NVIDIA Corp (NVDA) — Q4 2015 Earnings Call Transcript

Thank you. Good afternoon, everyone, and welcome to NVIDIA's Conference Call for the Fourth Quarter of Fiscal 2015. With us on the call today from NVIDIA is Jen-Hsun Huang, President and Chief Executive Officer. I'd also like to introduce Arnab Chanda, who joined us last week, as the Head of Investor Relations. Arnab and Raj will closely work to support the investment community. I'd like to remind you, 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 February 18, 2015. The webcast will be available for replay up until the next quarter's conference call to discuss Q1 financial results. The content of today's call is NVIDIA's property. It cannot be reproduced or transcribed without our written prior consent. During the course of this call, we may make forward-looking statements based on our current expectations. These forward-looking statements 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 Form 10-Q and the reports that we may file on Form 8-K with the Securities and Exchange Commission. All of our statements are based as of today, February 11, 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 may 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's start. We achieved record revenue for the fourth quarter and the full year, with impressive growth in each of our markets' specialized platforms. NVIDIA's strategy of creating an ecosystem in gaming, enterprise graphics, accelerated computing, and automotive continues to make excellent headway. Our gaming platforms grew more than 30% year-on-year and accelerated in the fourth quarter, driven by our 10th-generation Maxwell processors and the vibrant gaming market. Launched in September, our GeForce GTX 980 and 970 GPUs continue to be enthusiastically received. The Maxwell architecture provides new capabilities such as real-time, dynamic, global illumination, which simulates the true interaction of light with the environment and is the most energy-efficient GPU architecture ever created. We added to our lineup the GTX 960, which brings Maxwell to $199, what we see as the gaming market sweet spot. Powered by 1,024 CUDA Cores, it plays even some of the most demanding titles at 60 frames per second on full HD displays and it leaves plenty of headroom for over-clockers. The GTX 960 will let gamers enjoy the latest games at quality and performance greater than even the best gaming consoles at half the price. There are now more than 50 million PCs with GeForce Experience, our application for the gaming ecosystem that optimizes user settings for each title and enables players to record and share their victories. In the month of November alone, GeForce Experience users downloaded our drivers over 30 million times. GeForce Experience brings the simplicity and community benefits of the console to the PC. NVIDIA powered the holiday season's 2 best-reviewed Android tablets, the Google Nexus 9 and our own SHIELD Tablet, which are based on our Tegra K1 processor. More than a dozen media outlets recommended SHIELD in their holiday gift-giving guides. Over the course of the quarter, SHIELD, which has pioneered a premium gaming experience for tablets, became one of the very first devices to be upgraded to Google's latest OS, Android 5.0 Lollipop. The NVIDIA GRID graphics virtualization platform continues to gain momentum. More than 300 companies worldwide tested the product in recent months and experienced the benefits of GPU-accelerated virtualization. They range from large institutions such as the University of Southern California, leading manufacturers such as Airbus and Lockheed Martin and defense contractors like Raytheon. Enterprise virtualization leaders continue to deploy GRID. Just last week, VMware rolled out the latest version of its vSphere virtualization software, including full support for our GRID vGPU graphics virtualization technology. Quadro professional graphics continue to maintain market leadership. Refreshed in Q3, the lineup is now being shipped in all major OEMs. For the sixth year in a row, every film nominated for an Academy Award for special effects was made using Quadro. Our accelerated computing platform performed very strongly. Tesla's revenue growth from a year ago exceeded 50%, as HPC customers and cloud service providers continue to deploy large GPU-powered systems. These provide deep learning capabilities that help enable our amazing capabilities, like voice recognition and natural language processing. Using Tesla, our customers have seen dramatic benefits in application performance, radically reducing run times. During the quarter, we introduced the Tesla K80 Dual-GPU, our new flagship Tesla offering. It provides nearly twice the performance and double the memory bandwidth of its predecessor, the Tesla K40. A single server with Tesla K80 delivers the same performance as 8 servers with our old Fermi generation GPU accelerators for a range of applications. A powerful endorsement of Tesla came just ahead of the recent Supercomputing 2014 Show. The U.S. Department of Energy announced that its next generation of supercomputers will utilize Tesla GPU accelerators in conjunction with our powerful new NVIDIA NVLink high-speed interconnect technology. These systems will be deployed at Oak Ridge and Lawrence Livermore National Laboratory. We'll be at least 3 times faster than today's most powerful systems. These supercomputers will serve scientists across our nation in every field of science to accelerate their research for the next 20 years. Finally, NVIDIA's automotive platforms remain on a sharp upward trajectory, registering better than 80% growth on the year. More than 7.5 million cars with our technology are now on the road, up from 4.7 million a year ago. At CES, we announced NVIDIA Drive, a computing platform for next-generation advanced driver assistance systems and digital cockpits. NVIDIA Drive makes driving safer and more enjoyable by introducing surround vision and auto valet capabilities. Its cockpit computing platform includes 3D navigation systems, natural speech and image processing capabilities, dramatically transforming the automotive user interface. NVIDIA Drive is the first car computing platform, utilizing advances in deep learning. Drive will leverage our dramatic advances in deep learning