NVIDIA GTC Automotive Special Address

0:09 Good afternoon, everybody.

0:11 Welcome to GTC.

0:13 Hopefully, you are enjoying it.

0:15 And first of all, I want to say, what a year, especially for us AV developers.

0:23 In 2025, I would say that it's also

0:26 the chat GPT moment for autonomous driving technologies.

0:31 Mostly because of the rapid progress of AI technologies.

0:37 We have VLA, vision language action models,

0:40 and we have word model foundation models, and we also have reasoning models.

0:47 With these technologies, I would say over the last few years,

0:51 we have been seeing the way we develop AV have been fundamentally changed.

0:59 Sorry.

1:03 Good timing.

1:10 On the other hand, these changes didn't really slow us down, though.

1:17 And on the contrary, we are seeing, you know,

1:21 the whole industry is moving much faster towards full autonomy.

1:26 And at this moment, I would say that the path

1:31 to full L4 level autonomy has never been this clear.

1:37 And also the mission to make everything

1:40 moves autonomous have never been this real.

1:45 And NVIDIA actually have been working on this problem.

1:48 I would say the autonomy challenge for a long time,

1:52 for more than a decade actually.

1:54 I remember 10 years ago, we produced our first,

1:58 I would say, ADAS device or ADAS chip which is called, sorry.

2:08 which is called the Parker,

2:10 and that device actually enabled the first generation of autopilot.

2:17 And 10 years from that, today,

2:20 when we're looking at a pivotal moment of AV industry,

2:25 obviously NVIDIA is working super hard,

2:28 my team is working super hard with the rest of the industry

2:34 and the ecosystem to find a path towards scalable and a safe level for autonomy.

2:45 And in this presentation, I will just walk you through what we have been doing.

2:53 To kick off the presentation,

2:56 actually I would like to first show a video which is what my team did recently,

3:04 a drive from San Francisco around the North Bay based

3:10 on a reasoning model in a prototyping vehicle from our partner, Mercedes.

3:16 All right, let's get the video rolling.

3:20 Today we're going to experience the reasoning capabilities

3:23 of the Alpameo model that powers the intelligence in our cars.

3:27 Throughout the drive,

3:28 you'll see the car is constantly evaluating its surroundings in real time.

3:35 You can also ask the car for information about your surroundings.

3:39 Hey Mercedes, why is the Transamerica Pyramid shaped like that?

3:44 Architect William Pereira chose the tapered design so more sunlight

3:47 could reach the streets below instead of casting a huge shadow.

3:52 We've also asked the car to narrate its thinking out loud today,

3:55 so you'll hear it call out what it's observing and why.

3:57 I'm changing lanes to the right to follow my route.

4:14 Alpamayo is continuously reasoning in the background,

4:16 so that when something unexpected happens, the car is already ready to act.

4:20 I want to turn left, but there is oncoming traffic, so I'm waiting for a gap.

4:30 There's a double parked vehicle in my lane.

4:33 I'm going around it.

4:37 There's a pedestrian in the crosswalk ahead.

4:39 I'm going to yield.

4:48 There's a vehicle crossing my path ahead.

4:50 I'm going to yield.

4:56 Thank you for watching.

5:03 What sets Alpamayo apart is how it reasons through complexity,

5:07 breaking down edge cases before they become problems.

5:17 Hey Mercedes, can we speed up?

5:20 Sure, I'll speed up.

5:25 Hey Mercedes, make a lane change to the right.

5:30 Sure, initiating lane change.

5:34 You know, I think maybe we should go to Sausalito to grab a coffee.

5:37 Hey Mercedes, can you take the next exit please?

5:41 Sure, turning right ahead.

5:56 Hey Mercedes, pull over here.

6:00 Sure, I'll find a safe place to drop you off.

6:11 Thank you for watching.

6:18 From dense city streets to highways and bridges,

6:21 in both real and simulated environments, Alpameo reasons through it all,

6:25 continuously building a safer, smarter driving experience.

6:37 There's a pedestrian in the crosswalk ahead.

6:39 I'm going to yield.

6:44 Hey Mercedes, can you pull over here?

6:49 Sure, I'll find a safe place to drop you off.

7:04 All righty, hopefully this video, thank you.

