Just to put that in perspective, exascale is literally 30 times faster than the fastest supercomputer we have in our country. With President Obama's executive order, we're going to be able to create not only the world's fastest supercomputer, but 30 times faster than what we currently have in the U.S. Just to help you understand some of the sentiments around that executive order, it's really to help the nation find a path forward in the post-Moore's Law era, and it says it so explicitly. In order to achieve an exascale computing capability in some reasonable time without power going through the roof.
Let me just give you one example. In order to reach approximately one exascale in the next several years or half-decade, if we just use CPUs alone, we would need a power plant of about a billion watts. That's just effectively powering a computer with a nuclear power plant. However, using accelerated computing, which is the way the Titan supercomputer, which is the fastest supercomputer in America today, powered by 18,000 NVIDIA GPUs, that consumes only 10-20 MW, which is about the power envelope, if you will, of a supercomputer today. What we're going to do is we're going to really endeavor to continue to advance in accelerated computing. It's very clear now, this is the path forward.
For many applications, whether it's life sciences research or energy research or work in big data and artificial intelligence, that this approach to computing is really a fantastic way forward. That's very helpful. Thank you. I'm very enthusiastic about Tesla. Thanks. Terrific. Thanks.
Thank you. Our next question comes to the line of Hans Mosesmann, calling with Raymond James. Your line is open. Please proceed.
Thanks. Hey, Jensen, can you give us a rundown on the competitive dynamics over the past quarter on the auto side? There's lots of players that are out there, it gets a little confusing. If you can give us a rundown, who are the real guys long-term that could emerge as a competitive dynamic for you guys? Thanks.
Yeah. Let's see how to answer that. I guess there's several different places in the automotive industry where innovation is happening. I think largely infotainment is a relative commodity. It is possible to build infotainment systems relatively easily. However, a virtual digital cluster where the digital dashboard, the mapping system, the infotainment system, all unifies into one virtual display, that is relatively complicated work. The reason for that is because there's several operating systems at play. It's a very powerful virtual computer, virtual machine that's running. The quality of software needs to be utterly exquisite. Otherwise, problems in one functionality like infotainment system could cause all kinds of trouble for you with your digital cluster. That's an area where we're doing a lot of very good work, we're finding a great deal of interest.
Our software capability and the performance of our processors is a real advantage. Another place where we see a lot of traction is autonomous driving. In the first implementations, of course, it is driver assistance, observing something, an object in front of the car, and applying the brakes if the car is not slowing down. That's relatively easy to do. That's far away from autonomous driving. Autonomous vehicles are going to require sensor fusion. Radar's going to get involved, LiDARs will get involved in the future, sonars get involved, of course, and cameras all over the car. This particular approach of autonomous driving, autonomous vehicle, I think is a rather promising approach and is recognized around the world as a reason forward.
Where sensor fusion comes in, the processing capability of Tegra and CUDA, connected with smart cameras, is a really wonderful approach going forward. That's one of the reasons why we're engaged with just about every autonomous platform that I know of anyways. The Drive PX is really doing well there.
Great. Thank you.
Yeah. Thanks a lot, Hans.
Thank you. Our next question comes to the line of Ambrish Srivastava, calling with BMO Capital Markets. Your line is open. Please proceed.
Hi. Thank you. Colette and Jensen, if I missed it, I apologize. I'm just trying to get a little bit more of my arms around the guidance for the third quarter. Based on the comments that you've provided, sounds like PC gaming will be up Q over Q, then auto's up. What about PC OEM and the enterprise business? Thank you.
Well, the PC OEM and the enterprise business, there are reasons to be optimistic. In the PC OEM business, of course, Windows 10 has come out, and before it came out, there's really no reason to buy a PC in advance of it. Skylake is out, and both of these platforms should contribute to growth. We'll see how it turns out. We don't control the PC OEMs, and we stay very close to them, and whatever opportunities comes our way, we surely be delighted with it. With respect to enterprise, I guess your guess is as good as ours, and when the enterprises come back and purchase. I think our position with workstations, Quadro visualization, is surely excellent. Our position with accelerated computing with Tesla is surely excellent. Both of them are highly differentiated in the marketplace.
Their niche is very clear, and the value we deliver is very clear. When the enterprise comes back and buy, I think we're going to do very well. I think the answer is we should just wait and see.
