NVIDIA Corp (NVDA) — Q3 2019 Earnings Call Transcript

Thank you. Good afternoon, everyone, and welcome to NVIDIA's conference call for the third quarter of fiscal 2019. With me on the call today from NVIDIA are Jen-Hsun Huang, President and Chief Executive Officer; and Colette Kress, Executive Vice President and Chief Financial Officer. I'd like to remind you that our call is being webcast live on NVIDIA's Investor Relations website. It's also being recorded. You can hear a replay by telephone until November 22, 2018. The webcast will be available for replay until the conference call to discuss our financial results for the fourth quarter of fiscal 2019. The content of today's call is NVIDIA's property. It can't be reproduced or transcribed without our prior written consent. During this call, we may make forward-looking statements based on current expectations. These are subject to a number of significant risks and uncertainties, and our actual results may differ materially. For a discussion of factors that could affect our future financial results and business, please refer to the disclosure in today's earnings release, our most recent Forms 10-K and 10-Q and the reports that we may file on Form 8-K with the Securities and Exchange Commission. All our statements are made as of today, November 15, 2018, 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, I'd like to turn the call over to Colette.

Thank you, Simona. In the third quarter, revenue was $3.18 billion, an increase of 21% compared to last year, with all four of our market platforms showing double-digit growth. Data center, professional visualization, and automotive segments all achieved record results. However, gaming fell short of expectations as the post-crypto channel inventory took longer to sell through. Gaming card prices, which had remained high after the significant decline in cryptocurrency mining, have not normalized as quickly as anticipated. Our forecast for gaming in the fourth quarter anticipates very few shipments in the midrange Pascal segment to allow for normalization of channel inventory. We also expect minimal sales of Tegra chips for game consoles this quarter due to the typical seasonal build cycle. While the current channel inventory situation poses a short-term challenge, it does not alter our long-term outlook. Our competitive position remains robust, and we are expanding our potential market with Turing and our recent software updates. We are enthusiastic about growth possibilities in ray-traced gaming, rendering, high-performance computing, AI, and autonomous vehicles. GAAP gross margins increased by 90 basis points year-over-year, while non-GAAP gross margins rose by 130 basis points. This growth reflects our ongoing transition to higher-value platforms and includes a $57 million charge for previous architecture components and chips following the sharp drop in crypto mining demand. Both GAAP and non-GAAP net income exceeded $1 million for the fourth consecutive quarter. Analyzing our reporting segments, GPU revenue increased by 25% from last year, reaching $2.77 billion, while Tegra processor revenue decreased by 3% to $407 million. Focusing on our gaming sector, revenue of $1.76 billion was up 13% year-over-year but down 2% sequentially. The year-on-year growth was fueled by initial sales of our new Turing-based GPUs and strong notebook sales, which offset declines in gaming consoles. In mid-September, we launched the GeForce RTX series, the first gaming GPUs featuring our Turing architecture. Turing RTX technology offers up to twice the performance of its predecessor, Pascal, and six times more for ray-traced graphics, marking the largest generational leaps we've made in gaming GPUs. The initial GeForce RTX gaming cards, the 2080 Ti and 2080, provide 4K HDR gaming and 60 frames per second on the most advanced AAA titles, a significant achievement for gamers. This is rapidly becoming the new standard as 4K displays become more affordable. Following these high-end cards, we also released the GeForce 27 D. NVIDIA RTX technology brings unprecedented realism to games. The much-anticipated Battlefield V, which launched this week, is the first game to incorporate RTX ray-tracing, providing lifelike reflections on GeForce RTX GPUs. With a lineup of upcoming games that support NVIDIA RTX features, RTX is on track to become a revolutionary architecture. Even though the cryptocurrency surge has ended, the normalization of the channel is taking longer than expected. Pascal high-end cards have largely sold out before the arrival of RTX, while midrange Pascal gaming cards still see higher-than-expected channel prices and inventory levels. Pascal is well-positioned as the go-to GPU in the midrange for the holiday season, and we anticipate reducing channel inventories over the next quarter or two. In the data center segment, we achieved another