7:09 I think my team does deserve a round

7:11 of applause because they have been working super hard.

7:14 And hopefully, towards the end of this year,

7:18 we can really have the customers and other folks

7:22 to be able to experience this in a real car.

7:28 So, hopefully this video, again, gives us a glimpse of the future,

7:34 how a reasoning model can be used to tackle the autonomy task.

7:40 And this slide, obviously,

7:42 I believe Jensen has been using the slides in a few of his keynotes,

7:48 and, you know, this shows the journey we

7:51 have been through in the AI technology development.

7:56 With LXNet in 2012, I think, and then Transformer in 2017,

8:01 we have been seeing perception AI evolve into generative AI,

8:06 and with the reasoning model, which is happening right now,

8:11 We are introducing agentic AI,

8:13 and my team and obviously the rest of the industry are

8:18 working on also putting this AI technology into a physical embodiment,

8:22 which is the physical AI.

8:25 And we believe that building and safely deploying

8:28 physical AI is going to be the defining

8:31 challenge of the coming decade and Since automotives

8:36 are just basically driving robots on the road,

8:41 I'm pretty sure AV will be the first

8:45 mass-produced and mass-deployed physical AI technology and applications.

8:52 And why is autonomous driving so important?

8:56 I think there's a few numbers here.

8:59 So the most important number here is actually the middle one.

9:04 Every year, across all different segments, roughly speaking,

9:09 there's 13 trillion miles being driven by a car.

9:16 Or a delivery motorcycle, a delivery car.

9:24 And obviously there's other use case as well.

9:27 And as of today, only 0.006% of the mileage are driven autonomously.

9:35 And we do believe that in the near future,

9:40 every mile, everything that moves will become autonomous.

9:46 And this gap is the huge opportunity in front of all of us,

9:50 even for the size of NVIDIA.

9:57 And to tackle this challenge,

10:00 as I said in the beginning, at this pivotal moment,

10:05 we are also pivoting our main, basically, product,

10:09 which is NVIDIA Drive, to a slightly different form.

10:17 As Jensen mentioned, physical AI problem is a three computer problem.

10:23 So in Drive AV, in NVIDIA Drive, we have really three computers,

10:28 the training computer and the simulation computer in the cloud,

10:32 and the inference computer in the car.

10:36 And we are going to build five layers of services on these three computers.

10:41 And on the bottom layer, obviously, it's our hardware.

10:45 And here, it's the Drive Hyperion.

10:48 And it provides a unified reference design for computer and sensors.

10:56 And on top of that, it's the operating system and the platform software.

11:01 This is what we call the Halos.

11:05 Halos provides the middleware that interacts the model and our hardware,

11:12 and also in the future we want Halos to be able to provide safety

11:20 guardrails for anybody who wants to deploy

11:24 an end-to-end model on our Hyperion hardware.

11:30 And on top of that, we have the model layer,

11:33 where we use, I guess you guys have heard about AppMio,

11:37 and we have open-sourced a version of AppMio,

11:40 and we will keep developing on that.

11:44 And on top of that, it's applications.

11:46 This is where DriveAV come into play.

11:50 And we are working our partners to deliver from partial

11:55 to full autonomy in our basically in-house full stack.

12:01 And, last but not least, infrastructure is very important in the age of AI.

12:08 We are building a bunch of foundation models and tools in the cloud to help

12:15 developers to develop efficient models to be able

12:20 to run our hardware and the full stack.

12:28 Let's get into a little bit into the details.

12:31 And as I said, Hyperion is a unified vehicle reference architecture towards L4.

12:37 And as a call of it, obviously,

12:41 it's a L4 ready, you know, computer architecture,

12:44 which is based on a dual, basically, AGX SOAR,

12:49 which is our flagship device of current generation for in-car SOC.

12:56 And on top of that, we are also providing

12:59 a reference design for the sensor suite in Hyperion as well.

13:05 And this is very important because only

13:08 when you have a unified sensor architecture,

13:12 the data will become shareable across different OEMs, different platforms.

13:17 And on top of that, we are also working with our partner,

13:22 Uber, to build a fleet of basically Hyperion,

13:25 the current generation called Hyperion 10,

13:28 Hyperion 10 sensor set and to do data collection

13:31 and give a sort of jumpstart to the whole industry.