For both Tesla and for Quadro, Jensen?
Both of those are highly dependent on enterprise buying cycles and enterprise purchases. I think, as I mentioned, I'm optimistic about both of them. I'm enthusiastic about both of their positions. With respect to the guidance, the best thing to do is just wait and see.
Okay, thank you.
Thank you. Our next question comes from the line of Rajvindra Gill, calling from Needham & Company. Your line is open. Please proceed.
Yeah, thanks for taking my questions, congrats on good results in light of-
Yeah
the tough macro environment.
Thank you.
Given your new segment reporting or relatively new segment reporting, it kind of represents a kind of a transformation in your business. You didn't give out the PC OEM revenue, but if I could back into it does seem like the PC OEM revenue is down another 50% year-over-year, if I did it correctly, after following about 40% year-over-year in Q1. Now the mix of the business is where PC OEM is now only maybe 15% of sales, down from, say, 30% of sales at the start of last year. I just wanted to get a sense of that. Is that correct? The new growth segments, if you could talk a little about what's going on in gaming and auto, because gaming and auto and enterprise are now almost 85% of revenue.
Yeah, Raj, first of all, I appreciate the observation. If you look at it year-to-year, I'll just use rough CEO math here. Year-to-year, our OEM business, PC OEM business, and mobile OEM business are down nearly $200 million. Yet the company grew about $50 million, which basically says that the specialized platform strategy and these four applications, these four markets that we've targeted, have made up and some, the decline of OEMs. The benefit is even better than that, of course, as you surely would know. The resilience of the business is greater. The value that we add to the marketplace is much, much greater, which will eventually reflect itself in gross margins. The business model is more resilient, and our end market engagement is much deeper.
I think in just about every possible way, it's a higher quality business, it's a more resilient business, and it's a business that our shareholders, frankly, would enjoy better long-term. It gives us much larger growth opportunities. Whereas a long time ago, I guess maybe a couple of years ago, people asked why we could grow a $4 billion graphics business in a market that was only $4.8 billion large. Nobody would think twice if we could grow into a $100 billion gaming market. If we could add more and more value to a much larger end market, I think it's just a much better company.
I think that's all you're observing is the transformation, the successful transformation of our business model from a component business model to OEMs to a platform model with a specialty in an area that is related to visual computing.
Just as a follow-up to that point, your gaming segment is now 57% of sales versus 38% of sales this time last year. The growth rate actually accelerated year-over-year from Q1 to Q2. I think what the market underappreciates is the gaming market itself and how fast that's growing. I was wondering if you could discuss, you hit a little bit before, but why is the gaming market so robust, and is there anything going on by region? Any color there as well in addition to your commentary earlier in your call.
There are several dynamics that are happening, and it's very, very mathematical. The first dynamic is the size of the market is growing, and it's driven a lot by esports and MOBA. This is a type of game that's quite viral, and it is also very social. It has the ability to grow to larger numbers whenever it reach larger numbers. It has a network effect. The reason for that is because you want to play with your friends. If two of your friends are already playing a particular game and you want to hang out with them, and when they're playing, they're talking and they're enjoying each other's company, they're hanging out, then the third friend's going to get invited and the fourth and the fifth, and before you know it, just about all your friends are playing.
It tends to work in a social effect, and it grows from large numbers. The second is the production value of games. I don't know if you've noticed or not, they're almost cinematic now. That's all made possible because of the rich algorithms that we put on top of our GPUs, we call GameWorks because of our GPUs. When the production value is high, it consumes a lot more GPU, which is very different than movies. A Blu-ray movie, whether it's beautiful or not so beautiful, is basically Blu-ray. For computer games, in order to have beautiful games, you need more powerful GPUs. That's very mathematical. The third is 4K displays is a lot more pixel than 1080p, eight million pixels over two million pixels. It's a lot more pixels.
The number of pixels we can process is directly related to the power of the GPUs. All of a sudden you add VR on top of that. You need very high-resolution displays. You need two of them because you're in stereo, and you also want to render it at a much, much higher frame rate. Instead of 30 frames a second, you want to render it at 90 frames a second, and so it's times two, times three. Many of these type of experiences are driving the GPU adoption upward, and we're definitely seeing that trend. Both the larger market and increasing adoption of higher-end GPUs, and that's what you're basically seeing.