impressive quarter with revenue of $792 million, a year-over-year increase of 58% and a sequential increase of 4%. Demand for Volta architecture products, including Tesla V100 and VGX systems, remains strong, and our inference business has continued to grow, benefiting from the recent launch of the Turing T4 Cloud GPU. Just two months post-launch, the T4 is seeing the fastest adoption rate of any server GPU, integrated into 57 server designs, and is also available on the Google Cloud Platform. The T4 delivers record performance for deep learning inference and enhances a variety of cloud workloads, including high-performance computing, deep learning training and inference, machine learning, data analytics, and graphics. We also unveiled an updated TensorRT software stack and NVIDIA TensorRT Hyperscale Platform, which introduces two critical capabilities to enhance the deployment of the NVIDIA inference platform in hyperscale data centers. The first allows multiple models and frameworks to operate simultaneously on the same GPU, boosting data center utilization and directly leading to significant cost savings. The second integrates with Kubernetes, the leading orchestration layer for hyperscale data centers. Completing our inference platform, the new T4 GPU provides 12 times the peak inference performance of its predecessor. Our inference platform showcases a 40 times improvement in performance compared to CPUs, making it ideal for hyperscale data centers with the TensorRT software stack. With this launch, NVIDIA is well-positioned to capture the data center inference market, aiming at every server node within hyperscale data centers. Another significant product launch this quarter was the NVIDIA RTX Server reference architecture, which can host up to eight Turing-based RTX 8000 GPUs. This offering opens new opportunities for GPUs in photoreal rendering, essential for industries like media and entertainment, retail, product design, manufacturing, and architecture. Prior to Turing and its ray-tracing capabilities, GPUs were unable to handle this workload effectively, resulting in most rendering being done on CPUs. An RTX-accelerated render farm can achieve equivalent performance for one-fourth the cost, one-tenth the space, and one-eleventh the power compared to CPU render farms. NVIDIA's RTX platform has received substantial industry backing from key developers such as Adobe, ANSYS, Autodesk, and Dassault, among others. Lastly, NVIDIA introduced a GPU acceleration platform for data science and machine learning called RAPIDS, enabling companies to analyze vast amounts of data and make accurate business predictions at unprecedented speeds. Until now, data analytics and machine learning have been the largest high-performance computing applications that haven't been accelerated. Nearly all enterprises utilize data analytics to derive insights from large datasets for various applications, including detecting credit card fraud, forecasting retail inventory, and understanding customer purchasing behavior. RAPIDS is an open-source collection of libraries for GPU-accelerated analytics, machine learning, and forthcoming data visualization. With RAPIDS, NVIDIA GPUs can accelerate machine learning up to 50 times faster than CPUs, opening a $20 billion server market for data analytics and machine learning workloads to GPUs, with broad industry support from Oracle, IBM, SAP, Dell EMC, Hewlett Packard Enterprise, Microsoft Azure machine learning, Google, Q-Flow, and the open-source community. With a unified architecture and ecosystem, NVIDIA GPUs can meet the needs of the evolving high-performance computing market, encompassing scientific computing, deep learning, and machine learning. Our GPUs and software stack enhance a wide range of workloads, from software in supercomputers to deployments in hyperscale data centers. This week, we showcased this capability at the Supercomputing Conference in Dallas, where the number of systems using NVIDIA GPUs on the TOP500 supercomputer list increased by 48% from last year, including the top two systems globally. Moving to pro visualization, revenue reached a record $305 million, an increase of 28% compared to the previous year and a 9% increase sequentially. The strength extended across desktop and mobile platforms, impacting several key industries such as the public sector, manufacturing, and architecture, engineering, and construction. At SIGGRAPH in August, we announced our Quadro RTX 8000, 6000, and 5000 GPUs based on the Turing architecture. Earlier this week, we introduced the Quadro RTX 4000, the most advanced professional GPU priced under $1,000. These GPUs will transform the work of 50 million designers and artists by allowing them to render photorealistic scenes in real-time and utilize AI in their workflows. The Quadro RTX series began shipping in Q3, with the high-end server-grade products