1 00:00:00,080 --> 00:00:03,600 NVIDIA just released an open source AI model for self driving 2 00:00:03,600 --> 00:00:06,560 cars. At CES 2026, Jensen Wang 3 00:00:06,680 --> 00:00:10,920 unveiled Alpa Mayo, a 10 billion parameter reasoning model that 4 00:00:10,920 --> 00:00:14,000 can explain why it makes every driving decision and the first 5 00:00:14,000 --> 00:00:17,880 production vehicle to use it will be the 2025 Mercedes-Benz 6 00:00:17,880 --> 00:00:21,640 CLA, launching in the United States in the first quarter of 7 00:00:21,640 --> 00:00:23,920 2026. It's a real You can buy it in a 8 00:00:23,920 --> 00:00:27,960 few months with Nvidia's full self driving stack built in. 9 00:00:28,040 --> 00:00:32,240 Jensen Wong called this the ChatGPT moment for physical AI, 10 00:00:32,240 --> 00:00:35,360 the point where machines begin to understand, reason and act in 11 00:00:35,360 --> 00:00:36,920 the real world. And the question is whether 12 00:00:36,920 --> 00:00:40,840 NVIDIA just leapfrog Tesla in the race to autonomous driving 13 00:00:40,840 --> 00:00:44,720 by giving every other automaker access to technology that Tesla 14 00:00:44,720 --> 00:00:47,200 has spent a decade developing in secret. 15 00:00:47,240 --> 00:00:49,840 Now, the technical specs are pretty good. 16 00:00:49,920 --> 00:00:53,800 Nvidia's platform runs on two Drive 4 chips delivering over 17 00:00:53,800 --> 00:00:58,840 2000 teraflops of compute power, roughly 1000 trillion operations 18 00:00:58,840 --> 00:01:01,080 per second. Tesla's Hardware 4 is believed 19 00:01:01,080 --> 00:01:05,360 to be in the few 100 tops range and mobilize upcoming a key 20 00:01:05,360 --> 00:01:09,760 ultra is rated at 176 tops. Nvidia's offering five to six 21 00:01:09,760 --> 00:01:13,080 times more processing power than anyone else on the market. 22 00:01:13,160 --> 00:01:16,560 They up IO model weights are available on hugging right now. 23 00:01:16,560 --> 00:01:19,200 You can check them out yourself and the simulation framework is 24 00:01:19,200 --> 00:01:21,800 on GitHub. NVIDIA released 1700 hours of 25 00:01:21,800 --> 00:01:24,240 real driving data for anyone to use. 26 00:01:24,320 --> 00:01:26,880 And in this episode we're going to be talking about what the 27 00:01:26,880 --> 00:01:29,560 demo looked like, who's partnering with NVIDIA and why 28 00:01:29,560 --> 00:01:33,240 this open source approach could change everything. 29 00:01:33,280 --> 00:01:36,440 And we're right into that Right after this very short break, 30 00:01:36,440 --> 00:01:39,480 NVIDIA demonstrated the system live in San Francisco. 31 00:01:39,480 --> 00:01:42,440 The transportation editor from The Verge got a ride in a 32 00:01:42,440 --> 00:01:46,400 Mercedes-Benz CLA equipped with Nvidia's Level 2 Plus system. 40 33 00:01:46,400 --> 00:01:49,040 minutes to the city traffic and the route included delivery 34 00:01:49,040 --> 00:01:52,840 trucks, cyclists, pedestrians, four way stops, traffic lights, 35 00:01:53,000 --> 00:01:55,640 double parked cars and unprotected left turns. 36 00:01:55,720 --> 00:01:58,800 The car handled all of it without disengagements and 37 00:01:58,800 --> 00:02:01,560 according to this, there were no crashes and no hiccups during 38 00:02:01,560 --> 00:02:04,400 the entire ride. They said that NVIDIA showed 39 00:02:04,480 --> 00:02:07,520 would go toe to toe with Tesla's Full Self driving under the most 40 00:02:07,520 --> 00:02:11,160 complex circumstances. That's a significant claim right 41 00:02:11,160 --> 00:02:13,000 now. Now, the demo showed specific 42 00:02:13,000 --> 00:02:16,600 behaviors that suggests the AI is reasoning rather than just 43 00:02:16,720 --> 00:02:18,640 pattern matching. At one point, the vehicle 44 00:02:18,640 --> 00:02:20,520 approached an intersection blocked by a truck. 