Excellent, Colette. Thank you.
Thank you. Our next question comes to the line of Ross Seymore with Deutsche Bank Research. Your line is open. Please proceed.
Hi, Jensen. Given that gaming is slated to grow up over 20% year-on-year this year, how should we think about growth longer term? Colette, how can we think about the gross margin impact from the growth of gaming?
Yeah, that's a good question, Ross. Let's see. How do you think about it? I guess the way I think about it is it's related to the number of gamers. It's related to the production value of games, and so long as the number of gamers continue to grow, and the number of gamers have practically doubled in mobile in the last three years to almost 100 million gamers. Yet 100 million gamers is not that many gamers in the context of people who watch television. We know that just about every young adult is a gamer today. So it stands to reason that gaming is going to be a very, very large industry, and almost everybody is going to game. It's about $100 billion today.
It's hard to say exactly how big it's going to get, but it's hard for me to imagine that it would stay at $100 billion. So those things, hopefully, will continue to drive the growth of gaming and our dedication to the space. The investments that we've made over the course of nearly two decades in this space, has really put us in this opportunity to benefit from this growth. Okay, go ahead.
Further on your question of gross margins for gaming, we reviewed our gross margins by our market platforms on our Analyst Day, really kind of highlighting the success of our strategy in these platforms, and so many of them have higher gross margins than our company average. Yes, we can benefit from gaming, which does have a slightly higher gross margin company average right now. Of course, there's always mix factors that come into play each quarter, but it does help contribute to our great success that we've been showing on our gross margin over the last couple of quarters.
Thank you, and sorry. Yes, this is G for Ross Seymore. As for the follow-up question, you discussed the lumpiness of spending from customers in deep learning. We've seen CapEx cuts from the likes of cloud customers. What are you hearing from cloud customers in terms of slowing of their data center and cloud build-outs for the year, or are they cutting CapEx in other areas besides data center and cloud?
I really appreciate that question. In fact, we see the most success when there's pressure on CapEx. It seems counterintuitive, but the reason for that is this. The consumption of flops and the need for fast searches, whether it's voice search, natural language processing and translation of voice, image search, transcoding of images. When you send video up in one resolution, they want to maybe store it or transcode it or stream it at a different resolution. The ability to take poor video and enhance it, herky-jerky video, and all of a sudden you stabilize the video and make it beautiful. All of that video processing and all of that search is going to require the data center to be much, much more powerful.
The benefit of GPUs is that by adding GPUs, you don't have to incur a great deal of expense, but you can increase the performance of your data center tremendously. Let me give you the case in point. We increased the performance of the Oak Ridge supercomputer by a factor of 10 with just 18,000 GPUs. It was a retrofit. The most powerful supercomputer in our nation today, it's a retrofit. It's a retrofit of an existing supercomputer. If you could imagine retrofitting data centers with GPUs, it would increase the throughput, the computational throughput, arguably by a factor of 10, just like we've done for the Oak Ridge supercomputer. As a result, the CapEx doesn't have to go up very much. That's one of the benefits of acceleration.
It takes advantage of your current infrastructure, and you can improve its capability by upgrading it with a GPU.
Thank you.
Thank you.
Thank you. Our next question comes to the line of David Wong with Wells Fargo Securities. Your line is open. Please proceed.
Thanks very much. What % of desktops sold currently do you consider to be gaming desktops? What would be the answer for notebooks?
David, I'm not sure. One more time, what is-
Well, you're talking about-
In dollars, the gentleman earlier had already decoded our OEM sales. Our PC OEM sales represents only a few percentage of our total revenues now.
That's the way to think about it. Gaming, of course-
No
is a very large percentage of our revenues.
No, I'm not talking about your revenues. I'm talking about the overall market, if one considers the market of desktops. What percentage of units of desktops do you consider gaming that are taking up your gaming GPUs?
I see. Very good question. Incredibly small. Surprising. The reason for that is this: most gamers buys a PC and then upgrades it maybe three, four cycles. Does that make any sense to you?
Yes.
They buy our GeForce cards directly from e-tail. They take it home, they pull it out of the box, they install it into their computer, and sometimes they cut a hole in it so that they could see the graphics card. Sometimes they paint the chassis with car paint, so that they could personalize it, fall in love with it, and they upgrade it for years to come.