13:35 I think this is very important of our strategy as well.

13:42 A little bit more regarding our current flagship device, Ajax SOAR.

13:48 This is really a in-vehicle AR supercomputer.

13:52 It's based on the Blackwell architecture, which is the same architecture as most

13:57 of the company are using to train the foundational models.

14:02 And it supports basically, you know,

14:05 obviously different positions, starting from 32 to 16.

14:09 It also supports, you know, floating point 4, 4-bit floating point,

14:14 which is very important because for in-car devices,

14:17 you are always limited by memory.

14:21 memory bandwidth, and this can

14:23 effectively increase your in-car computer capability.

14:28 And as compared to the previous generation, if we can, you know, use, you know,

14:33 FP4, that's really basically a 20x boost

14:36 in terms of performance as compared to Orion.

14:42 And the census suite.

14:44 Again, to make a census suite common is very important,

14:48 as I said, from a data-shareable perspective.

14:53 And we have two models.

14:56 On the right side is the high model,

14:58 which we believe will be able to sufficient for a high-level economy,

15:03 including highway and urban L3 and L4.

15:06 And right now, this sensor suite has 14

15:11 high-definition cameras and up to four internal cameras,

15:15 and it has nine radars and one LiDAR.

15:19 This sensor is designed to have full sensor redundancy,

15:23 so that whenever you have a single point of failure,

15:26 the system will still be able to operate safely.

15:32 On the left side is our base model,

15:36 and this is obviously more focused on the cost, you know,

15:41 saving in terms of at a bomb level, but with 10 cameras and 3 radars,

15:48 this system will be able to deliver a full, hand-free...

15:56 Highway and urban, address-to-address,

15:57 even park-to-park capability, which is, you know,

16:01 what we are building on with our customers in DriveAV.

16:11 And on top of Hyperion, we build Halos.

16:17 Halos, as I said, is a unified software safety foundation for L4.

16:23 It actually have multiple layer of software, you know, defined within Halos.

16:28 At the bottom layer, it's our operating system,

16:31 which is certified to the highest ASO level, which is ASO Delta.

16:38 And this is based on the legacy DriveOS, basically software,

16:43 and DriveOS will become part of Halos moving forward.

16:49 And it provides a library, including CUDA, TensorRT, and LMSDK,

16:54 so that we can enable developers to deploy large language

17:00 models and AI models for AV in the most effective way.

17:07 And it also supports both Linux and the QNX, basically, operating system.

17:14 And on top of that, it's middleware.

17:17 So middleware is very important.

17:19 In our current generation, which is Orem-based generation,

17:23 we used to have a separate responsibility between NVIDIA and OEM.

17:29 And OEMs actually have to spend hundreds

17:32 of engineers to define this sensor and vehicle, let's say, abstract layer.

17:38 And moving forward, in Halos, as part of Halos operating system,

17:44 Halos OS layer, we will take more responsibility.

17:49 And because much of this software is generalizable across different OEMs.

17:54 We can significantly reduce the engineering burden on the OEM

17:58 side to be able to deploy their stack on Hyperion.

18:04 And on top of that, in the future, in the SOAR generation,

18:09 we also want to introduce, as I said, the safety guardrails for the system.

18:16 This will include our five-star EN-CAP and N-CAP active safety software,

18:22 and it will have our classical stack to serve as a guard rail.

18:30 So in the future, hopefully for every AV developer,

18:34 They can, you know, put their own end-to-end model on halos,

18:39 you know, and the benefit from the safety

18:44 foundation we have built across different layers within halos.

18:55 Now let's go to the model layer,

18:58 and today I'm very proud to announce that we are going to release our Mario 1.5.

19:06 And our Mario 1.0 was released only, I think,

19:09 two months, slightly more than two months ago.

19:13 And it was a 10 billion parameter model, which is also the first.

19:21 Recently model for AV and over the last two months it was actually number two

19:29 in the most downloaded application or model

19:33 in the whole robotic space on Hugging Face.

19:38 It was downloaded more than 160,000 times.

19:43 I didn't even know there are so many people working on AV.

19:47 But, you know, we are very happy about, you know, what we, you know, the result.