45 00:02:20,520 --> 00:02:23,560 The NVIDIA system for a slow to let pedestrians cross then 46 00:02:23,560 --> 00:02:25,880 executed a wide maneuver around the obstacle. 47 00:02:25,920 --> 00:02:29,040 That kind of nuanced decision making, combining caution and 48 00:02:29,040 --> 00:02:32,320 assertiveness the way a human driver would, is what the Epamio 49 00:02:32,320 --> 00:02:35,280 model is designed to produce. The AI uses chain of thought 50 00:02:35,280 --> 00:02:38,000 reasoning, which means it generates a step by step 51 00:02:38,000 --> 00:02:41,240 explanation of its logic before executing each action. 52 00:02:41,240 --> 00:02:43,600 That internal monologue is not just for debugging them. 53 00:02:43,680 --> 00:02:46,960 It's designed to help the system handle rare edge cases by 54 00:02:46,960 --> 00:02:49,840 explicitly reasoning about cause and effect. 55 00:02:49,880 --> 00:02:54,080 Tesla's FSD does not do this. Tesla's neural networks are 56 00:02:54,080 --> 00:02:57,200 black boxes that produce outputs without explanations. 57 00:02:57,320 --> 00:02:59,440 The search strategy is also different from Tesla's. 58 00:02:59,480 --> 00:03:02,040 Nvidia's platform supports cameras, radar, lidar and 59 00:03:02,040 --> 00:03:04,440 ultrasonics, and a full 360° array. 60 00:03:04,520 --> 00:03:07,680 Tesla famously removed radar from his vehicles in 2021, has 61 00:03:07,680 --> 00:03:10,400 never used lidar, relying entirely on cameras for full 62 00:03:10,400 --> 00:03:12,920 self driving. Nvidia's arguing that redundancy 63 00:03:12,920 --> 00:03:16,000 is necessary for safety, and the demo vehicle used a combination 64 00:03:16,000 --> 00:03:18,680 of cameras and radar to perceive its surroundings. 65 00:03:18,800 --> 00:03:21,840 Now this approach is safer and more robust than camera only 66 00:03:21,840 --> 00:03:25,160 systems, especially in poor visibility or when objects are 67 00:03:25,160 --> 00:03:27,240 occluded. NVIDIA has partnered with sensor 68 00:03:27,240 --> 00:03:32,000 suppliers like Eva, High C and Sony and RB to create a 69 00:03:32,000 --> 00:03:35,400 validated ecosystem of hardware that works with its software 70 00:03:35,400 --> 00:03:38,080 stack, and automakers can mix and match components without 71 00:03:38,080 --> 00:03:41,120 worrying about compatibility. There's no walled garden for 72 00:03:41,120 --> 00:03:41,920 this thing. Now. 73 00:03:42,040 --> 00:03:44,120 I've been digging through the analytics of the show. 74 00:03:44,160 --> 00:03:47,080 I've noticed 37% of you are following this right now. 75 00:03:47,160 --> 00:03:48,440 And for you, I'm forever grateful. 76 00:03:48,480 --> 00:03:50,240 Thank you so much for joining this community. 77 00:03:50,280 --> 00:03:53,160 The other 63% of you haven't hit the follower subscribe button. 78 00:03:53,240 --> 00:03:56,760 And I'm going to tell you one thing that I think is really 79 00:03:56,760 --> 00:03:58,480 important. Supporting independent 80 00:03:58,480 --> 00:04:01,720 journalists is important in this day and age because we used to 81 00:04:01,720 --> 00:04:04,440 have people that would block everything. 82 00:04:04,440 --> 00:04:06,800 We had the gatekeepers. Independent journalists like 83 00:04:06,800 --> 00:04:10,480 myself, I've been doing this for, I don't know, 6-7 years and 84 00:04:10,480 --> 00:04:14,280 covering tech, Elon Musk, SpaceX, Space Flight, anything 85 00:04:14,280 --> 00:04:15,680 like that, like any sort of tech. 86 00:04:15,760 --> 00:04:18,399 I've been doing that for the last 6-7 years and I'm going to 87 00:04:18,399 --> 00:04:20,040 continue doing it for the next 10 years. 88 00:04:20,160 --> 00:04:23,200 And all I need from you is literally one second of your 89 00:04:23,200 --> 00:04:26,080 time, which is going to help the show tremendously. 