NVIDIA Corporation (NVDA) Earnings Call Transcript & Summary

Hi, everyone. Happy New Year, and welcome to our special event. I will kick it off with gaming and creators, and then Ali Kani will talk about developments in autonomous vehicles. So let's get started. [Presentation]

Gaming is huge. People no longer ask if it will keep growing. Of course, it will. Now the question is, how big will it get? There are almost 3 billion gamers in the world with 150 million more people gaming this year. Today, the market for games and gaming hardware is almost $300 billion, up $70 billion from just 3 years ago. eSports continues to be an important driver. League of Legends Worlds 2021 Finals grew 74 million peak concurrent viewers. The 2021 Dota 2 International paid out a $40 million price pool. All told, eSports has an online audience of 460 million viewers, and people are tuning into all types of gaming content. There are now 10 million unique streaming channels on Twitch, doubling over the past 2 years. 750 million people watched gaming content online in 2021. And the gaming universe is expanding into virtual worlds, with our own economies. Roblox has attracted 200 million monthly users. Lil Nas X hosted a concert inside Roblox in front of 33 million fans. This is also fueling virtual reality, where headsets are expected to grow 4x in the next 4 years. But it's not just about playing. There are 45 million professionals and freelance creators, fueling an explosion in digital content. They will become the builders of our 3D digital worlds. And as the gaming experience extends into these vast open virtual worlds, some see an opportunity in the trillions. Over the past 20 years, we have dedicated ourselves to build the best platform for gamers and creators, not just to delight our customers, but to develop graphics technologies that enable game developers and creative ISVs to do some really amazing work. At the heart of the platform is our GPU and a history of revolutionary architectures like RTX. All told, we have shipped almost 1.5 billion GPUs and on top of the hardware comes a massive investment in our software stack that harnesses the GPU to deliver graphics innovations like game-ready drivers, DLSS, G-SYNC, Reflex, NVIDIA Broadcast, Max-Q, Omniverse and GeForce NOW. With billions of devices unable to play the latest games, it's no surprise that cloud gaming is expected to grow to over 100 million subscribers by 2024. For these gamers, we built GeForce NOW, delivering GeForce class gaming to over 15 million members around the globe on Chromebook, MAX, SHIELD TV, Android and iOS phones and underpowered PCs. GeForce NOW has over 1,100 PC games, including 90 free-to-play. From the leading PC game stores, including Steam, Epic Games, Ubisoft Connect and EA Origin. And extending our partnership with EA, I'm excited to announce that Battlefield 4 and Battlefield 5 are available on GFN starting today. Be sure to tune in to GFN Thursdays where every week we onboard new games for our members. We've also been working with the world's largest TV manufacturers to bring GeForce NOW to millions more living rooms. Last month, we launched a beta version of GFN on LG 2021 webOS Smart TVs. Today, we're pleased to announce a partnership with Samsung to integrate GeForce NOW into their smart TVs. Starting in Q2, Samsung Smart TVs will be streaming PC class gaming with GFN. The rapid expansion of 5G offers the potential for low-latency cloud gaming everywhere. So we've teamed with AT&T and their fast reliable 5G network to bring the power of PC gaming to mobile devices. As a 5G technical innovation partner, GeForce NOW on the AT&T 5G network offers ultra-responsive cloud gaming on the go. We also have an exclusive offer to announce. Starting today, AT&T customers with a 5G device on a qualifying plan can get a 6-month GeForce NOW priority membership at no charge. GeForce NOW members get access to the latest GeForce technology. Our new RTX 3080 tier service powered by the GeForce NOW Super Pods is the first to support 120 frames per second at 1440p on PCs, up to 1600p on MacBooks and 4K with HDR on SHIELD TV, available today at just $99 for a 6-month membership. Play the most demanding PC games, no waiting, no downloads on the latest GeForce RTX hardware in the cloud. RTX momentum continues to build. Ray tracing and AI are defining the next generation of content. Since last CES, we've added over 150 RTX games and applications, and gamers have logged over 1 billion hours of RTX play time. The Day Before from FNTASTIC is an open world MMO survival set in a deadly post-pandemic America, overrun by flesh-hungry infected and survivors. It's in the top 10 most anticipated games on the Steam wishlist. Today, we are announcing The Day Before will launch June 21 with RTX on, including ray tracing and DLSS on GeForce and GeForce NOW. Here's the first look at The Day Before with RTX on. [Presentation]