for voice, language, and image recognition to enable surrounding and situational awareness in cars. We believe deep learning is the key enabling technology to continue advancements in autonomous driving capabilities. At the heart of NVIDIA Drive is the recent announcement of Tegra X1, a 256-core super chip with over 1 teraflop of floating-point processing power built on the same Maxwell architecture. Rolled out only months earlier for high-level gaming, Tegra X1 provides the computational horsepower and graphic capabilities for cars that can see and learn. Now let's take a more detailed look at the financials. Fourth quarter revenue increased 9% year-over-year to a record $1.25 billion. Growth was driven by strength in gaming, data center, cloud, and auto. Fourth quarter revenue was up 2% sequentially. The GPU business grew 8%, helped by Maxwell GPUs and the seasonal increase in consumer PCs. Overall, Tegra processor sales decreased 33%, as growth in our automotive and SHIELD devices was more than offset by lower revenue from smartphones and tablets. Fiscal 2015 revenue grew 13% to a record $4.68 billion. In our GPU business, revenue grew 13% in Q4 from a year earlier. Revenue from our PC gaming platforms rose 38% due to the popularity of our high-end Maxwell-based GTX GPUs. Tesla GPUs for accelerated computing increased strongly, driven by large project wins with cloud service providers. Tegra processors declined 15% in Q4 from a year ago, as smartphone and tablet designs reached the end of their product life cycle. This was partially offset by the increased revenue from auto infotainment systems, which more than doubled, and from SHIELD devices. Moving to gross margins, GAAP gross margin was 55.9%, above our outlook for the quarter and up 70 basis points from last quarter. Non-GAAP gross margin was 56.2%, also above our outlook and was up 70 basis points sequentially. Strength in margins from gaming and accelerated computing was partially offset by those from Tegra processors and certain inventory provisions for prior Tegra architectures. For fiscal 2015, GAAP gross margin was 55.5%, and non-GAAP gross margin was 55.8%. These full year record levels were led by the strength of our high-end GPUs for gaming and accelerated computing. GAAP operating expenses for the fourth quarter were $468 million. Non-GAAP operating expenses were $420 million, in line with our outlook and inclusive of legal fees associated with our litigation against Samsung and Qualcomm. For fiscal 2015, GAAP operating expenses were $1.84 billion, and non-GAAP operating expenses were $1.66 billion. The low single-digit growth from the previous year represents discipline from our continued management of our investments in both R&D and capital expenditures to enhance return on invested capital. The U.S. Federal R&D tax credit was reinstated for calendar 2014. For the full year, the GAAP effective tax rate was 16.5%; non-GAAP was 16.9%, incorporating the U.S. Federal R&D tax credit. For the fourth quarter, GAAP net income was $193 million, up 31%, driven by increased operating profit from strong revenue and margins. GAAP earnings per diluted share of $0.35 increased 40% from a year ago quarter, reflecting net income growth and share repurchases. For the full year, GAAP net income was $631 million, up 43%, and GAAP earnings per diluted share were up 51% from a year earlier. Now turning to some key balance sheet items. During the full fiscal year of 2015, we paid $186 million in cash dividends and repurchased 44.4 million shares for $814 million. As a result, we returned to shareholders $1 billion during fiscal 2015. Since restarting our capital return program in the fourth quarter of fiscal 2013, we have returned approximately $2.22 billion to shareholders. This represents 111% of our cumulative free cash flow for fiscal year 2013 through '15, reflecting the acceleration of our capital return program from cash generated in prior years. In November of 2014, we announced our ongoing commitment to deliver shareholder value through capital return, with the intention to return approximately $600 million to shareholders through ongoing quarterly cash dividends and share repurchases in fiscal 2016. Accounts receivable at the end of the quarter were $474 million, down from $563 million in the prior quarter. Inventory at the end of the quarter was $483 million, up from $408 million in the prior quarter. This sequential increase included the ramping of our new Maxwell-based GPUs and Tegra-based SOCs and SHIELD devices. Cash flow from operating activities was $443 million, up from the $216 million in the prior quarter, reflecting higher net income and the annual payment from Intel. Free cash flow was $412 million in the fourth quarter. Depreciation and amortization expense amounted to $54 million. Capital expenditures were $31 million. Now turning to the outlook for the first quarter of fiscal 2016. We expect revenue for the first quarter of 2016 to be $1.16 billion, plus or minus 2%. Growth year-over-year is expected from our key platforms: gaming, data center and cloud, and automotive. Our GAAP and non-GAAP gross margins are expected to be 56.2% and 56.5%, respectively, plus or minus 50 basis points. The gross margin outlook is slightly higher than our Q4 performance, driven by our gaming and accelerated computing platforms. GAAP operating expenses are expected to be approximately $478 million, with non-GAAP operating expenses of approximately $425 million, inclusive of legal fees associated with our litigation against Samsung and Qualcomm. Excluding these legal fees, we expect to maintain our operating expense levels near Q4 levels. GAAP and non-GAAP tax rates for the first quarter of fiscal 2016 are expected to be 20%, plus or minus 1%, excluding the benefit of the U.S. Federal R&D tax credit, which expired December of 2014. In summary, we are pleased with our record revenue for the quarter and the full year, which reflects strength across our platforms and businesses. Profitability as measured by earnings per diluted share increased 51% in fiscal 2015, as our revenue and gross margin growth continued to outpace our investment. This concludes our remarks, and we will now turn it back to the operator for questions. Please be sure to limit your questions to one with one follow-up.