19:53 And in our Pomayo 1.5, it's still a 10 billion parameter model,

19:57 but we are going to make it more powerful.

20:01 So what are the changes?

20:03 Number one, we are going to introduce

20:06 routing capability into the model in different formats.

20:09 It can be waypoints or it can be a nav guidance from a, you know, let's say,

20:17 a navigation software or in other some, let's say, vectorized format,

20:22 then the model will be able to follow the route.

20:28 And the number two, we are also

20:30 introducing text prompts capabilities as you what

20:32 do you have seen in the in the in the video in the opening

20:37 video The driver or the passenger can ask what the car is trying

20:41 to do And we are also introducing multiple configurations in the camera support.

20:50 It can support one or two or four cameras

20:53 with different FOVs and placements and so on and so forth.

20:57 We believe that with this model, it can hopefully add one more capability.

21:07 One more capable model to our ecosystem and help the industry,

21:12 you know, build more advanced models.

21:16 And as a matter of fact, you know, our Mario 1.5,

21:20 you know, ranked number one in the open dataset lingo QA.

21:32 This is going a little bit behind the scenes regarding how we train our model,

21:36 the open source model.

21:38 And we trained our model with 80,000 AV data.

21:44 Maybe it doesn't sound too much,

21:46 but we are not training our Mario model from scratch.

21:50 Our Mario model was actually the backbone

21:53 was taken from the Cosmos Reason model,

21:56 which was obviously part of the Cosmos Open model family.

22:02 And Cosmos, the foundation model, was trained with internet-scale data,

22:07 with 20 million hours of real-world data.

22:11 And then on top of that, there's basically a Cosmos predict transfer

22:15 and then the Cosmos reason model was trained on top of that.

22:20 So this pyramid shows us basically how much data

22:24 the actual model of Mario inherit from the foundation model.

22:30 And hopefully it will also show great

22:32 generalization capability in the field as well.

22:39 However, taking the Mario open source model to production

22:43 will still take quite a bit of work.

22:48 Hopefully, this is not very surprising to anybody here.

22:52 So, to build a production-ready, basically, model,

22:55 you have to make sure the model is

22:59 fine-tuned to work well in all different scenarios.

23:03 And this is a typical pipeline,

23:05 which is what we use for internal effort as well,

23:07 which I'm going to show you in a later slide,

23:10 regarding how to fine-tune the out-model.

23:13 and get it ready for production.

23:15 So you would need lots of data set,

23:18 obviously, either through data collection or through synthetic data,

23:21 which Cosmos can actually help over there as well,

23:24 which is also open source to public.

23:27 And...

23:30 In the middle, obviously,

23:32 you have to prepare the data set and train the model and then basically when

23:38 you release to the car and then basically you have to get the, let's say,

23:45 the failure mode from the road and basically

23:49 feedback the data to the model training.

23:52 So, the tools are very important here.

23:55 One is basically, obviously, when we prepare the dataset,

23:58 you need a good search and curation capability.

24:02 And also, on the right side, you know,

24:05 simulation and the validation is super important in the gen AI,

24:09 you know, kind of way of developing AV.

24:14 And so, To help, you know, again, the ecosystem to be able to train the model,

24:23 so we are not only open sourcing the model itself,

24:28 we are also open sourcing data, which is right now, at this point,

24:33 it's 7,000 hours of high-quality data,

24:36 which we collected in the driver AV development phase.

24:41 To the best of my knowledge,

24:43 this is by far still the largest data set, open source data set out there.

24:47 And this is probably not enough for training,

24:50 but this data has good, basically, diversity.

24:52 It's collected into, you know, in 25 countries.

25:00 And it will be very well suited for evaluation and testing.

25:07 And we are also, you know, as I said, tooling is very important.

25:11 You need good tooling for search and curation,

25:14 and you need good tooling for validation and assimilation.

25:19 And towards this end, we are also basically make, you know, Neurac,

25:24 which is neural reconstruction, as a tool to be available to public soon.

25:31 Not yet, but in the next few weeks.

25:34 And the Neurac is a super powerful, basically, tool,

25:38 which is absolutely needed for developing AV stack,

25:41 especially the end-to-end model.