With over 70 million players, Tom Clancy's Rainbow Six Seige from Ubisoft has become one of the biggest competitive shooters. The next game in the series, Rainbow Six Extraction, gamers will compete in a tactical co-op FPS as they engage, contain and eliminate the mysterious creatures known as the Archaeans. Today, we are announcing that Rainbow Six Extraction launching January 20 will be RTX on featuring DLSS and available on GeForce and GeForce NOW. Here's a first look at DLSS in Rainbow Six Extraction. [Presentation]

Escape from Tarkov is an online first-person action game. Players battle through the war-torn city of Tarkov, which has been sealed off by UN and Russian military forces. Escape from Tarkov quickly became one of the top 10 most played competitive shooters. Last June, RTX gamers gained a competitive edge with NVIDIA Reflex. We are extending our partnership with Battlestate Games to bring more performance with the addition of DLSS. Let's take a look. [Presentation]

Dying Light, the tremendously successful open-world action RPG by Techland, had over 23 million players worldwide. The highly anticipated sequel, Dying Light 2, will be launching February 4 with RTX on featuring ray tracing and DLSS. It's available on both GeForce and GeForce NOW, Here's an exclusive look at never before seen RTX on action. [Presentation]

For competitive gamers, milliseconds are the difference between winning and losing. We ask gamers to check out the benefits of lower latency gaming with our system latency challenge. Early results show up to 2x aiming improvement, and this is why 8 of the top 10 competitive shooters have integrated NVIDIA Reflex, our low-latency gaming platform, and gamers love it. Over 20 million GeForce players compete with Reflex on each month. And the hardware ecosystem continues to grow. Our Reflex Latency Analyzer allows players to easily measure system latency with a single click of the mouse. We now have over 50 reflex mice and monitors from 16 partners. While Reflex is critical for competitive matches, lower latency is important in any game as it better connects the gamer and their physical actions to those in the game. Today, we are announcing 7 more Reflex games and extending Reflex into new genres. These include iRacing where you can steer with better precision in the world's premier online racing simulator. Rainbow Six Extraction where Reflex has your Six as you confront Archaeans. And Sony's award-winning God of War, beat the boss with better dodges and combos when it launches on PC on January 14. Let's take a look. [Presentation]

With the global growth in eSports, the demand for eSports displays has been doubling each year. We believe the time is right to take eSports to a new level. Today, we are announcing a new category of eSports displays, 1440p, 27-inch G-SYNC and up to 360 hertz, with new eSports vibrance and Reflex latency analyzer. These displays are also dual format. They can switch to 1080p 25-inch if that's preferred for certain games. Here's Seth to fill in the details.

1080p displays have ruled eSports for over a decade, as the lower resolution has enabled higher frame rates and faster refresh rates. But our research shows that larger higher-resolution panels improve aiming performance, providing larger and more detailed targets to acquire. In fact, 1440p 27-inch displays can improve aiming by up to 3% over traditional 1080p 24-inch displays. For competitive gamers, that's huge. With today's high-end GeForce RTX GPUs rendering eSports games well above 360 FPS at 1440p, the industry is primed for a change. Today, NVIDIA is announcing a new class of 1440p 27-inch G-SYNC eSports displays. These displays feature industry-leading refresh rates up to 360 hertz with the smoothest, lowest latency game. eSports tuned vibrance color profiles, powered by mini LED backlights and oxide transistors, make targets pop from the screen and enhance easier to use Reflex Analyzer automatically measure system latency. So gamers can head into competition with full confidence in their PC's performance. And a gamer still wants to play a match at 1080p, the monitor features a special dual-format 25-inch 1080p mode. The future of eSports monitors is here with new displays available soon from AOC, Asus, MSI and ViewSonic.