I appreciate the strong results and guidance. I guess the first question, Jen-Hsun, one for you on the Tegra side of the business. I know that can be inherently lumpy in how it operates, but can you talk about what you envision the biggest drivers in fiscal 2016 will be to get that business on a growth trajectory versus the year-over-year drop that it's currently running at?

Yes, Ross. First of all, thanks. Our strategy for Tegra is to focus on automotive, gaming. And wherever opportunities arise in OEMs, we surely will entertain it, but our fundamental focus is automotive and gaming. And you already heard Colette say that our automotive business nearly doubled year-over-year. Our gaming business is really strong, and we have some exciting things that we'd love to share with you in the near future. But those are our 2 focuses for Tegra. And if you look at the opportunities that we're pursuing in the automotive area, obviously, more and more cars are becoming computerized cars. At CES, we announced a brand new platform called Drive. And Drive is basically a mobile super chip, a mobile supercomputer, with a ton of software on top. It provides for 3 basic functionalities: The first is a very advanced digital cockpit. You know that we're incredibly good at computer graphics, and the things that we can do in the car with more displays is really wonderful. The second is a technology called surround vision. Beyond surround view, which is basically cameras connected around the car, surround vision not only allows you to see the images from the cameras around the car, but it also creates a virtual environment around it and keeps the car alert of its surroundings using computer vision technology. The third is using a technology called deep learning that has been revolutionized recently and it's something that we're very much at the center of. Using deep learning, we can do voice processing, natural language processing, and understanding of imagery so that the car can be more situational and environmentally aware. All of those technologies that I just described, we believe, will help pave the path to the self-driving car. In the area of mobile gaming, the gaming market is quite large and is growing rapidly internationally. It’s about a $70 billion business now. We know that mobile cloud is probably one of the most important disruptions in the history of computing, and yet there's really no computer gaming architecture that serves mobile cloud very well. And that's what SHIELD is about, to create the platform that allows mobile cloud to bring gaming to a lot more people. And we think that by doing so, we'll engage a much larger part of the overall gaming market than just PC gaming.