recognized in data centers. We have already engaged a variety of customers on RTX, including major movie studios and game developers, and the feedback has been very positive. Turning to automotive, automotive sales in Q3 reached $172 million, an increase of 19% from a year ago and a 7% increase sequentially. This growth reflects our progress in autonomous vehicle development and production, as well as the rollout of next-generation AI-based cockpit infotainment systems. At GTC Europe, we announced that Volvo Cars has selected NVIDIA's DRIVE AGX Xavier for their next-generation vehicles. The initial production release scheduled for the early 2020s will deliver Level 2+ assisted driving features, including 360-degree surround perception and a driver monitoring system. This marks our first Level 2 mass-market car design win. In addition to Volvo, global automotive suppliers Continental and Veoneer have also chosen NVIDIA DRIVE AGX for their autonomous driving systems. Furthermore, our DRIVE AGX Xavier development kit started shipping this quarter. This is the world's first autonomous driving platform capable of running our NVIDIA DRIVE software for autonomous driving, which includes data collection, 360-degree surround perception, advanced driver monitoring, and in-vehicle visualization. With this platform, customers can develop and test their autonomous driving solutions and transition smoothly to production. We are enthusiastic about the AV opportunity as we look ahead to next year and beyond. Moving to the rest of the P&L and the balance sheet. Q3 gross margins was 60.4% and non-GAAP was 61%, below our outlook due to the $57 million charge for prior architecture components and chips following the sharp falloff in crypto demand. GAAP operating expenses were $863 million and non-GAAP operating expenses were $730 million, up 28% year-on-year. We continue to invest in the key platforms driving our long-term growth, including gaming, data center, and automotive. GAAP net income was $1.23 billion, and EPS was $1.97, up 48% from a year earlier. GAAP net income benefited from the reduction of $138 million in our U.S. tax reform transition tax amount as well as other discrete tax items. Non-GAAP net income was $1.15 billion and EPS was $1.84, up 38% from a year ago, reflecting revenue growth, gross margin expansion, as well as lower income tax expense. Accounts receivable was $2.22 billion compared to $1.66 billion in the prior quarter as Turing's RTX shipments began in the latter part of the quarter. Inventory at the end of the quarter was $1.42 billion compared to $1.09 billion in the prior quarter, reflecting the ramp in production of Turing products. Quarterly cash flow from operations was $487 million. Capital expenditures were $150 million. This fiscal year, we have returned $1.13 billion to shareholders through the end of Q3. We've announced a $0.01 increase in our quarterly dividend to $0.16 effective in Q4 of fiscal 2019. We are also pleased to announce an increase of $7 billion to our share repurchase authorization and that we intend to return an additional $3 billion to shareholders by the end of fiscal 2020. With that, let me turn to the outlook for the fourth quarter of fiscal 2019. As noted earlier, our revenue outlook is impacted by the expected work-down of Pascal midrange gaming card inventory in the channel. In addition, we expect a decline in our gaming console revenue given seasonal build patterns. Keep in mind that the midrange desktop portfolio is typically about 1/3 of our gaming business. Our outlook assumes that channel inventory weeks approach normal levels exiting Q4 and that gaming demand increases in Q4 compared with Q3. Now in total, we expect revenue to be $2.7 billion, plus or minus 2%. GAAP and non-GAAP gross margins are expected to be 62.3% and 62.5%, respectively, plus or minus 50 basis points. GAAP and non-GAAP operating expenses are expected to be approximately $915 million and $755 million, respectively. GAAP and non-GAAP OI&E are both expected to be income of $21 million. GAAP and non-GAAP tax rates are both expected to be 8%, plus or minus 1%, excluding discrete items. Capital expenditures are expected to be approximately $190 million to $210 million. Further financial details are included in the CFO commentary and other information available on our IR website. In closing, I'd like to highlight some upcoming events for the financial community. We'll be presenting at the Barclays Global Technology, Media, and Telecommunications Conference on December 6, and we will be meeting with the financial community at the Consumer Electronics Show in Las Vegas from January 8 through 11. And our next earnings call to discuss our financial results for the fourth quarter of fiscal 2019 will take place on February 14. With that, we will now open the call for questions. Operator, will you please poll for questions?