A decade ago, we introduced the 50-class GPU, offering gamers great performance on the latest architecture at an affordable price. It was the first step up into GTX gaming. In fact, 3 of the top 5 GPUs on Steam are GTX 50-class. And one of the most successful is the GTX 1050. When it was released in 2016, the GTX 1050 could power the top games of the time at over 60 frames per second, but the production value of games continues to rise, and modern games need much more GPU horsepower. The GTX-1050 struggles to keep up. Even a 1050 Ti or the more modern GTX 1650 can't power the latest games at 60 frames per second. Today, we are announcing the RTX 3050. The GeForce RTX 3050 brings the performance and efficiency of the Ampere architecture to more gamers than ever before. It powers the latest games at over 60 frames per second. Based on our Ampere architecture, the 3050 comes equipped with second-generation RT cores for ray tracing and third-generation Tensor cores for DLSS and AI. For the first time, you can play ray-traced games on a 50 class GPU at over 60 frames per second. Ray tracing is the new standard in gaming and the 3050 makes it more accessible than ever. The RTX 3050 comes equipped with 8 gigabytes of G6 memory. And starting at just $249, it will be available worldwide on January 27 from all of our partners. With 75% of gamers still on GTX GPUs, this is the perfect time to step up to RTX. We are at the dawn of the next digital frontier, interconnected 3D virtual worlds, whether for commerce, entertainment, creativity or industry are being built today. These worlds are boundless and will be populated with shops, homes, people, robots, factories, museums, an infinite amount of 3D content, and they will be built by an expanding number of creators, collaborating across the globe, constructing these amazing virtual environments. For this future, we've built NVIDIA Studio, a fully accelerated platform for creators. It starts with RTX, the most advanced GPU architecture with hardware acceleration for ray tracing, simulation and AI and NVIDIA high-performance video processors. The NVIDIA Studio software includes specialized drivers and dozens of SDKs, which accelerate over 200 of the industry's top creative applications. We also offer our own studio apps, with the latest addition NVIDIA Omniverse for 3D design and collaboration. Today, we are announcing that NVIDIA Omniverse is out of beta and generally available to GeForce RTX Studio creators. 3D content is challenging to produce. It typically requires multiple tools, which are often incompatible. And then you have to wait for long render times. Omniverse brings over 20 years of NVIDIA's groundbreaking work in graphics, AI, simulation and compute into a single platform to transform 3D workflows. Let me explain how it works. Omniverse connects independent 3D design worlds together into a shared virtual space. The connecting language of Omniverse is USD or Universal Scene Description, which you could think of as the HTML of 3D worlds. Today, a 3D artist typically work sequentially across multiple applications, like 3DS Max for modeling, then Substance Painter for texturing and finally, Unreal Engine to arrange the scene, exporting and importing large files many times along the way. With Omniverse, artists connect their apps and then compose the combined scene using Omniverse create. Once in Omniverse, an artist can draw on NVIDIA SUPER Powers, like physics, which can let artists use true to reality simulations that obey the laws of physics. And RTX Renderer to see the scene in real time, fully ray traced or path traced. Omniverse also lets you collaborate with another artist from across the room or across the globe, connecting their favorite apps into a single shared scene. Changes made by one designer are reflected back to the other artists, like working in a cloud shared document, but in 3D. This is the future of 3D creation and how virtual worlds will be built. [Presentation]