25:43 So this video.

25:46 Hopefully, I haven't clicked it yet.

25:48 Okay, so this video shows the, I guess, the cameras.

25:54 And you can see, you know, we can really reconstruct the pixel when

25:58 the vehicle is changing a different pose.

26:01 This is very critical for any, basically,

26:04 closed-loop evaluation of the stack of all of the end-to-end model.

26:10 So this...

26:12 This technology is pretty mature.

26:15 It's getting more and more mature as we speak as well,

26:19 partly because it's massively utilized

26:21 in our internal effort to drive AV development.

26:25 We are running two million simulation tests based

26:29 on NeurIQ on a daily basis to develop our stack.

26:33 So there's a, I guess, famous phrase, we are eating our own dog food.

26:40 And I'm very confident about the quality of this tool,

26:44 and hopefully this can help the community.

26:52 Well, Nurex is not only about basically doing just replay,

26:56 you know, different trajectory and reconstruct the pixel.

27:00 You can do more about that.

27:02 And as part of the, you know,

27:05 Nurex open-source effort, we also open-sourced Fixer.

27:10 These are two Gen AI models.

27:11 One is Fixer.

27:13 The other one is basically Harvester.

27:16 The Harvester can actually harvest any object you see

27:19 from the field and save it for other use.

27:23 For example, you can plant it in other videos.

27:27 So this slide shows how we can leverage this capability,

27:30 not only use NeurIQ for replay,

27:32 but also use it to insert and modify behaviors of the object.

27:37 By doing this, we can increase the data diversity or variety by auto-magnitude,

27:42 you know, from the data we collect from the field.

27:48 So, there's a few examples.

27:50 I will just quickly click through them.

27:54 Essentially, you can, you know, obviously you can just do the reconstruction.

27:58 You can also basically insert a scooter here.

28:01 I'm not sure if you guys can see.

28:03 I cannot see very well.

28:04 You know, the screen in front of me is extremely small.

28:08 And then, basically, you can also modify the trajectory when this kind

28:12 of object shows up and then feed it back,

28:16 train a different model, and deploy it to the car.

28:22 And the other one which is extremely

28:25 important and extremely powerful is Cosmos Transfer.

28:29 Again, Cosmos Transfer, as I showed you guys briefly in the data pyramid,

28:34 is a part of the basically Cosmos family.

28:37 And what it does is basically you

28:40 can render any sequence into a different environment.

28:43 And is there some example here?

28:46 Can we go back to the previous slide, please?

28:49 Yeah, so there's a multiple example here.

28:52 Obviously, you can just add a prompt of different weather,

28:57 different lighting, and even in different cities,

29:00 in different essential terrain,

29:02 so this is, again, another way to bring a lot more

29:07 variety and diversity into a data set and train the model.

29:15 All right, last but not least, let me spend time talking about the driver AV.

29:22 And we have a pretty sizable team working on driver AV right now.

29:27 And obviously, NVIDIA is pretty well known

29:30 to work with the industry to provide AV SOCs.

29:34 As I mentioned, 10 years ago, Parker enabled the first version of Autopilot.

29:40 And the next generation, basically Xavier and Orin, they really...

29:48 started the massive transform or bring up of you know AV

29:52 technology in China and the rest of the world and right

29:57 now basically at least in the China market you cannot really

30:01 sell a car new car without a good AV capability in it.

30:08 And we are obviously we announced the SOAR as well.

30:12 It has been in the market for a few

30:15 years and you know actually there's already OEMs,

30:18 you know taking SOAR to to SOP And we are also seeing very good trend

30:24 in the rest of the world market

30:27 as well of SOAR adaptation adaptation and SAW is,

30:32 as I said, part of the Hyperion,

30:35 basically, ecosystem, Hyperion platform and, you know,

30:38 for everybody who's adopting Hyperion,

30:40 we'll use SAW as an, you know, in-car AI supercomputer.

30:46 It's probably less well known that NVIDIA worked

30:49 on AV stack also for a long time.

30:53 DriveAV was kicked off 10 years ago as well.

30:57 And in 2017, actually, we had our internal,

31:01 which was a little bit ahead of its time,

31:05 I would say, end-to-end model-based approach, basically,

31:08 for AV, which is called a pilot net.