Within your guidance for the January quarter, is the team anticipating continued sequential growth in your data center business? There seems to be some concern around a near-term slowdown in cloud spending. But on the flip side, we're hearing that the NVIDIA team is actually seeing pretty strong demand, near term, from some of your China cloud customers for your Tesla-based products. So I just wanted to get your views on cloud data center dynamics and the trajectory into the January quarter.

We expect to keep performing well in data centers. With Moore's Law coming to an end, demand for computing is still rising, especially as data centers increasingly run machine learning algorithms, which require significant computational power. To meet this demand, the only option is to acquire more computers and CPUs, as the speed of individual CPUs is not significantly increasing. Consequently, capital expenditures for data centers will need to rise. However, our alternative approach, which contributes to the rapid adoption of NVIDIA's accelerated computing platform, offers a way to move beyond Moore's Law. Recently, we've achieved several significant milestones. First, we launched a new accelerated computing platform called RAPIDS. The industry is enthusiastic about deep learning as a method for artificial intelligence, which is very data-intensive and requires substantial domain expertise across sectors like retail, finance, healthcare, and logistics. Traditional machine learning methods have been effective, but they have not been accelerated until now. We've collaborated with the open-source community over the past few years to develop a comprehensive stack that includes Apache Arrow, the Dask parallel distributed computing engine, and all our algorithms that work alongside these tools, creating an accelerated machine learning platform that has generated considerable excitement. Additionally, our Turing architecture enables affordable film rendering beyond what Moore's Law could have achieved. We also recently announced our first Turing-based T4 Cloud GPU, accompanied by an extensive software stack, including Kubernetes, Docker, and our second-generation Tensor Core AI accelerator, which has fostered significant interest in data centers. Therefore, I anticipate our data center business to continue to do very well.

Great. And then just on the high-end Turing products that the team started rolling out in October, early demand actually seems to be quite strong. And I think part of it is just the lineup of AAA-rated games. eSports continue to be strong as well, obviously a big motivator for your enthusiast-class gamers. I know the team is, near term, kind of working down midrange Pascal cards, but do you anticipate your Turing-based RTX product families to drive sequential growth in the January quarter just what appears to be pretty strong demand pull for these products?

Yes, the Turing launch took place at the end of the quarter and represents the most significant generational advancement we have ever achieved. It features real-time ray-tracing and is the first gaming GPU to incorporate artificial intelligence. At every price point, it delivers significantly higher performance than its predecessor and is the highest performing GPU globally. We're seeing great content being released, including Battlefield V, which was launched with real-time ray-tracing, marking the world's first application to support this technology. We anticipate Turing will perform exceptionally well. As we move forward, we plan to integrate Turing more deeply into the mainstream market. While we don't have any announcements today, our goal is to make this new architecture available to as many gamers as possible.

I had a question on the gaming outlook and as it relates to channel inventory. Colette, you mentioned that typically, the midrange is about 1/3 of gaming. How much was it in the October quarter? And are you effectively assuming close to 0 in the January quarter? And related to that, I think with your game console business, I'm estimating you did something around $200 million to $250 million in October. Again, is that coming down pretty hard into January?

Yes. I discussed the overall size of the midrange as we've seen it over a long period. It's been consistently about one-third over approximately eight quarters, and we believe that's a useful figure for our Q4 guidance. From a broader console perspective, we've typically seen figures in the hundreds, sometimes exceeding that in earlier quarters. What we're experiencing now is a usual buildup before the holidays in Q3, which then tends to slow down as we progress into Q4 during the holiday season.

Great. And then as a follow-up, you saw nice acceleration in growth in your ProVis business this quarter. Jen-Hsun, I think you've talked extensively about the RTX and the long-term opportunity there. What are your expectations going into 2019 for that segment?

I expect ProVis to grow. The ProVis platform has an increasing amount of digitally created content. Most images that seem like photographs are actually rendered using software. Almost every catalog, video, movie, and TV show now incorporates significant digital rendering. Until now, rendering has not been possible in an accelerated manner. RTX is the world's first accelerated ray-tracing GPU, and the response from the digital content creation market has been very positive. I definitely anticipate that, as I mentioned earlier, the workstation industry hasn't seen a fundamental platform architecture change in nearly a decade, and RTX represents that change. Therefore, I expect ProVis to perform very well.