Okay, great. On a second matter, Tesla, I think last quarter you kindly provided us the revenue, I think it was $79 million. What was it this quarter?
Hold on in our thingy. It is $62 this quarter.
Great. Thanks very much.
You're welcome.
You're welcome, David.
Thank you. Our next question comes to the line of Alex Gauna with JMP Securities. Your line is open. Please proceed.
Thanks for getting me in. Congratulations on the strong quarter. Jensen, I'm wondering, you talked about 50 incremental designs in a driver-assist, as well as maybe 30 million units. Does that 50 customers totally encompass that 30 million, or is there more on top of that to come? Thanks.
Yeah. Thanks for that question, Alex. Most of the 50 new customers, the reason for that is because the number of companies that are working on autonomous driving cars are not just necessarily car companies. Car companies are going to invest in more and more autonomous capabilities. However, there are many companies who are creating autonomous cars that could be used as a service. You're gonna see the car industry bifurcate in some really exciting and interesting ways. On the one hand, car companies are gonna make cars more capable and safer and more joyful to drive. On the other hand, you're gonna see new startup companies, and maybe exciting startup companies, that create fleets of autonomous vehicles that are operated as a service. They come at the world very differently. Autonomous vehicles come in a lot of shapes and forms.
They're not all passenger cars. This is an area of a lot of very exciting development, and we're working with a lot of them because we have a platform that was really designed to fuse computer vision cameras from all around the car, as well as radars and LIDARs and sonars, and be able to do path planning and all of those kind of capabilities that you're gonna need for an autonomous vehicle.
Okay, great. Congratulations again. Great quarter.
Yeah. Thanks a lot, Alex.
Thank you. Our next question comes to the line of Sanjay Jha, calling from Nomura Research. Your line is open. Please proceed.
Hey, Jensen. One question on the gaming growth. If I look at combined discrete GPU market between NVIDIA and AMD, if my estimates are right, AMD lost roughly $300 million in discrete GPU market from 2014 to 2015, and you are gaining $300 million roughly. If I look at the whole discrete gaming market between both of you, it would appear it's stable. How do I reconcile this and the commentary that the gaming is growing for you? Is it growing between both of you, or is it growing for you? I just wanted to understand how would you reconcile this.
Well, you know AMD's business a lot better than I do, Blayne Curtis. I can't imagine the math working out. The GPU business is larger than it has ever been. I can't imagine that being a share gain. The applications that we're in are applications that have never been conceived of before, and so I can't imagine it being something that is a share gain matter. Whether it's in deep learning, whether it's the growth of MOBAs, the number of gamers around the world that are engaging in esports and MOBA is surely an increase. The gaming market in Southeast Asia is growing. The gaming market in Latin America is growing. The gaming market in China continues to grow, even through a slower economy.
People are just getting born, and they grow up, and one day they discover that their other friends game, and they get into gaming. I think the number of gamers of the next generation is more than the gamers of the previous generation. I think all of that is very logical and is kind of sensible. That's kind of how we see it. We're really not correlated at all to anything that AMD's doing anymore. Not that I can tell.
Okay. As a follow-up question on GRID. You are at trial at several thousands of customers, and just wondering what feedback, what issues, concerns that you have seen that may be taking longer for it to ramp?
I appreciate that. Thanks for that opportunity to answer the question. Enterprise trials surely takes a long time, but the benefit of an enterprise business is that it remains sticky for a very long time. This time last year, we've deployed 100 customers, and today we've deployed 300 customers. Our partnership with VMware is on full steam. You probably heard from their conference call and their commentary, how enthusiastic they are about end user computing. Since Horizon supported vGPU, in the last couple of quarters, the engagement across the board is just incredibly high, and the pipeline is very healthy. I think this is an area in enterprise computing that's just going to take time.
This is a long-term bet, but there's no question in my mind that the value of virtualizing an enterprise end-to-end, because of heterogeneous computing, because of security, because data is getting so large, it is better to keep it in the data center, and because people's work styles are more mobile and more flexible than ever. These dynamics are not going to go away. Virtualization of the PC applications that are not going to go away either, that are relied upon by enterprises all over the world, virtualizing those applications is really the right way to go forward and modernizing your company. I have a lot of enthusiasm and faith in this product line, and it appears that the rest of the IT industry is equally excited about it.
Great. Thanks so much, Jensen.