31:12 And then, with many years of progress,

31:17 we secured a partnership with Mercedes in 2020, which was five years ago.

31:26 And then, as the JLR happened and Lucid happened, and I cannot say much yet,

31:34 but we'll have more OEMs getting into the DriveAV ecosystem.

31:41 And this year, no, last year, 2025 was the first year,

31:46 after five years of partnership,

31:48 we deployed our technology to all Mercedes vehicles globally, except China.

31:55 And we are expecting to roll out our address-to-address

32:00 L2++ capability throughout this year in Mercedes vehicles,

32:04 which is gonna be super exciting.

32:08 And we are also working on, as we speak, JLR and Lucid customer fleet as well.

32:19 Our driver AV stack is actually a hybrid end-to-end classical stack.

32:25 This is very important because we get

32:28 the human-like drive experience from the Android model.

32:31 This is very well known.

32:32 But the whole industry has been developing AV

32:37 and ADAS based on basically FUSA and ISO 26262 standards.

32:43 And this kind of interpretable safety is very important to the industry.

32:47 So having a hybrid stack, you can achieve both.

32:51 Number one, you can still have human-like driving experience.

32:56 And meanwhile, without compromising, basically, the safety interoperability.

33:04 So this...

33:06 Slightly busy diagram shows our stack.

33:09 What's important here is, you know, our Mario,

33:11 which is the production version of Mario,

33:13 is our end-to-end model running the stack.

33:15 By the way, we are running everything here in one ORN.

33:19 And basically, this is a LIDAR list and a mapless stack.

33:23 And on the right side, it's our classical stack.

33:26 And on top of that, basically, we have a safety arbitrator,

33:31 which can guarantee safety, you know, when the stack is running.

33:38 And with this stack, obviously, we can, we are delivering parking,

33:43 we are delivering active safety, and we are also delivering advanced,

33:48 you know, AV features, including Highway L2+, with handle-free capability,

33:53 and L2++, which is the address-to-address or park-to-park in urban scenarios.

34:05 And this slide shows a little bit how we

34:09 take our model to production in the internal DriveAV effort.

34:14 It's very similar to the slide I showed before, and we're iterating super fast.

34:21 We have iterated, basically,

34:22 we started developing this end-to-end model about a year ago,

34:26 and now we have iterated 3,500 times.

34:28 And roughly speaking, we are iterating 10 versions every day.

34:39 And basically, we are also using, obviously,

34:42 a sizable GPU cluster to be able to iterate this fast.

34:49 And also, as I mentioned,

34:52 simulation is very important for this end-to-end model development.

34:59 And right now, basically,

35:02 our technology has been enabled from Test in the field until a model is ready.

35:09 Again, every new rack is a model.

35:12 You have to build this model and use

35:15 it in replay and the closed-loop evaluation.

35:18 Right now, we can contain the whole

35:20 hours within turnaround time within six hours.

35:22 And we believe that we can take it down to a few hours in the near future.

35:29 So, again, this is a massive effort, and the middle block is,

35:33 you know, is similar to the AlpaSIM we open sourced to public,

35:38 but this one is slightly more because it have to be

35:42 able to do a closed-loop evaluation of a hybrid stack.

35:47 So you can run a classical stack and the Android

35:50 model and use this to get a good evaluation result.

35:57 Well, this one explains roughly how

36:00 guardrails work with a classical stack running,

36:04 again, side by side by the end-to-end model.

36:08 We have a cyclist right on the side,

36:11 and because his shoulder is ending, to end up basically coming to the road.

36:18 So, when it's green, it shows how Mario was running it.

36:20 When it's classical, the arbitrator changed the model to classical.

36:25 Essentially, it will say, okay, this current trajectory is not safe enough.

36:31 The classical, you know, trajectory is better.

36:34 Then it will seamlessly, almost like with zero delay, it can transfer,

36:39 it can transit, I would say, to a classical trajectory and to maintain safety.

36:49 And as we speak, we just started basically,

36:52 you know, worldwide scaling as well for our L2++ stack.

36:56 And this is showing we are basically

36:58 driving pretty much everywhere in the world.

37:01 And we expect that by the end of this year,

37:03 we should be able to scale the technology to all of the U.S.