Great. I guess, as my follow-up, a bigger-picture question for the GPU side of the equation. You talked about the GeForce business being up 38% year-over-year, and GPUs, as an entirety, were up 13%. Is there any color you can provide just into the size of the buckets within your GPU segment so we can get a little more granular in how we model those very disparate growth rates, whether it's the 4 segments you break out in your segment definition and the CFO commentary? Any other colors you can give would be helpful.

We serve 4 markets: gaming, automotive, enterprise graphics, and high-performance computing and cloud. Those are the 4 major segments that we serve. GPUs serve gaming, enterprise graphics, and cloud. Gaming, as we've said earlier, has grown, as you mentioned, 38% year-over-year. Our enterprise graphics business, the workstation business, is relatively flat. However, our enterprise virtualization virtualized graphics platform called GRID doubled year-over-year. Our Tesla business, which is our accelerated computing business, also doubled year-over-year. The PC OEM business declined, representing about low teens maybe approximately part of our overall business. We've been talking about its decline for some time now. This year, it declined again. Obviously, the PC OEM business, representing a low percentage of our overall GPU business, has much lower profit margins because the gross margins at OEMs tend to be much lower. But nonetheless, it declined year-over-year.

This is Shankar on behalf of Vivek. I would like to follow up on the GPU breakdown. Can you provide details on how the individual segments are projected to perform based on the guidance given for Q1?

Sure. We guided seasonally down. And as usual, we'll see how it turns out. But if you look at the position of each of our platforms, our position in gaming is very strong. The gaming market globally is quite robust. We're still seeing gaming growth all around the world, and the production value of games is increasing all the time. Whenever the production value of a game increases, it requires more GPU performance. For the very first time in history, a game console has come out that uses essentially a PC architecture, and its performance is substantially higher than the average PC performance. And as a result, we see a lot of adoption of our higher-end GPUs, and that's good. We also expect that the Tesla platform continues to be quite robust. We know now that GPUs are wonderful processors for massively parallel applications like deep learning. Deep learning can be used for voice recognition, natural language processing, image recognition, video tagging, all kinds of applications made possible because of cloud computing and the abundance of sensors and images being uploaded into the cloud. That's one of the drivers for Tesla, and the Tesla business has doubled year-over-year from a substantial base now. I expect that Tesla will continue on that trajectory this coming year. GRID, another part of our GPU business, is all about enterprise virtualization. With GRID, companies are able to virtualize beyond the data center all the way to the end-user computing device. For the very first time, it is now possible to virtualize graphics-intensive applications and make it possible for people to enjoy their applications wherever they are. So now you can be virtualized, it can be more secure, and you can be more mobile, enjoying graphics on any device.

Great. I just have a follow-up on the autos business. I know you have talked about different products and features that you have been working on, and those seem to be the high-end side of auto. But can you talk about the products that you have for the mid and low end because it looks like that might be the area where the scale is? And also talk about how you plan to increase the gross margin for auto systems because it looks like it's in the low 30s if I just back out all the numbers based on your reported numbers.

Well, the answer to the second question I know better than the answer to the first, so I'll just answer that one first. The way to add to increase our gross margins is to continue to innovate and increase the value we bring. And if you recall a long time ago, NVIDIA's gross margins were about 30-something percent. And over time, we made the platform more software-rich. Today, our gross margins are surely much higher than 35%. My expectation is to do the same on the NVIDIA Drive platform. It's intensely software-rich. I think somebody said that today, a car is about 10% software, but soon, the car will be about 80% software, and I think that's right. I mean, in the future, with electric cars, you have obviously still sophisticated drivetrains, but the car is going to be largely a computer with software. That's where we can add significant value, and that's what Drive is all about. Adding value through a very powerful processor, but most of the capabilities I described—the digital cockpit, surround vision, and deep learning—are all about software. This is what we provide. On your question about high-end versus midrange, I believe more and more of the car's value will be delivered through software. If you have to deliver it through software, then the processor underneath it is increasingly important. Great computing platforms will extend well into the midrange and low-end part of the marketplace.

I think the first one, Jen-Hsun, you made some interesting comments earlier about the need for a fairly virtualized mobile gaming experience. I assume you're meaning across desktop tablets, smartphones, cloud, etc. One of the things that strikes me is, obviously, there's the suit going on between yourselves and Qualcomm, Samsung, etc., and your market share within the mobile graphics business when you think about smartphones is fairly small from a hardware perspective. So do you see that changing dramatically going forward? And if not, do you see the gaming mobile device market bifurcating from the traditional smartphone market going forward?