We are partnering with over 40 ISVs to transform 3D workflows. Today, Omniverse is connected to leading applications like 3DS Max, Maya, Blender, Unreal Engine and many more to come. We now have several new additions to Omniverse. Today, we are announcing an even easier way for artists to collaborate with Nucleus Cloud. Now in early access, Nucleus Cloud simplifies Omniverse scene sharing. It's one click to collaborate and your entire 3D scene is online. Leading 3D marketplaces are now featuring Omniverse ready assets. Check out the collection of free assets now available in the Omniverse launcher. Omniverse Machinima has been a big hit with creators who love the game. It lets you remix and recreate your own game cinematics with thousands of game assets and environments. We are now adding MechWarrior 5 and Shadow Warrior 3 assets to the Machinima library. Audio2Face is a revolutionary AI-enabled app that easily animates a 3D face with just an audio track. We are now supporting blend shape and direct export to Epic's MetaHuman. AI powered by NVIDIA Studio opens up new capabilities for creators. Individual artists can now do the work of large teams. Application developers are quickly embracing the power of AI for content creation. Adobe Lightroom adds AI to quickly create detailed masks around subjects. Topaz Labs video enhanced AI uses AI to upscale videos, retaining higher details than traditional upscaling algorithms. Adobe 3D Substance Sampler uses AI to generate high-quality photo-real 3D materials from photographs. And Notch uses NVIDIA Maxine pose estimation to enable creators to drive a full body rig with a single webcam. NVIDIA Studio also features Canvas, our application that instantly converts brush strokes into photorealistic images. Today, we are announcing a major update to the Canvas AI built from NVIDIA's GauGAN 2 research, the latest app produces images that are 4x higher resolution with more detailed elements. It also adds new materials like flowers and bushes. The new Canvas app is available free to download at nvidia.com. Gamers and creators have made GeForce high-performance laptops, the fastest-growing category in PCs. In the last year alone, 35% more RTX laptops were sold and creators have fueled a 3x jump in studio laptop sales. We are announcing over 160 new laptops, all built around our RTX 30 Series GPUs. They include a wide range of form factors like dual screen, easels and convertibles. There are new 14-inch portable powerhouses and even more G-SYNC and 1440p displays. As more gamers and creators turn to laptops for their primary PC, performance expectations continue to increase. At the same time, they are demanding sleek form factors and longer battery life. In 2017, we reimagined gaming laptop design with an extreme focus on efficiency. We developed a system design approach that delivers high performance in thin gaming and studio laptops. We called it Max-Q. It has fundamentally changed how laptops are built. With the help of AI, we optimize everything. The GPU, CPU, fans, display, board design, laptop construction, even optimal settings for games. Today, we are announcing the fourth-generation Max-Q technologies. Let me show you what we've done. On laptops, power is shared between the GPU and CPU. CPU efficiency is critical for maximizing performance. So we developed CPU optimizer. We've worked with CPU vendors to create a new low-level framework, enabling the GPU to further optimize the performance, temperature and power of next-generation CPUs. As a result, CPU efficiencies improved and power is transferred to the GPU for more gaming performance. For creators and students who rely on compute-heavy apps like Adobe Premiere, Blender or Matlab, we've developed rapid core scaling. It enables the GPU to sense the real-time demands of the application and use only the core it needs. This frees up power that can be used to run the active cores at higher frequencies, delivering up to 3x more performance for intensive creative work on the go. On top of the great performance, laptop gamers also want more battery life. Battery Boost 2.0 has been totally rearchitected. Now, AI controls the whole platform, finding the optimal balance of GPU and CPU power usage, battery discharge, image quality and frame rates, all in real time. The result is great playability on battery, with up to 70% more battery life. Our fourth generation of Max-Q technologies with CPU optimizer, rapid core scaling and Battery Boost 2.0 continues our commitment to bring thin, high-performance laptops to gamers and creators. Today, we are announcing the RTX 3080 Ti laptop GPU, bringing the flagship 80 Ti class of GPUs to laptops for the first time, featuring 16 gigabytes of the fastest DDR6 ever shipped in a laptop, the RTX 3080 Ti delivers higher performance than the desktop TITAN RTX. RTX 3080 Ti laptops start at $24.99. The RTX 70 family is our fastest-growing class of laptop GPUs, introducing the RTX 3070 Ti. It's 70% faster than the RTX 2070 super laptops and delivers 100 frames per second at 1440p. RTX 3070 Ti laptops start at $14.99. And laptops powered by both of these new GPUs will be available starting February 1. For creators looking for the best laptop for your work, we are also announcing new NVIDIA Studio laptops. With the latest RTX GPUs, these laptops are 7x faster than the newest MacBook Pro 16 in 3D design, with RTX hardware accelerated ray tracing, AI, NVIDIA's high-performance video processor, they are the perfect tool for any creator workflow. Together with NVIDIA Omniverse, Studio Laptops are ready to build the virtual worlds of the future. With over 100 Studio Laptop designs, there's a wide variety to meet every creator's needs. This includes ultraportable 14-inch Thin and Light, perfect for remote workers or students like the new Razer Blade 14. Dual-display laptops for video editors, like the new ASUS ZenBook Pro Duo, and the most powerful laptops for creators like the new MSI Creator Z17, featuring RTX 3080 Ti. Let's see what digital artists Jacob Norris could create in 48 hours with a powerful 3080 Ti laptop. [Presentation]