90 00:04:26,120 --> 00:04:28,440 Hit the subscribe or follow button on whatever podcast 91 00:04:28,440 --> 00:04:30,160 platform you're listening or watching on right now. 92 00:04:30,200 --> 00:04:32,040 That's going to help me out tremendously and help out the 93 00:04:32,040 --> 00:04:34,280 show and continue the growth of the show. 94 00:04:34,320 --> 00:04:37,640 We've grown so much because of you and I can't wait to go to 95 00:04:37,640 --> 00:04:40,560 this next part of the journey because of you. 96 00:04:40,560 --> 00:04:42,400 Now. The partnership list is very 97 00:04:42,400 --> 00:04:44,920 impressive. Mercedes-Benz is first launching 98 00:04:44,920 --> 00:04:48,680 the CLA with what they are calling MB Drive Assist Pro 99 00:04:48,760 --> 00:04:51,840 capable of point to point city navigation under driver 100 00:04:51,840 --> 00:04:54,960 supervision, and Huang said the CLA is safest car in the world, 101 00:04:55,000 --> 00:04:58,040 sending its Euro NCAP 5 star rating. 102 00:04:58,040 --> 00:05:00,920 Jaguar Land Rover has a multi year deal to use Nvidia's 103 00:05:00,920 --> 00:05:03,680 platform for level 4 autonomy and future vehicles. 104 00:05:03,760 --> 00:05:07,160 Lucid is in granting the full stack into upcoming models. 105 00:05:07,240 --> 00:05:10,440 Uber is back in the autonomous vehicle game, working with 106 00:05:10,440 --> 00:05:13,600 NVIDIA to potentially deploy robo taxis or automated delivery 107 00:05:13,680 --> 00:05:15,400 vehicles. The Tier 1 suppliers are on 108 00:05:15,400 --> 00:05:18,720 board too. Bosch, ZF, Magna, and Quanta are 109 00:05:18,720 --> 00:05:21,800 all building electronic controls and units around Nvidia's 110 00:05:21,800 --> 00:05:24,000 architecture. That means car companies can buy 111 00:05:24,000 --> 00:05:27,000 pre integrated hardware from their existing suppliers and 112 00:05:27,000 --> 00:05:30,120 just drop in Nvidia's system without building everything from 113 00:05:30,120 --> 00:05:32,000 scratch. I'm an open source advocate. 114 00:05:32,080 --> 00:05:35,480 I love that Nvidia's doing this and the open source approach is 115 00:05:35,480 --> 00:05:38,200 what makes this different from everything else in the market. 116 00:05:38,280 --> 00:05:40,600 Tesla's full self driving software is entirely 117 00:05:40,600 --> 00:05:42,760 proprietary. You can't buy it, you can't 118 00:05:42,760 --> 00:05:44,480 license it. They're by giving it away. 119 00:05:44,520 --> 00:05:46,920 You can't study it, we don't know how it works. 120 00:05:47,000 --> 00:05:49,080 Mobilized technology is also proprietary. 121 00:05:49,080 --> 00:05:51,920 Waymo and Cruz develop full self driving stacks, but they use 122 00:05:51,920 --> 00:05:54,280 them exclusively for their own robotaxi fleets. 123 00:05:54,360 --> 00:05:57,400 NVIDIA is the first company to release a production grade 124 00:05:57,400 --> 00:05:59,480 autonomous driving. EI is open source. 125 00:05:59,600 --> 00:06:03,240 This is a clear bid to become the kind of like Android of 126 00:06:03,240 --> 00:06:07,480 autonomy while Tesla continues to keep its FSD stack completely 127 00:06:07,480 --> 00:06:10,840 closed, similar to Apple. And the implication is that any 128 00:06:10,880 --> 00:06:14,160 automaker without Tesla's decade of in house AI development can 129 00:06:14,160 --> 00:06:17,440 now get equivalent technology right off the shelf. 130 00:06:17,600 --> 00:06:21,560 It's free, you get it, all you have to do is buy the parts and 131 00:06:21,600 --> 00:06:23,640 build this software around it and you're good to go. 132 00:06:23,640 --> 00:06:28,160 Now if Mercedes actually ships a car in Q12026 that has similar 133 00:06:28,160 --> 00:06:31,600 capabilities to Tesla's FSD, it is based on an open source 134 00:06:31,600 --> 00:06:36,400 system any automaker can buy. It could commoditize level 2 135 00:06:36,400 --> 00:06:39,120 plus autonomous systems. Advanced driver assistance would 136 00:06:39,120 --> 00:06:42,360 no longer be the secret sauce of a few companies, would be a 137 00:06:42,360 --> 00:06:45,120 feature any car maker can implement with Nvidia's help. 138 00:06:45,160 --> 00:06:48,320 That puts pressure on Tesla, which has never faced an equal 139 00:06:48,320 --> 00:06:51,200 capable rival system in production vehicles. 