Yes, Matt, thanks a lot for that question. First of all, mobile is much more than phones. Mobile is a fundamentally new way of designing computers, and I believe that mobile will impact almost every segment of computing as we know it. It'll impact refrigerators. That's not a phone. It'll impact drones. That's not a phone. It'll impact earrings. That's not a phone. It'll impact watches. That's not a phone. It'll impact game consoles, which are not phones. It'll impact cars, and that's not a phone. I think mobile is going to be important in all kinds of computing devices. When I say mobile, that's what I mean. I don't mean mobile as in the mobile phone. Mobile technology is very important. I also believe that mobile cloud in combination is one of the most powerful computing forces that the computer industry has ever known. Because of mobile cloud, we've been able to extend the capabilities and benefits of computing to billions of people, whereas in the PC area, we could only benefit hundreds of millions. Yet no one has created a game platform around mobile cloud technology that allows us to extend gaming not to tens of millions of game console users, but billions of users. That’s the opportunity. I don’t have anything to announce today, but that's our endeavor. I appreciate you asking that question. I think it’s going to be a really big opportunity for us.

As a follow-up, this one is for Colette. Obviously, you guys have done a great job in returning cash to shareholders and announcing new plans for the next fiscal year. Maybe as the last fiscal year ends, maybe you could update us on your onshore and offshore cash balances, if you could, and any perspectives on the difference in cash flow onshore and offshore.

Thanks, Matt, for the question. Again, we are looking at a cash balance lower than where we ended at fiscal year '14, due to our $1 billion return to shareholders, but also the increase in cash flow. Nothing has materially changed in terms of our cash flow from international and our cash flow in the U.S. So again, the majority of our cash flow tends to arrive from international operations. We have about the same mix of overall U.S. cash versus international cash, a little lower than where we ended at the end of fiscal year '14. But we feel very confident with our overall capital return program for fiscal year 16 and the cash flow prospects, so I think we're well aligned there.

Can you give us some idea of what percentage of Tegra is now automotive? And just a clarification, when you say that automotive infotainment more than doubled year-over-year, do you mean all automotive? Is there some other segment than infotainment?

Most of our automotive today is infotainment. More people are starting to do what Audi had done with their really amazing virtual cockpit, where infotainment and digital clusters merged into one. Digital clusters are much harder than infotainment. With Tegra, we're able to easily merge the two. You'll see more examples of that in the future. But initially, it's largely infotainment and some amount of digital clusters. Between automotive and SHIELD, our gaming platform, that represents the vast majority of Tegra now.

Okay, great. And you also mentioned Tesla more than doubled from an already substantial base. Can you give us some idea of the size of Tesla today?

Tesla is several hundred million dollars annually now and grew—nearly doubled year-over-year.

Can you guys talk a little bit about your eventual move to 20 nanometers? It seems like you guys are staying on 28 for perhaps a bit longer than you have prior nodes. If you are staying here for longer, is it safe to assume that you can get some gross margin uplift year-over-year?

Well, let's see. We're always moving to the next node. Tegra X1, the mobile super chip that we announced at CES, is 20 nanometer, and we're always moving forward. But the equation is much more complicated than that, and the reason for that is because we have the ability, thanks to the fact that we architect and innovate from architecture all the way to software, to improve our performance, efficiency, and features on all those dimensions. We can improve at the architectural level using the same process. We can innovate at the design level using the same process and architecture. We can innovate at the process level. We can innovate at the software level. We can innovate at the algorithm level. By controlling all of that, we can move the needle forward in multiple dimensions. Maxwell is an incredibly energy-efficient processor. Thanks to that energy efficiency, we were able to take the GPU and put it into a Tegra X1 in just a few months, achieving amazing performance.

Okay, great. And then also at CES, I loved the automated driving stuff, but I was a little confused on how you're positioning yourself in that market, particularly against or partnering with Mobileye. I talked to one person who said you're going head-to-head. Another person said you're partnering. I talked to another person at the booth who said you're enabling third parties to go after Mobileye. How do you see yourself positioning in that market? Are you going after them?