To recap, GFN delivers the ultimate cloud platform for PC gamers. Today, we are bringing more games, more devices and more networks to the GFN ecosystem. RTX is the new standard. And the GeForce RTX 3050 makes it more accessible than ever. With 75% of gamers still gaming on GTX GPUs, this is the perfect time to step up to RTX. Laptops are the fastest-growing PC platform. Today, we announced 160 new laptop designs, including our latest laptop GPUs, the 3080 Ti and 3070 Ti and our fourth-generation Max-Q technologies. We announced NVIDIA Omniverse for 3D design collaboration is now generally available for GeForce RTX and Studio creators. This is the future of 3D content creation and how virtual worlds will be built. Over the past 20 years, we have dedicated ourselves to building the ultimate platform for gamers and creators and today, we continue that journey. But there's one more thing, the RTX 3090 Ti, a monster GPU, 40 shader teraflops, 78 RT teraflops, 320 tensor teraflops, 24 gigabytes of 21 gigabit per second G6X memory, the fastest ever, the RTX 3090 Ti, our next BFGPU. Tune in later this month for more details. Now let me hand it over to Ali Kani.

Thanks, Jeff. I'm here to talk about autonomous vehicles. It's perhaps the most intense AI challenge, but it's also one with the greatest benefits to society. It will save lives, make roads less congested and change the way billions of people move around the world. [Presentation]