140 00:06:51,280 --> 00:06:54,120 It also challenges Mobileye, whose business model depends on 141 00:06:54,120 --> 00:06:56,400 being the dominant ADAS supplier. 142 00:06:56,440 --> 00:06:59,880 Qualcomm announced new AI features for cars at CES, but 143 00:06:59,880 --> 00:07:02,920 Snapdragon Ride platform is still significantly behind 144 00:07:02,920 --> 00:07:06,320 NVIDIA in raw performance and NVIDIA laid out it that 145 00:07:06,320 --> 00:07:09,560 compresses the timeline to widespread autonomy. 146 00:07:09,600 --> 00:07:13,840 By mid 2026, they expect Level 2 plus urban and Hwy. driving with 147 00:07:13,880 --> 00:07:17,360 automated lane changes, traffic light recognition, and point to 148 00:07:17,360 --> 00:07:19,960 point navigation. By the end of 2026, they plan to 149 00:07:19,960 --> 00:07:23,120 cover the entirety of the United States in terms of operational 150 00:07:23,120 --> 00:07:25,640 design domain and add autonomous parking. 151 00:07:25,640 --> 00:07:29,840 Late 2026 will bring a small scale Level 4 trial similar to 152 00:07:29,840 --> 00:07:34,360 Waymo's early ROBO taxi pilots and by 2027 NVIDIA expects 153 00:07:34,360 --> 00:07:36,400 partner robotaxi deployments to begin. 154 00:07:36,440 --> 00:07:39,680 By 2028, they're targeting personally owned Level 4 155 00:07:39,680 --> 00:07:43,240 vehicles and Level 3 hands off driveway and Hwy. driving into 156 00:07:43,240 --> 00:07:46,120 production cars. That is an aggressive schedule. 157 00:07:46,120 --> 00:07:48,480 It depends on regulatory approvals and the confidence of 158 00:07:48,480 --> 00:07:51,200 automaker partners though. But the message is very clear. 159 00:07:51,200 --> 00:07:54,520 NVIDIA is committing to rapid iteration and deployment of this 160 00:07:54,520 --> 00:07:57,440 open source software. Now, the business model is 161 00:07:57,600 --> 00:08:00,000 significant. Nvidia's automotive division 162 00:08:00,000 --> 00:08:03,560 currently generates about $600 million in revenue, compared to 163 00:08:03,560 --> 00:08:06,240 51 billion in their core AI and cloud business. 164 00:08:06,360 --> 00:08:08,640 Now, if this works, every vehicle that uses Nvidia's 165 00:08:08,640 --> 00:08:11,520 platform brings recurring revenue for chips, software 166 00:08:11,520 --> 00:08:15,520 licenses and cloud services. NVIDIA also introduced the Rubin 167 00:08:15,520 --> 00:08:18,200 Platform, a new data center system for training autonomous 168 00:08:18,200 --> 00:08:20,120 driving model. That means they capture value 169 00:08:20,120 --> 00:08:22,120 both in the vehicle, in the cloud. 170 00:08:22,240 --> 00:08:26,680 As fleets scale up, Jensen Wong said the goal is that someday 171 00:08:26,760 --> 00:08:28,920 every car in every truck will be autonomous. 172 00:08:28,920 --> 00:08:32,039 NVIDIA is working toward that future with the foundation laid 173 00:08:32,039 --> 00:08:35,600 out at CES 2026. They have made a strong case 174 00:08:35,600 --> 00:08:39,000 that will make the key architectures of that future. 175 00:08:39,080 --> 00:08:43,559 I think it's probably the best announcement at CES 2026 as far 176 00:08:43,679 --> 00:08:47,480 as a regular person goes, as far as automakers go, because 177 00:08:47,480 --> 00:08:50,280 everybody uses a car, right? And if you have ever used Uber, 178 00:08:50,320 --> 00:08:53,200 you know how important it is to get your car there on time and 179 00:08:53,200 --> 00:08:54,360 also get you to where you're going. 