Wow. I appreciate that question. The answer is actually really simple. We are making two different things, and we’re very complementary. Mobileye, think of it as a computer vision signal processor ISP connected to the camera, and they do wonderful work in computer vision. The Drive platform is a computing platform. Just as a phone has an ISP and the application processor, we’re the application processor. This application processor has been designed to excel in the three areas that I talked about: digital cockpit, surround vision, and deep learning-based computer vision. Many of the customers, in fact, most of the customers that I know working on ADAS projects are using both technologies. Even if you are able to detect images and objects, you still need to process that information, reconstruct the surroundings, drive infotainment in the digital cluster, and apply higher-level deep learning algorithms. So I see ourselves very complementary.

Jen-Hsun, just want to follow up in terms of the auto opportunity, just your dollar content progression. You're talking about selling modules, multiple processors and then more importantly, monetizing the software. And then if you could talk about some timing here, I think the auto OEMs have become a little quicker. Where are you today in terms of the generation of products you're selling, Tegra 3 or K1? And when could you see some of these more advanced systems actually hitting revenue?

Yes, I really appreciate the question, and you're exactly right. The dynamics are that more and more of the value of a car is going to its computing platform. Our computer car and the software capabilities we can bring is going to grow substantially. Second, the fact is, you can't spend 5 years designing a car anymore. Consumer electronics moves way too fast. Consumer expectations are set by what they have in their phones, so we must move faster. The most progressive car companies are moving at the speed of light now, taking their design cycle from 5 years to 3, and I see them moving it down to 2, potentially 1. This will revolutionize the automotive industry. The third point you've mentioned is the content. If you think of cars as a chip opportunity, it's an application processor opportunity like an infotainment system like a smartphone. It can be in the tens of dollars. I see the car as a multi-thousand dollar computer opportunity. Some cars have one computer, some have many. We're a visual computing company. We're software-rich, and all the applications I've mentioned are all software. We don't just provide the chip; we provide the entire solution. So, I anticipate considerable growth in our automotive opportunities.

And just to reflect quickly on gross margin. I think you mentioned gaming as the tailwind for gross margin. Is that more a function of just Maxwell and pricing, or are you actually seeing gaming not down as much seasonally? And then as you look out the rest of the year, how should we think of that tailwind?

Yes, thanks for the question. Our gross margin guidance is just for Q1. We do see seasonal effects as we usually go from Q4 to Q1. We believe gaming is still a driver due to the mix of type we're selling into the market. We've also talked about accelerated computing and cloud as drivers of our guidance in Q1.

Maxwell is driving significant expansion into the notebook market. I think you started to see that in Q3, with that segment up 2 times year-over-year. Customer gaming notebook platforms were pretty visible during the holiday season. So I apologize if I missed this, but how much did the notebook gaming business grow for you guys in Q4? And maybe, Jen-Hsun, could you help us quantify where the attach rates could go and how much this expands your GPU gaming opportunity?

Yes, I appreciate the question. Maxwell made it possible. First, Maxwell allowed high-end GPUs in laptops to deliver a desktop-quality experience. You can now get a laptop with a GeForce in it, run AAA titles at 1080p, and exceed the performance of state-of-the-art game consoles. That's pretty amazing. You can now have a thin laptop that's better than a game console. This business is currently growing over 100% per year due to its small base. Every laptop company is now entering this opportunity due to its significance. We have nearly every laptop manufacturer worldwide supporting this opportunity. Where do I see the future? I don't think of attach anymore, as the PC OEM business is not our focus anymore.

And then, Colette, nice job on your part and the team's part on OpEx discipline in Q4 and here in Q1. How should we think about the OpEx trajectory for the remainder of the year?

Yes, we really are here to focus on our Q1 guidance and what we can see at this time. When we get further into the year, we’ll talk about the full year. But again, we just want to balance our Q1 with the investments we need to make while looking at efficiencies across the organization. This will remain a consistent theme, as you've seen us play out in fiscal year '15, and our guidance for Q1 is about that same.

Jen-Hsun, I have a question regarding your GRID GPUs. I thought you were very enthusiastic about the upcoming VMware release, especially with G8. I'm curious if this provides you with better visibility for forthcoming deployments. Do you believe that all the challenges customers face while trialing GRID have been addressed, or are there still some gaps? I would like to hear your thoughts on this.