There are 3 core pillars of the machine learning pipeline being developed at NVIDIA to create automated driving solutions that can be used by the transportation industry. First is the real-time AI-based software stack running on an efficient, high-performance supercomputer that goes inside every vehicle; second, there are AI models trained on our DGX servers in the cloud; and finally, a simulation platform is required, and we've built it on Omniverse. It's used both for synthetic data generation as well as to test and validate the AV system. This machine learning pipeline is never ending. Autonomous driving software will need to continuously develop and improve. New data will be collected and synthetically generated, models will be retrained in the data center, then validated in simulation before new software is ultimately updated for use on the road. We've created this end-to-end automotive platform and modules, so that our partners can use exactly what they need to speed time to market and build a product that can stay true to their brand. We have some partners that just buy our chips and core operating system while developing their own software applications. Other partners like Mercedes-Benz rely on us across this entire stack, from our self-driving software running on NVIDIA DRIVE computers in each car to training AI models in the cloud, synthetic data generation, vehicle validation and testing of new features through simulation, which finally get pushed over the air into each Mercedes-Benz vehicle. Developing an autonomous vehicle requires an entirely new platform architecture and software development process. Both the hardware and software must be comprehensively tested and validated to ensure they can handle the harsh conditions of daily driving with the stringent safety and security needs of an automated vehicle. This is why NVIDIA has built and made open the DRIVE Hyperion platform, which specifies a high-performance computer and sensor architecture that meets the safety requirements of a fully autonomous vehicle. Today, we are in our eighth generation of Hyperion. It has been adopted by hundreds of automakers, truck makers, Tier 1s and robotaxi companies. Hyperion 8 is designed with redundant drive Orin computers, 12 state of the art surround cameras, 9 RADAR, 12 ultrasonics, 1 front-facing LiDAR and 3 interior sensing cameras. It's architected to be functionally safe so that if one computer fails, there is a backup available to ensure that the autonomous vehicle can drive its passenger to a safe place. We have spent many years developing a data collection system with high-quality calibration and time synchronization in 4D, so our partners can leverage NVIDIA's AV investment and focus their resources on their own AI application development. We have also defined Hyperion with our standard AGX form factor so that customers can design their autonomous vehicle to scale across generations of their vehicles. With this standard form factor, an automotive OEM can go to production with a DRIVE Orin computer today and easily plug in hardware and software compatible DRIVE Atlan computer in the future. Today, I'm pleased to announce that leading automotive Tier 1 suppliers Desay, Flex, Quanta, Valeo and ZF as DRIVE Hyperion 8 platform scaling partners. Together, these partners provide the automotive ecosystem with world-class Tier 1 options who can manufacture production-ready Hyperion platforms, with the highest levels of functional safety and security. We're very proud of these partnerships as they demonstrate how Hyperion 8 has quickly become the automotive industry's most open and adopted platform architecture, and you will see it start hitting production as early as this year. The transportation industry is going through a rapid transformation. This change started with a move from internal combustion engines to electric motors. Electric vehicles are not just better for the environment. They fundamentally improve the driving experience for consumers. They drive more quietly, accelerate faster and they will last longer. Over the course of the next several decades, we will see the majority of cars sold be electric. This transformation has given way to dozens of new energy vehicle startups. They have reimagined the car, starting with a new vehicle architecture based on programmable software-defined computers. I'm very pleased to announce that our DRIVE Hyperion platform has been adopted by leading new energy vehicle OEMs in the industry. Volvo-established Polestar is bringing a storied heritage of safe driving into the modern era with electric vehicles that emphasize the joy of driving. The upcoming Polestar 3 SUV features a centralized compute architecture powered by NVIDIA DRIVE. China, now the world's largest auto market has also been very successful driving the transition to NEVs, with nearly 20% of all cars sold in China this past quarter being electric. We're really excited to work with the vast majority of the leading NEV companies in China, including NIO, XPeng, LI Auto, R Auto and SAIC's premium electric vehicle brand, IM. These new electric cars will get better and better over time with each over-the-air update. Such companies can benefit from new business models that are software driven. Rather than selling a car and seeing it degrade once it's driven out of the dealership, a tech-based automotive company manages a fleet of vehicles that will improve with new services and functions offered over the life of the vehicle. The potential of these relatively young electric automotive companies and their innovative business models is really disruptive. It has resulted in the market valuing them higher than car companies that are well established and produce millions of vehicles each year. AI is not just transforming passenger vehicles, but also opening up a major opportunity for disruption in the commercial sector. Over the last decade and accelerated by the pandemic, consumer shopping has dramatically shifted online, resulting in increased demand for trucking and last-mile delivery. This trend is expected to continue with projections starting at 170 billion packages being shipped in 2022 and increasing to 280 billion packages in 2027. Simultaneously, the trucking industry has been experiencing a significant shortage of drivers. Over