180 00:08:54,440 --> 00:08:57,160 But also, it's not the perfect system. 181 00:08:57,240 --> 00:08:59,920 Once people are taken out of that equation and we can have 182 00:08:59,920 --> 00:09:03,640 robotaxis and we'll have to work with those robotaxis or work as 183 00:09:03,640 --> 00:09:06,000 those robotaxis, it's not just going to be Tesla anymore. 184 00:09:06,080 --> 00:09:09,040 People will be able to compete and NVIDIA is helping them. 185 00:09:09,040 --> 00:09:11,440 So this is a strategic frame too. 186 00:09:11,440 --> 00:09:14,760 Listen to this. Jensen said that Chet TBT is 187 00:09:14,760 --> 00:09:17,280 similar to this, and he believes this is the moment that 188 00:09:17,280 --> 00:09:20,280 reasoning AI will do for robots and vehicles what large language 189 00:09:20,280 --> 00:09:23,960 models did for text. For us, the ability to explain 190 00:09:23,960 --> 00:09:26,840 decisions is not just about debugging anymore. 191 00:09:26,840 --> 00:09:28,960 Governments want this. They want to understand why an 192 00:09:28,960 --> 00:09:31,400 autonomous vehicle made a particular choice before they 193 00:09:31,400 --> 00:09:34,560 approve it for public roads. And by exposing that reasoning 194 00:09:34,560 --> 00:09:38,320 trace, NVIDIA is addressing concerns about black box AI in 195 00:09:38,320 --> 00:09:40,360 safety. Critical application regulators 196 00:09:40,360 --> 00:09:44,120 and researchers can inspect how the system drives at each 197 00:09:44,120 --> 00:09:47,520 decision, and that transparency can make NVIDIA powered vehicles 198 00:09:47,560 --> 00:09:51,040 easier to validate and certify than systems that cannot explain 199 00:09:51,040 --> 00:09:53,240 themselves. Now, the challenges are 200 00:09:53,240 --> 00:09:55,720 absolutely real here. There's a lot of work to do. 201 00:09:55,760 --> 00:09:58,640 Automakers traditionally want to control their own stack. 202 00:09:58,640 --> 00:10:02,120 Handing over the core driving AI to NVIDIA means giving up a lot 203 00:10:02,120 --> 00:10:04,800 of that control. But the presence of Mercedes, 204 00:10:04,880 --> 00:10:08,480 JLR and Lucid on board suggest a shift towards collaboration. 205 00:10:08,480 --> 00:10:11,720 If the big players can do it, if they jump on board, I mean 206 00:10:11,720 --> 00:10:14,760 Mercedes come out now. Public acceptance will require 207 00:10:14,760 --> 00:10:17,600 flawless safety records though. NVIDIA focus on safety, 208 00:10:17,600 --> 00:10:20,640 including their Halos framework for validation and functional 209 00:10:20,640 --> 00:10:23,080 safety is aiming at building trust with government. 210 00:10:23,080 --> 00:10:26,080 The competitive response will also shape outcomes here. 211 00:10:26,160 --> 00:10:27,880 And also they're building trust with the public. 212 00:10:27,960 --> 00:10:31,880 They want you and me to feel comfortable in a car with NVIDIA 213 00:10:31,880 --> 00:10:34,240 stacks. Kind of Tesla opens parts of its 214 00:10:34,240 --> 00:10:36,200 stack. Or a mobile eye undercuts NVIDIA 215 00:10:36,200 --> 00:10:39,000 on cost. The market could go wild here. 216 00:10:39,000 --> 00:10:42,040 NVIDIA also released ELPA SIM, which is an open source 217 00:10:42,040 --> 00:10:45,080 simulation framework for testing autonomous vehicles in virtual 218 00:10:45,080 --> 00:10:47,120 environments. And this is cool because 219 00:10:47,120 --> 00:10:50,000 engineers can create realistic scenarios with configurable 220 00:10:50,000 --> 00:10:53,240 traffic, whether in sensor modes that run the AI through edge 221 00:10:53,240 --> 00:10:54,840 cases, without putting anyone at risk. 222 00:10:54,840 --> 00:10:56,600 They don't need an actual car to do this. 223 00:10:56,680 --> 00:11:00,280 It's going to be training the AI on the infinite amounts of 224 00:11:00,280 --> 00:11:02,600 situations that we could hit on the road. 225 00:11:02,920 --> 00:11:04,920 Then they're going to run the AI through edge cases without 226 00:11:05,040 --> 00:11:09,000 putting any people in a vehicle. The 1700 hours of driving data 227 00:11:09,000 --> 00:11:13,440 released alongside it covers diverse graphical locations and 228 00:11:13,440 --> 00:11:16,360 conditions, including rare and complex situations that are hard 229 00:11:16,360 --> 00:11:19,120 to capture a normal testing. Like just bad drivers. 