Well, with VMware's integration, we can now engage probably 80% of the world's enterprises. Just about anyone who has virtualized their enterprise did so with VMware's technology. This is a significant deal. Our trials went from about 400 last year to over 1,000 this year. Our revenues doubled. My sense is that there's no reason for this pace to slow. But the most important part of enterprise virtualization is to have deep integration with VMware, and we've done that, so I see it continuing to grow.

And as a follow-up, Jen-Hsun, you indicated some large project wins with cloud service providers on Tesla's side. Could you give more color on what exactly is the use case? You talked about deep learning and image processing. What's the ramp like? Was that a lumpy deployment, or is it early in the ramp and expected to grow?

Yes, many of these projects are confidential, but I'll highlight two that have been public. Baidu announced that they built a supercomputer based on Tesla for deep learning, used for natural language processing and real-time translations. A recent announcement, Microsoft reported that for the first time, a computer is able to understand images better than humans. This is a momentous event in neural nets and computer vision, announced yesterday. For the first time in history, a computer can recognize over 95% of the images presented, better than a human. We'll look back on this day as a milestone. Their deep neural nets were trained using NVIDIA GPUs. These are examples, and there will be more revelations at our upcoming GPU Technology Conference on March 17.

Could you discuss your transition to a platform company? Over the last few years, you have transformed from a component supplier to a platform-based business with higher margins. Where are we in that development cycle? How do you view the overall business model over the next 3 to 5 years for the company?

Yes, I appreciate the question. If you look at our company 10 years ago, we were almost entirely PC OEMs, and now we're less than 20%. The company has been transitioning to a new model with our own platforms that serve four vertical markets. These markets benefit greatly from visual computing technology. Those four markets are gaming, automotive, enterprise graphics, and high-performance computing and cloud computing. Our PC OEM business continues to decline because PCs have become good enough and are fragmented. Most people use PCs for information access, and a smaller group for gaming. So, in these four markets, our gaming business is growing at over 30%. Our automotive business has doubled, and our Tesla business has doubled as well. The high margins are a result of this shift from PC OEM to these higher margin sectors.

So very good. Lastly, with regards to GRID, could you update us on design engagements with GRID in 2014 compared to 2013? As you continue on this trajectory, how are enterprises beginning to realize the advantages of a cloud-based visual computing architecture? Do you think this year will be the inflection point or are we past that and gearing for acceleration?

Yes, let me break it down. The number of OEM platforms we were designed into last year was about 50. It's nearly 100 this year. Trials were about 400 last year and over 1,000 this year. I believe we will see a significant year from GRID this coming year. The benefits of GRID include mobility meaning you can access your PC applications from anywhere on any device. There’s improved security since digital downloads are no longer a concern. And collaboration improves because the computer can now reside in the data center, allowing teams to work on the same data set without downloading it locally.

Jen-Hsun, can I ask about the gaming business in terms of PCs in fiscal Q4? Can you talk a bit more about the geographic demand across developed and emerging markets?

Yes, Stephen. Our largest market is China. The GeForce is the largest game platform in China because it hasn't had access to game consoles. Prices are still extremely high for consoles. Most titles come from the West, and role-playing games tend to dominate. China remains our largest market, but the fastest-growing market is likely Southeast Asia. These regions become vibrant quickly in their computer and Internet history, similar to the developments we saw in China five years ago.

I was wondering, Jen-Hsun, if you could update us on what's going on in the licensing front. Perhaps comment on whether it is getting to the point where it might provide a tailwind to gross margins this year. Or if not this year, when?

Our licensing discussions are ongoing. We are unquestionably the world leader in digital computing. We've pioneered modern computer graphics from the invention of the GPU to general-purpose computing with GPUs like CUDA. We are open to licensing technology to companies wishing to build their differentiated products while protecting our IP. The most public discussion is with Samsung and Qualcomm, and I look forward to reporting on developments in that area over the next several quarters.

Could you talk about your foundry strategy a little bit? I mean, TSMC is a good supplier, but not considered the best FinFET out there. Would you consider diversification as something to evaluate?

First, TSMC is a fabulous supplier. Their FinFET technology is excellent, and we've been evaluating it. We take our test chip technology seriously, ensuring rigorous testing. We've collaborated with TSMC on FinFET for a couple of years, and we have confidence in their ability to deliver amazing FinFET transistors. We continually evaluate other foundries, but TSMC remains our most strategic partner.

Okay, operator, I think this is the end of our call. To all who joined us, I thank you, and we'll see you next quarter.