the next few years, this gap is expected to widen even further, increasing to a shortage of more than 140,000 drivers in the U.S. alone by 2027 and more than 1 million worldwide. We are now seeing a large number of autonomous driving services being built to enable these opportunities that are together essential to our transportation future. Today, I'm pleased to announce that TuSimple will be building its autonomous trucking platform on NVIDIA DRIVE Orin. TuSimple is an exciting company that's meeting the challenges of today's trucking and logistics industry with its scalable autonomous freight network. TuSimple is already working with the leading names in delivery, including UPS, Navistar and Penske, and its technology has already improved arrival times for long-haul routes of the U.S. Postal Service. Now I mentioned earlier that developing AV is a grand AI challenge. At the heart of AI model development, we need the ability to collect a lot of data and accurately label it to allow our AI engineers to build accurate and safe AI models. Labeling real-world data can be time-consuming, costly and difficult. In fact, often, the data we are most interested in using is also data that's the most challenging to label. Here are some examples from our own data set that are difficult to label, including scenes that are dark, blurry or hazy. Others contain scenes that are challenging to understand like occluded vehicles or pedestrians. We also have many scenes that are not often encountered like construction zones. These issues are the reasons why AV developers need a mix of real data and synthetic data when developing their AI models. Because of all these labeling challenges, NVIDIA has built DRIVE Sim Replicator, a synthetic data generator for autonomous vehicle development. Replicator helps our AI engineers to build up hard to label ground truth data by synthetically generating them from virtual cameras, LiDAR and RADAR sensors within our Omniverse simulation platform. In this way, engineers can train AI models even before any real data has been collected. And Replicator can label ground truth in ways that humans cannot, tracking moving objects across sensors, velocity, distance, occlusion and severe weather conditions. It's a really powerful tool for the AV developer. It's accurate, low cost, and it fills in gaps for data not easily found in the real world. The industry is just getting started on its path to truly self-driving technology. Most vehicles today are only able to deliver Level 2 driving, commonly referred to as Advanced Driver Assistance Systems that require drivers to still be fully attendant. We are now starting to see cars classified as Level 3 that allow drivers to take their eyes off the road but still remain in the driver's seat. Some companies are also developing Level 4 and Level 5 solutions that will be truly self-driving, allowing the driver to become a passenger and even sleep in the back seat. These include autonomous last-mile deliveries, hub-to-hub trucking and passenger robotaxi. The complexity of AV software development is ever increasing for these different use cases. Vehicles targeted for higher levels of autonomy consume much more data as they're designed with more sensors, each with higher resolutions for better accuracy. AI networks and software architecture also increasing complexity as you require true self-driving. All of these factors result in a large increase in data center infrastructure required for self-driving software development. Companies investing in basic L2 systems can have data centers that need just 1,000 or 2,000 GPUs. And then we see some companies needing upwards of 25,000 GPUs when developing full self-driving systems. Now our strategy at NVIDIA is unique to the industry. We help our partners with computers and software in the car. We also help them with their infrastructure in the cloud. And when our partners align to our DRIVE Hyperion architecture, we can help them connect their platform in the car with their infrastructure in the cloud seamlessly. We believe the speed of an OEM's end-to-end autonomous driving software development flow is the most critical asset and differentiator. Software-defined vehicles will incorporate 2 key computers built with NVIDIA DRIVE Orin. So far today, I've talked about the computer for autonomous driving, the DRIVE Chauffeur and how it is an AI-assisted driving platform that can handle both highway and urban traffic with the utmost safety. Using the high-performance compute architecture and sensor set of DRIVE Hyperion 8, DRIVE Chauffeur can drive you from address to address. But there's another equally important computer that is needed for a next-generation software-defined vehicle, and we call it DRIVE Concierge. With DRIVE Concierge, vehicle occupants have access to intelligent services that are always on using NVIDIA Omniverse Avatar for real-time conversational AI. [Presentation]

Omniverse Avatar connects speech AI, computer vision, natural language understanding, recommendation engines and simulation. The technology of Omniverse Avatar enables DRIVE Concierge to serve as everyone's digital assistant, helping them with recommendations to make reservations, safely use mobile devices and provide alerts, like if a purse is left behind in the vehicle. It's personalized to each driver and passenger, giving every vehicle occupant their own personal concierge. [Presentation]

DRIVE Concierge also needs to be tightly integrated with DRIVE Chauffeur to provide high-quality 360-degree 4D visualization inside the vehicle. So the driver can comfortably sit back and trust the Chauffeur to safely drive. DRIVE Chauffeur and Concierge together can do really beautiful things. As an example, they can serve as a valet to let you out at the entrance of your destination and enable the car to search for a parking spot on its own. And then when you're ready to leave, you simply summon your Concierge who will ask your Chauffeur to drive your car back to you. From AI in the car to AI in the cloud, NVIDIA is paving the way to safer and more efficient transportation. Now we're still in the early stages of this exciting journey. And I can't wait to see what our partners develop and deliver to consumers on our Hyperion 8 platform starting later this year and beyond. Thank you.