230 00:11:19,240 --> 00:11:21,120 You see them every day. You see a bad driver or you'll 231 00:11:21,120 --> 00:11:23,600 see somebody who, this is a green light and they're on their 232 00:11:23,600 --> 00:11:25,360 cell phone for a while. You have to hog at them. 233 00:11:25,400 --> 00:11:27,280 Those kind of things happen all the time. 234 00:11:27,280 --> 00:11:30,640 But you can do this virtually so nobody gets hurt, nobody gets 235 00:11:30,640 --> 00:11:33,960 mad. So by providing both a simulator 236 00:11:34,040 --> 00:11:37,600 and real world data sets and videos, enabling what they call 237 00:11:37,600 --> 00:11:40,800 a self reinforcing development loop, developers train models on 238 00:11:40,800 --> 00:11:44,280 a wide range of situations, test them in simulation, and 239 00:11:44,280 --> 00:11:47,040 continuously improve the system in all of this. 240 00:11:47,120 --> 00:11:49,840 Which is crazy. It's public. 241 00:11:49,880 --> 00:11:55,120 You and I can check it out. Alpa SIM ALPASIM is on GitHub. 242 00:11:55,160 --> 00:11:58,320 The data is on a hugging face. We can check it all out. 243 00:11:58,320 --> 00:12:00,600 The barrier to entry for autonomous vehicle development 244 00:12:00,680 --> 00:12:04,200 just drops significantly. Test like it has some absolute 245 00:12:04,200 --> 00:12:07,320 killer competition. NVIDIA has planted their flag. 246 00:12:07,400 --> 00:12:10,880 The technology that only a few companies had access to is 247 00:12:10,880 --> 00:12:14,000 beginning to be available for everybody. 248 00:12:14,000 --> 00:12:17,000 Now this is going to be a crowded race here. 249 00:12:17,080 --> 00:12:20,400 Every vehicle manufacturer has access to this and why would 250 00:12:20,400 --> 00:12:24,080 they spend years developing their own from here on out when 251 00:12:24,080 --> 00:12:26,560 they can just plug in NVIDIA stuff and then plug in some 252 00:12:26,560 --> 00:12:28,840 sensors, plug in their own cameras, plug in their own 253 00:12:28,840 --> 00:12:31,320 stacks of other things. It seems like a no brainer to 254 00:12:31,320 --> 00:12:33,080 me. They're going to be not the 255 00:12:33,080 --> 00:12:36,080 Apple, but they're going to be the Android of vehicles. 256 00:12:36,080 --> 00:12:37,240 They're going to make so much money. 257 00:12:37,240 --> 00:12:39,080 They're going to be selling their cloud services, their 258 00:12:39,080 --> 00:12:42,040 sensors, their chips, etcetera. This is going to be so much 259 00:12:42,040 --> 00:12:44,840 money for NVIDIA. I can't wait to see what happens 260 00:12:44,920 --> 00:12:47,200 in the future because one, I want to take a ride in this 261 00:12:47,200 --> 00:12:49,000 thing. I want to see how great it 262 00:12:49,000 --> 00:12:51,000 actually performs compared to a Tesla. 263 00:12:51,080 --> 00:12:54,040 And two, just think about how easy it's going to be when you 264 00:12:54,040 --> 00:12:56,320 need to go someplace and you don't actually have to own a 265 00:12:56,320 --> 00:12:57,440 car. Any that's going to be great. 266 00:12:57,520 --> 00:13:00,800 That's going to be 1 asset that doesn't depreciate when you 267 00:13:00,800 --> 00:13:04,040 drive it off the lot. It's going to be democratization 268 00:13:04,120 --> 00:13:04,800 of driving.