I'm curious, Jen-Hsun, what needs to happen to reduce the midrange Pascal inventory? Is it pricing, or is there something else involved? The expectation was that this inventory could be cleared by the end of the October quarter, but that hasn't happened. Do you think people were holding off for the Turing launch, which might have caused a delay? Additionally, Colette, how should we view seasonality in the April quarter considering you mentioned it might take one or two quarters to reduce this inventory?

Yes, we entered Q3 with excess inventory due to the aftermath of the crypto situation. We anticipated a decline in pricing in the market. However, this decline happened more slowly than we had projected. While we were expecting sales volume and demand to increase, it seems that the adjustments in pricing and the growth in volume both took longer than anticipated. Currently, most pricing has dropped to levels that are at or slightly below where they were before the launch. I am hopeful that with pricing now stabilized, customers will return to making purchases. When there's price volatility in the channel, it tends to cause a pause as people wait for prices to level out, which took longer than we predicted. But now that prices are at appropriate levels, we expect the market to normalize. The 1060 remains the top-selling graphics card globally, and we've decided not to increase supply to the channel in the upcoming quarter to allow existing inventory to be sold. We will monitor the situation, but we anticipate that inventory levels will return to normal by the end of the quarter.

Okay. Vivek, to answer your question also regarding Q1 in terms of what we're going to see in terms of the expectation, as the channel inventory normalizes at the end of Q4, we do believe going into Q1, we will probably be up from where we end in terms of Q4. So we won't follow that normal seasonality between Q1 and Q4. We do expect to be up as we go into Q1.

I guess, a follow-up question on the channel inventory side. It looks like it's roughly $600 million kind of a drawdown here. And just curious, does that sound right? Number one. Two, does that fit with what you are hearing from your channel partners in terms of what's excess? And then as part of that, are you drawing down inventory in the current quarter ahead of Turing architecture launch into the mainstream?

I'm not sure I fully understand the last question. I believe the answer to your first question is yes, and you articulated it well. What was the last question again?

The last question was regarding our midrange. Is there any statement about future Turing products that were taken into account?

We haven't announced our future Turing product, but we expect to create a Turing GPU that targets the mainstream parts of the marketplace. While we're not making any announcements at this time, it is typical for us to do so.

I guess, the question was whether you were drawing down inventory perhaps below normalized levels in the current quarter.

No, we're really not shipping into the midrange segment of Pascal so that we give the channel an opportunity to sell through the product that has. And we would like to see channel inventory get normalized by the end of Q4, and then we'll get back to doing our work.

My first one is for Colette. I just wanted to be a little more explicit. If I think about your business split into sort of gaming and non-gaming, are you looking for the non-gaming pieces in aggregate to grow sequentially into Q4?

Stacy, yes. I think the answer to that is yes. In aggregate, yes, we do believe the rest of the business will grow sequentially.

Got it. I guess that fits with the kind of 1/3 you're talking about because that implies the gaming down 30%-plus. So that is what your kind of magnitude that you're thinking about at this point.

That is correct.

For my second question, I wanted to address that the concern about the channel potentially becoming saturated isn't a new one. However, over the past several quarters, you've maintained that you felt you had a good grasp of the channel, which now appears to be inconsistent with the current situation. Can you provide more clarity on what changed and when you noticed it during the quarter? Was this realization something that occurred later in the quarter, or did you enter the quarter aware that the inventories were high and needed adjustment? What's the situation, especially since the tone seems different from what we've heard in your previous earnings calls?

We were surprised by the situation, just like everyone else. The impact of the cryptocurrency decline lasted longer than we anticipated. Prices began to drop, and we expected this to happen much more rapidly, as well as the demand to pick up quickly once prices fell. The channel aimed to protect its pricing, and there was uncertainty surrounding crypto, which affected how soon we thought demand would stabilize. This overall uncertainty seemed to hold the market back longer than we had expected. Currently, pricing has dropped below pre-launch normal levels. I am optimistic that we will see demand recover and sales improve during the holidays, and we are already noticing this. We didn't foresee this situation, nor did we grasp its extent until near the end of the quarter. What was the other question? Was there another question? I believe that covers it.

With regards to Turing ramp, I guess, how is that going relative to your expectations? It seems like availability is quite a bit better now. And where do you stand with DLSS support? I know you've announced a number of games that will have DLSS support by year-end. Like how many of those are already supporting that technology?