Yep, thank you.
Thank you. Our next question comes to the line of Craig Ellis with B. Riley & Co. Your line is open. Please proceed.
Thanks for taking the question, and congratulations.
Thanks, bud.
Jensen, just starting off with a higher level question that cycles back to Analyst Day. At Analyst Day, you highlighted four platforms with secular growth dynamics to them. With some of the softness that we've recently seen in the macro, does that have any impact on the growth that you would expect intermediate term from those four platform groups, gaming, auto, enterprise, and HPC and cloud?
The interesting thing is wherever an enterprise gets involved, we're surely affected by slowdown. When our products are bought and sought out one at a time, we tend to experience what the large industry would experience. In the case of gaming, it's a consumer product, it's a consumer market, people buy the product not because they can afford it necessarily. They buy the product because they want to play the game. When a new game comes out and it's too good not to play, something like "Star Wars" or something like "Call of Duty" or something like "League of Legends," it's just too good not to play. People will find a way to afford one of the GeForce graphics cards.
In the case of cars, the car industry is moving forward in autonomous vehicles, cars are becoming more and more computerized, irrespective of the economy, the high-end segments of the car market seems to be doing quite well. Maybe oil prices have something to do with that as well, we're seeing just a really exciting adoption in the higher-end segments of the market where a new technology is really transforming the car. With respect to deep learning, I think the jury is now pretty clear that deep learning has an opportunity to make a real difference in computer science, using a large amount of data to write the software itself is a pretty astonishing development.
Our GPU, making it practical to train networks that are trained and taught by just an enormous amount of data, has really opened up this field for rapid innovation. We're seeing a lot of excitement there. These platforms that we talked about, I think are exciting and making contributions on fundamental levels, I'm hopeful that we'll continue to grow through the rest of the year.
Thanks. The follow-up is for Colette. Colette, on the baseband wind down, can you just provide some more detail with regards to what you'd expect with the cost for that coming out of the system? I understand that you said you're going to reinvest that, there's legal expenses this year, can you give us a few more timelines related to that exit?
We had started a good portion of the wind down of the business in Q2. We are still working on shutting down that piece. We have a couple pieces, which is reflective of our restructuring charges that we plan for Q3 associated with it. We're just nearly done, but we still have a couple open issues that we're working through on that. As related to what we are going to invest in, as we've talked about on the call, we talked about the deep learning, really thinking about the automotive business, and of course, in terms of gaming, and those investments to assure where we are well-positioned for going forward. However, our overall plan, excluding the overall litigation expenses, is to still be flat OpEx year-over-year with fiscal year 2016 compared to fiscal year 2015.
That's to give you a little bit of color in terms of what we're planning on there. If you have anything further, just let us know.
Just to follow up there, will restructuring charges conclude in the fiscal third quarter, or will they carry into the fourth quarter?
Again, it's a little hard for us to determine that. We commented on them mainly as the second half, as we are still working through with those employees on their exit. I can't conclude that they would be in Q3, and they may move into Q4.
Thank you.
Thank you. Our next question comes to the line of Christopher Rowland with FBR Capital Markets & Co. Your line is open. Please proceed.
Hey, guys. It looks like standalone GPU purchases and pure gaming products, they're becoming a larger percentage of sales and mix, particularly as discrete GPUs kind of mix down here. Should we be looking at Q4 differently now? Does the holiday season have more impact now than it's had in prior years? Could this be a year where we get better than typical seasonality?
Well, I think we ought to just enjoy a seasonality first, and then we'll come back and comment on it. What really drives our business are the games that come out, the quality of the games, the production value of the games, and the global growth of gaming. United States is not very much a Well, let me just say opposite. United States is very much a game console market, and the reason for that is very clear. We have the benefit of family rooms, and people have their own bedrooms. Around the world, especially markets that are just developing and growing, the PC platform is a platform that every family needs and every individual needs anyways. They tend to be a PC gaming market.
In addition to that, because of esports and MOBA, which is very clearly the most exciting area of gaming today, that is largely a PC phenomenon. Those are really the factors that drive our growth. This second half of the year, we're really fortunate to have some amazing titles coming out, whether it's "Star Wars" or "Call of Duty," "Assassin's Creed," "Metal Gear Solid." These are franchise titles with just an enormous number of players that are waiting for it. I'm looking for it as well, and hopefully it'll drive for a good season.