37:08 and a few cities in Europe.

37:11 And the next year, we want to basically make it, you know, available globally.

37:19 And another important point to hit is basically our current

37:24 software structure is actually very much scalable or extendable to L4.

37:29 So this slide shows what the architecture will

37:32 look like when we enable this kind of capability, our stack, in the L4 setting.

37:38 You will have a main ECU.

37:41 And you also have a satellite ECU.

37:43 In the main ECU, it's pretty much running

37:47 the L2++ stack we are running right now.

37:50 But on top of that, it will take additional sensors

37:54 from the satellite ECU to make the main ECU capability even better.

38:00 And on the satellite ECU, this is our safety redundant stack.

38:05 It's mostly going to be classical based.

38:08 But it will also have the MRM or minimum risk maneuver when basically,

38:14 you know, a single point of failure happens.

38:19 So in this way, basically we can use the same

38:23 code repository and delivering L2++ and L4 together to our partners.

38:32 All right, this is our high-level roadmap.

38:37 As I said, basically, this year,

38:40 we are trying to scale L2++ from address to address.

38:44 And starting from next year,

38:46 we want to start the initial deployment of our L4 fleet.

38:51 And by the end in 2008, You know,

38:56 we will basically deploy L4 in passenger vehicles with our partners.

39:07 And basically, I want to, this is a, you know,

39:11 Hyperion is, you know, catching a lot of attention,

39:15 I would say, and, you know, we are having new OEMs joining,

39:20 basically, Hyperion ecosystem almost, I would say, every month.

39:24 And, you know, we have four new OEMs joining the Hyperion ecosystem,

39:30 BYD, Geely, Nissan, and Hyundai.

39:34 All of them are, you know,

39:37 fall of the top 10, basically, OEMs globally, and again,

39:41 we see great benefit of, you know,

39:45 having a common sensor suite and a compute platform across different OEMs.

39:53 And we also tried, I'm very happy

39:56 to announce our partnership with Uber towards L4.

40:01 And by 2028, we agreed to deploy L4 technologies in 28 cities globally.

40:12 We are also basically, obviously NVIDIA is not only an AV company, you know,

40:19 of course we are not only a chip company, we are a full stack AI company,

40:26 and we are using our AI capabilities to work

40:30 with OEM globally on many different areas as well,

40:33 including basically using Omniverse Digital Twin

40:36 for design and manufacturing and enterprise.

40:43 And this is the ecosystem we are working with.

40:47 Autonomy is a daunting task.

40:49 It's a grand mission, but it really needs everybody to work together.

40:54 And we are very happy to work with everybody in the automotive,

40:59 basically, industry, including OEMs, tier ones, or AV software developers,

41:03 to work together to get from 0.006% to upper 90s.

41:07 This is the mission.

41:14 All right, I think I'm running a little bit out of time,

41:18 but this is the key takeaways of what I talked about today.

41:22 Again, Drive AV is working, sorry,

41:25 NVIDIA Drive is getting into a new phase, three computers.

41:31 Five layers of service, and we are trying to basically, on the Hyperion side,

41:38 we're trying to get as many as possible EMs

41:42 to share the same sensor in a computer platform,

41:46 and we are also trying to work with Uber

41:51 to create a large data set for the ecosystem developer AV.

41:58 And we are bringing halos, we are enhancing halos significantly from just

42:05 the operating system to basically a safety

42:09 guard rail so that everybody can deploy end-to-end or AV stack on Hyperion,

42:17 our hardware platform.

42:20 And we announced Alpamario 1.5, which is the next generation of Alpamario

42:25 and also basically on the application side,

42:28 NVIDIA is working super hard to take Drive AV to the next step,

42:34 which is really basically this year it's about scaling

42:39 of L2++ and also get the technology extended to L4.

42:46 And on the infrastructure side,

42:49 because simulation and the validation with AI technology is super important,

42:54 we are also open sourcing, you know, our foundation models and also new rack

43:00 and the Cosmos transfer models so the industry can help,

43:05 can use to enhance their data set and accelerate their AV development.

43:11 So this is what I want to talk about today.

43:14 Thank you for your time, and hopefully you can enjoy the rest of the afternoon.