Yes, the ramp is progressing well, and I believe this is the most significant generational leap we have experienced. This new technology in computer graphics is the most substantial advancement seen in the last decade. Real-time ray-tracing has been a long-held dream, and today, with Battlefield V, people are experiencing real-time ray-tracing for the very first time, and the visuals are stunning. The ramp is strong. Turing ramped towards the end of the quarter in a situation unlike any GPU before it. Nonetheless, demand for our high-end products is fantastic. The 2080 Tis are mostly sold out, and I believe they are still unavailable anywhere. The demand is impressive, and I anticipate a fantastic new generation. You’ve seen the first title released, FINAL FANTASY, and we have a pipeline of about 30 titles that we are diligently working on. These games will enable RTX upon release. I must emphasize that aside from content, RTX offers higher performance at the same price point compared to any graphics card available. Across all price points, it is the highest performing graphics card. Therefore, it is clearly the top-performing GPU in the world, and many great new features are on the way.

I'm still trying to understand the scale of the channel inventory reduction because if you haven't shipped the midrange cards for an entire quarter, that suggests your inventory is greater than a quarter's worth of sales, which is over 12 weeks. So my first question is whether I'm correct in thinking that you have enough inventory to cover a full quarter of sales. My second question is that while you're reducing this inventory, I expected to see the Turing high-end cards, specifically the 2080 and 2080 Ti, ramp up in the fourth quarter. However, it appears that, aside from the inventory, the overall GPU business in gaming will likely remain relatively flat sequentially.

I'm trying to figure out what the first part of the question was.

The first question was whether the midrange of Pascal had more than 12 weeks of inventory, if it's going to take more than a quarter to bring it down.

I believe the channel currently has over 12 weeks of inventory between us and the other brand. One challenge we face is estimating the inventory levels of other brands. Additionally, our quarter starts a month later, which means any actions we take or observations we make in the channel reflect a month after their quarter ends. The inventory concerns are not solely ours; they also involve other brands. Unfortunately, it is quite difficult for us to assess the inventory levels of those brands. We do our utmost to estimate, but clearly, our estimates have not been sufficiently accurate. Therefore, to answer your question, my perspective is that we have about 12 weeks of inventory remaining to sell through at this time.

Could you compare this product cycle transition from Turing to the one you experienced with Pascal? Is the main difference just the effects of the crypto downturn, or are there other factors affecting the transition? You've mentioned that Turing represents a significant generational leap, and I’m curious if that larger difference affects the transition as well.

Turing is the highest performance GPU at every price point, and its transition was unaffected. The main issue was the effects of the crypto hangover. Turing had a successful launch, and we ramped it up as anticipated towards the end of the quarter. The ramp was impressive, and everyone performed excellently, resulting in fantastic performance and excitement. However, the situation beneath Turing was inconsistent, which we didn't fully recognize until the end of the quarter. As we looked ahead to this upcoming quarter, we decided not to ship more products into this segment because of the existing inventory, allowing the channel to sell through the midrange Pascals. In about a quarter, we will resume normal operations. This is certainly a setback, and I wish we had identified it earlier, but it's hard to say what we might have done differently. The unexpected and unanticipated gradual decline in pricing in the channel and the delay in volume pickup after prices dropped, compounded by other brands' inventory in the marketplace, made the situation worse than we had anticipated.

Maybe I can ask the question a little bit differently on the gaming business. If I look back over the past several quarters, let's say you've been running at roughly a $1.6 billion to $1.8 billion revenue level since the October '17 quarter. Prior to that, you were at $1.1 billion, $1.2 billion. We look like we're now going back to that level. I guess, the question is, do we build off of that level? Do we bring back half of the inventory burn? How do I think about the return of year-over-year growth in that gaming piece of the business as we start to look into fiscal 2020? And then a quick second question. Over the past few years, you've had really strong seasonal sequential growth in the data center business, in Q4, about 20% sequentially. I'm just curious, how is your guide factoring in the sequential growth in that piece of the business into this current quarter?