Great. In autonomous driving, your main competitor, some consider a monopoly in the space, they're only spending maybe $40 million a year in R&D. You spend probably 30 times that in your total company, and I'm pretty sure you'd feel comfortable spending multiples of that just on the autonomous driving segment. Do you guys see this as an opportunity for you guys, kind of lower hanging fruit here? How should we view spending, how you're thinking about this? Just a smaller one, how do we sort of split your R&D between software and hardware spend in the area?
Well, we don't spend all that on autonomous driving, of course. We're a visual computing company with four platforms, and one of our platforms is autonomous driving or Drive, which is our car platform. Our approach to the current market for what is called ADAS driver assistance relies on a smart camera, if you will. It's a camera that can detect objects, and objects of a certain list. In the future, autonomous driving is surely more than detecting objects, and it will surely require more than cameras. You can't drive in the dark. You can't see in the dark if you can't see. Having cameras alone is not good enough. For cars using LIDARs and using radars and sonars and cameras all over the car, that is what the industry calls sensor fusion, to realize insight by fusing the input of multiple sensors.
In our case, we're going to use deep learning, the same approach that we've been using for much of the work that we do, that have been talked about in deep learning, to assist and help realize this vision of an autonomous vehicle. Along the way, of course, we're going to have all kinds of capabilities. The benefit of our approach is it's very compatible with smart cameras. Most of the platforms that we're in today, in fact, are harmoniously working with smart cameras that are either home brew or bought from a third party, and also processing on our platform. We see it as a compatible path to the future. The road to autonomous driving is there's lots of innovation yet. Your last question was, Chris?
Yeah, just in terms of your spend on that part, call it ADAS and autonomous, what's software and what's hardware? How do you sort of split that out or think about it?
Well, just the company largely is a software company, the reason for that is because we're about algorithms. If you look at the company, although we're well known for building some of the most complex processors in the world, that represents probably about a third of our engineers. Two-thirds of our engineers are algorithm experts and computational mathematicians and system software experts and so on and so forth. That's pretty consistent in ADAS as well. We're about two-thirds software.
Okay, great. Thanks.
Thank you. Our next question comes to the line of Joseph Moore with Morgan Stanley. Your line is open. Please proceed.
Great. Thank you. I wonder if you could talk about DirectX 12, what kind of impact you think that might have. I know these things don't come along that often, does that have a strong impact in any one short period, or is that sort of more drawn out over time?
Boy, I cannot wait for DirectX 12. The reason for that is DX12 has been rearchitected, particularly in the geometry primitive part of the pipeline. It can now process geometry with much, much lower load on the CPU than ever before. The benefit is that we're going to be able to pump through a lot more geometric fidelity, a lot more geometry processing into the GPU. You're going to see just much, much more beautiful graphics. On the other hand, it takes a lot of pressure off the CPU, and the CPU can now be used for other type of tasks, network more players, and deal with scenes that has a lot more characters, maybe more artificial intelligence algorithms, and making your opponent, if you will, the computer opponent, smarter. You're going to be able to do a lot more things with DX12.
I'm super excited about it.
Great. Thank you. Sort of a question, you guys have talked in the past a little bit about using FPGAs to accelerate servers versus graphics. Given the importance that Intel's placing now on FPGA server acceleration, how do you see the value proposition of Tesla in that same space?
Well, I think that first of all, it's, I guess at this point, a foregone conclusion that accelerated computing is an essential part of future scaling in what is called the post-Moore's Law era. Accelerated computing can come in a lot of different forms. An FPGA is a configurable computer. If you don't have that many different types of algorithms, and you can afford the expense and the difficulty of designing an FPGA, you can get results. It's basically designing your own custom chip that runs slower because ASICs are faster, but still, you're designing a custom chip. On the other hand, a GPU is not about designing, it's about programming.
If you have a large different types of algorithms, you're changing your algorithms on a regular basis, the throughput that you get with a GPU that we've built, which is arguably already the most complex that any processor has ever been built anyhow. It gives you a lot more flexibility. These are all spectrums, if you will, of accelerated computing. The thing that is really fantastic is finally on a very large scale and across the industry, the recognition that accelerated computing is the path forward.
Great. Thank you very much.