Our data center business is performing very well. The fundamentals of accelerated computing are aligning perfectly, especially as we reach the limits of Moore's Law. This year, we saw a significant increase to 127 systems in the TOP500, reflecting almost 50% growth year-over-year. We lead in the United States and globally, including Europe and Japan, and rank 22 out of the top 25 most energy-efficient computers in the world. Additionally, we've launched three new initiatives aimed at expanding our presence in the high-performance computing market, focusing on machine learning, which is currently the largest segment of artificial intelligence that has not yet been accelerated. We're also introducing photorealistic rendering for film and a new cloud GPU called T4, which is generating tremendous excitement. The speed at which Google incorporated T4 into their cloud was remarkable, taking only 30 days from production to availability. I anticipate T4 will perform very well. Overall, I feel positive about the dynamics of our data center business. Regarding the forecast, we'll see how it evolves, but I believe the foundational elements are strong. Now, going back to your question about gaming.

So the statement came in regarding, you've bumped up the overall gaming somewhere in mid of the year to about a $1.7 billion gaming business, where maybe if you look back 2 years, you were at about $1.1 billion. At this stage, when you come out of the setback that we have here to get through the overall channel inventory, where will you come out after that? And what type of growth could we expect?

Yes, I'm going to let you handle the modeling, but I want to emphasize that the gaming market remains fundamentally the same as we understand it. Cryptocurrency is a significant factor that we all need to come to terms with. We believed we managed the cryptocurrency dynamics well, but as prices began to fall, we expected demand to follow suit more quickly than it has, and that transition has taken longer than anticipated. In the gaming marketplace, our notebook gaming sales, which are unaffected by cryptocurrency, grew 50% year-over-year in China, indicating that the market remains strong. RTX will undoubtedly transform gaming computer graphics. Overall, the dynamics are positive, but we do need to address the channel inventories. This quarter, we made the simultaneous decision to halt shipments of midrange products while also following the usual seasonal console production schedules, which typically ramp up a quarter before the holidays. These two factors are happening at the same time. However, there isn't anything about the gaming marketplace or business that we perceive as fundamentally changed.

Yes. To kind of add to that, think about our gaming business in several pieces that we talked about in terms of the tremendous strength that is also continuing. In terms of our success in terms of Turing, our notebooks for gaming are growing extremely strong, and our overall console business is also extremely healthy as well. So to think about all of the different components, we just have a piece of channel inventory at the midrange, but overall, as you can see, gaming is also growing quite well.

A question with regard to inferencing and what we can expect from that for both Q4 and going forward. And perhaps I don't know if it's a valid comparison to compare what we might expect from inferencing after the new Quadro launches to what happened in training after the Volta launches. Is there any comparison there in terms of magnitude for how the ramp goes?

The ramp of T4 is entirely linked to customers adapting their models to our platform. Inference models are quite complex. Many people may perceive inference as straightforward due to the numerous ASICs being discussed. However, most of the complexity lies in the optimizing compiler. The fifth-generation optimizing compiler, TensorRT, that we recently announced took several years to develop. To achieve scaling similar to what is seen in Google Cloud, we needed to create a TRT server, an inference server that allows multiple models to operate on Kubernetes in the cloud. Developing that software is also highly intricate. The technology we are assembling has come together, and we are now collaborating with Internet companies globally to port their heaviest workload applications onto T4. Our efforts are focused on this, and ultimately, it's up to these companies to decide how many units they wish to purchase, which will inform us about our adoption rate. From a broader perspective, it's clear that Moore's Law has reached its limit, yet more data centers continue to implement deep learning and machine learning models, which are computationally demanding. Looking ahead, the T4 Cloud GPU stands out as the most efficient option available. It can manage various types of models, including image, recommendation, and speech synthesis models. It has the highest throughput for any processor at 70 watts, suitable for a hyperscale data center OCP server, and boasts the lowest latency for inference, below 1 millisecond. Given the architecture, the software technologies in place, and the growing need for Internet companies to accelerate their efforts, I believe T4 is extremely well positioned, and I look forward to sharing our successes with you soon.

First, Jen-Hsun, I appreciate all the details on the T4 for inference in the data center. Could you likewise highlight the current traction you're seeing and the long-term growth expectations for the Jetson product that's designed for really, I think, a different market? It's inference at the edge, right?