Thank you. Our next question comes to the line of C.J. Muse with Evercore. Your line is open, please proceed.
Good afternoon. Thank you for taking my question. I guess first question, Jensen, when you look at PC gaming and strength and sustainability into the back half, it looks like you came in about $150 million more than what you were thinking, at least initially for the guidance. I'm curious if there's any sort of color you can provide there in terms of mix, higher ASP strength geographically. Looking to the second half of the year, first half fiscal 2016 is growing about 40% year-over-year. How should we think about the second half versus the first half, particularly given those four drivers that you outlined at the early part of the call?
Let me answer the second one first, and maybe Colette can help me with the first one. On the second one, the thing that I feel much more confident in answering is our position in the market and the health of that market. I am pretty sure that gaming is a large-scale phenomenon and growing. I'm pretty sure that MOBA is a social network, that when you're playing games, you're hanging out with your friends, and when some of your friends are hanging out on MOBA, you've got to join it, otherwise you don't get to hang out. It's the reason why these large markets get larger. The benefit of a very large market is that finally, developers, publishers can invest several hundred million dollars in building a new game of extraordinary production value and launch it into a very large market.
The benefit of a large production value game to us is that it needs really great GPUs. Those positive feedback systems, those positive feedback effects, are things that I understand very well. How's the seasonal buying in the second half and exactly how many % and how many we're going to sell, it's kind of harder for me to tell. I do know that our gaming business is quite robust. On the other extreme, I know what I know about the autonomous vehicles and the processing that is necessary to achieve and realize that dream, and where we are on that dream. We're still quite a ways from it, in fact, to do something that we can really count on.
On accelerated computing, I'm pretty confident at this point that accelerated computing has reached a tipping point, and that it has been recognized around the industry and globally that this is a very good approach forward, both for supercomputing, already adopted by our nation's fastest supercomputer, but in also the recognition by the presidential initiative to adopt methods to bring exascale to the marketplace in the next several years. Accelerated computing will surely be required. I think that those things I feel very strongly about, and I think it's pretty clear our position within each one of those are very good.
To help answer your question regarding our performance in the quarter on gaming and how well it did and where did it come from in terms of, is that ASP, is that volume, is it region? The answer to that is yes, and yes. The value that we delivered in our high GTX cards were truly well received by the market. That high value affords us a great ASP and a great revenue growth. We've seen tremendous volume over this quarter as well, which has helped us. Thirdly, which I think is important, is all regions from a sell-through are also performing very well.
Very helpful. Thank you.
I think with that we have one more question for the group. Operator, if we can take that last question.
Certainly. The last question comes to the line of Matthew Ramsay with Canaccord Genuity. Your line is open, please proceed.
Thank you very much, and congrats on the quarter. Thanks for squeezing me in. Jensen, as you may appreciate, there's been a little bit of angst in the investor community regarding the Chinese consumer, given all that's gone on over there in the stock market, et cetera. It seems like your gaming business has obviously done quite well, and maybe you could talk a little bit about the concentration of your gaming business that maybe ends up in China and how the trends are playing out in that market overall. It seems you guys are doing much better than folks might have feared. Thanks.
Yeah, I appreciate the question. About 40%, I guess just shy of 40% of our total gaming business is in China. The Chinese market is surprisingly resilient because most of the games are free to play. It is probably one of the most affordable form of entertainment that you can possibly imagine. As I mentioned earlier, these networked games are really social environments. They're social networks. One of the reasons why Tencent grew so fast in China. Once the social network gets traction, it grows faster than linear. There are some reasons why the Chinese market is more resilient, gaming market is more resilient than expected. We're also seeing a shift to higher-end GPUs in China. Some of that probably has to do with the dramatic change and improvement in production value of games in China.
There was a time when the Chinese games were enjoyed and fun, however, the production value weren't very good. Now if you take a look at the Tencent games, the production value are just absolutely phenomenal. They're beautiful, they're artistic, and in those cases, require a lot more GPU capability. We're seeing multiple dynamics happening in China, and they all seem to be quite healthy.
All right. Thank you very much.
Thank you. I will now turn it back to the speakers for their closing remarks.
Thank you, everybody, and I'll talk to you next week, next quarter.
Okay. Thank you.
Thank you, gentlemen. That does conclude the conference call for today. We do thank you for your participation, and I ask that you please disconnect your lines.