Yes, Jetson is created for edge AI. One version of Jetson is self-driving cars, featuring high-performance capabilities with complex software. This quarter, we announced our first mass-market Level 2 success. We have excelled in robot taxis, Level 4s, trucks, shuttles, and high-end systems, which involve numerous processors and sensors, including LiDARs and surround cameras that require extensive computation. However, we have not succeeded until now in utilizing the DRIVE platform for Level 2 mass-market vehicles. Volvo is our first announcement, marking our initial victory in high-volume production ramping in the early 2020s, and I anticipate many more to follow. We believe we have developed a solution that is not only highly useful and easy to use but also capable of delivering Level 2 functionality with a single chip for the first time. Xavier is currently in production and is the only single-chip autonomous vehicle processor available today. This same platform can also be applied to various edge AI devices, such as manufacturing picking robots, autonomous retail systems, and medical instruments that can recognize and identify anomalies in medical imaging. These applications are increasingly focused on AI, which is why we built Jetson.

I'll ask a clarification and then a question. The clarification is just on the inventory issue, and thanks for all the color. But one, are Pascal 1070s and 1080s and in Ti flavors still selling? And if so, could they present any kind of inventory risk either later this year or in early fiscal '20? And then the question really, Jen-Hsun, is trying to get a better understanding of how you see the intermediate-term growth rate of the data center business. You had spectacular high-performance compute TOP500 accelerator penetration performance up 50-plus percent. That about matches the growth in the data center business. Those may be somewhat coincidental, but can you just talk about where you see penetration across key end markets like HPC, like cloud and hyperscale and like enterprise? Which offer you the best growth from here? And where do you feel like your penetration may be more mature? I'm just trying to get a sense if there's an acceleration coming off of the 50% year-on-year growth that we're seeing now or if, consistent with the recent trend, we might be moderating potentially down into the 40% or 30% range as we go into the next calendar year.

Our high-end Pascal GPUs are mostly sold out, and we've effectively managed the transition to ramp up our high-end Turing products. Regarding our data center business, I would describe last quarter as one of the best we've ever had, despite some inventory challenges. This success stems from the strong position we hold in data centers and accelerated computing, which are fundamental to our company. Initially, our journey into accelerated computing focused on scientific and high-performance computing, but over the past five years, we've proactively invested in deep learning and helped software developers worldwide adopt it. Currently, I'm particularly excited about three new areas we've established this past quarter with the launch of Turing and RAPIDS. First, we see significant potential in the film rendering market, where approximately 10 million CPU nodes globally can benefit from accelerated computing as Moore's Law reaches its limits. Second, we are addressing inference in the hyperscale data center market, which is projected to see around 15 million CPUs sold this year, growing at about 15% annually. Given the end of Moore's Law, these servers will need to embrace acceleration, and our T4 GPU is ideally suited for this, designed for power-sensitive environments typical in hyperscale data centers. We've invested years in developing a comprehensive software stack that includes Kubernetes, containers, a TRT compiler, and inference servers, all integrated into our NGC cloud. This work has positioned us favorably. The third and potentially largest opportunity lies in machine learning, where technologies like Hadoop, Spark, and various popular algorithms have not previously benefited from accelerated computing. After two years of effort, we've open-sourced RAPIDS, making it available on our NGC cloud, and it's being integrated by major players like IBM and various cloud providers into their machine learning platforms. Our growth rate reflects a significant expansion in our accelerated computing and data center opportunities this quarter. With the T4 GPUs and RAPIDS platform alongside our RTX server for film rendering, we have broadened our data center prospects. I anticipate continued success in the accelerated computing space for data centers.

Thanks, everyone. To sum up, the crypto hangover has left the industry with excess channel inventory. It will take 1 or 2 quarters to work through it. This is an unexpected near-term setback and doesn't change the fundamental dynamics of our company. The end of Moore's Law has cleared a way for NVIDIA accelerated computing as a great path forward. Turing opens up 3 exciting markets for us with ray-tracing games, film rendering, and hyperscale inference. And with our first win in mainstream Level 2 self-driving cars with Volvo, our DRIVE AV platform is gearing up for the mass market, and our competitive position has never been stronger. We look forward to updating you on our progress. Thank you.