Latest / eMotors: Electric Revolution / Stellantis Hi-Drive: Autonomous Driving and Robotaxi Future
Transcript
- 0:09Welcome to the Deep Dive. We've got a a really fascinating
- 0:13set of sources today focusing on a major player in the Automotive
- 0:18World, Stalantis, and they're extremely aggressive push into
- 0:24the future of mobility. For years, the industry has been
- 0:27talking about Level 4 fully driverless autonomy as some
- 0:31distant fantasy, you know. Tech demo stuck in a few sunny
- 0:35test cities. Exactly.
- 0:37But that narrative I think is really over now.
- 0:39Stellantis is showing a full commitment to Level 4 by
- 0:42tackling this massive challenge with a very strategic dual track
- 0:46approach. A dual track.
- 0:47Approach right? They aren't just incrementally
- 0:49improving assisted driving, they are simultaneously building a
- 0:52commercial robo taxi fleet for global deployment and running
- 0:56sophisticated R&D right now on the complex roads of Europe.
- 0:59And that's the core of our mission today, isn't it?
- 1:01We need to unpack this strategy. Why the dual approach?
- 1:04What is the the enormous business challenge they're
- 1:07trying to solve by aligning with Titans like Uber and NVIDIA?
- 1:11And what are they learning from these living lab vehicles that's
- 1:14going to change the tech in those future robotaxis?
- 1:17Yeah. If you're looking to understand
- 1:18the future of transportation, I mean, this really is the
- 1:20blueprint. This shift defines what some are
- 1:23calling the next era of mobility, because it moves
- 1:27Stalantis beyond being a car builder, right, and transforms
- 1:31them into an integrator of AI electronics and mobility
- 1:36services. The complexity is just
- 1:38staggering. So let's start with that
- 1:40alliance, right? What's the strategic foundation
- 1:43for Stalantis here? This is not just some simple
- 1:46supplier agreement. Oh, not at.
- 1:47All they announced a massive collaboration, a four way
- 1:50alliance designed specifically to unleash level 4 autonomy
- 1:54worldwide. It's the ultimate example of
- 1:56outsourcing expertise and mitigating strategic risk.
- 2:01Stalantis realized they couldn't afford to spend a decade and
- 2:03billions trying to catch up. Try to build it all themselves.
- 2:06Exactly so they established a non binding memorandum of
- 2:09understanding and know you with the very best NVIDIA for the AI
- 2:12brains, Uber for the operational logistics and Foxconn for the
- 2:15integrated electronics. And that combined strength is so
- 2:20essential. If Olantis were to try and build
- 2:23their own full stack autonomous software, they'd be competing
- 2:26directly against NVIDIA. A company that has poured
- 2:29decades into this field? Yes.
- 2:32So they're manufacturing the vehicles, but they're
- 2:34outsourcing the intelligence. Precisely.
- 2:36Stalantis brings the global vehicle engineering scale, the
- 2:39manufacturing muscle, the physical platform.
- 2:43NVIDIA is the intelligence layer.
- 2:45They provide the safety certified autonomous driving
- 2:48software and the powerful AI computing stack.
- 2:51This is the difference between a car that helps and a true
- 2:54autonomous robot. And then the crucial operational
- 2:57piece comes from Uber. They are the global leader in
- 2:59ride hailing. They understand fleet
- 3:01management, interfaces, dynamic routing.
- 3:04And how to deploy a service network in dense urban
- 3:07environments. They've done it for years.
- 3:09And that leaves the 4th pillar, Foxconn.
- 3:11Yeah, and they handle the electronics and system
- 3:13integration. I mean, think about it, a level
- 3:154 vehicle has dozens of sensors, high performance computers,
- 3:18redundant systems, all of it needs to be packaged reliably
- 3:22and affordably. Right Foxconn's expertise in
- 3:24massive, efficient electronic supply chains is what helps
- 3:28drive down the cost and ensure that reliability.
- 3:31OK, so that collaboration provides the expertise, but what
- 3:33about the payoff? When does this all translate
- 3:36into actual driverless cars on the road?
- 3:38It has a clear timeline. Uber plans to begin deploying
- 3:42these Dalantis autonomous vehicles starting in the United
- 3:45States. In select cities I saw.
- 3:47Yes, selected cities with an initial commitment of around
- 3:505000 units. The start of production, or SOP
- 3:54is targeted for 2028. 2028 so that's not that far away.
- 3:58No. And the pilot programs will be
- 4:00ramping up in the years before that.
- 4:02They're not waiting for perfection.
- 4:04They're building a commercial service based on a scalable
- 4:06platform. What's fascinating here is that
- 4:08this massive global deployment plan with Uber and NVIDIA, it
- 4:13doesn't stand alone. No, it runs in parallel with
- 4:16their other Level 4 effort, the partnership with Pony dot AI.
- 4:19Right, which is dedicated specifically to Co developing
- 4:22and testing L4 autonomous vehicles for European roads.
- 4:25And that's the essence of the dual strategy.
- 4:27Then it is the Uber alliance is about massive scale and global
- 4:32revenue. The Pony dot AI collaboration is
- 4:35about geospecific technical adaptation.
- 4:39Europe has narrow streets, historic city centers.
- 4:42And just different driving challenges.
- 4:44Different challenges, complex intersections, less standardized
- 4:48infrastructure. So Spellantis is ensuring they
- 4:51have a reliable solution for both massive global deployment
- 4:55and for the complex environments they know so well in Europe.
- 4:58So if they're running two separate level 4 tracks, they
- 5:00must be using a common technological core.
- 5:02Yeah, you can't just custom engineer the hardware for every
- 5:04market. Absolutely not.
- 5:06Standardization is the key to scaling this efficiently,
- 5:08especially the vehicle needs to be reliable enough to run 24/7
- 5:11without a driver. Yeah.
- 5:13So both the global push and the European R&D, they both leverage
- 5:17the Stalantis AV ready platforms.
- 5:19And we know which specific platforms they're starting with,
- 5:21which gives us a huge hint about their immediate priorities.
- 5:24The case are a medium sized van and the STLA small platform.
- 5:27And that choice is highly strategic.
- 5:29They're starting with commercial and smaller passenger vehicles.
- 5:33Things that are ideal for purpose built Geo fenced robo
- 5:36taxi and delivery services. Exactly.
- 5:38These are not high volume consumer cars, at least not
- 5:41initially. The focus is on commercial
- 5:43mobility because it's just it's easier to control the
- 5:46operational domain of a fleet than to deploy a consumer
- 5:50product everywhere all at once. So let's talk about the tech
- 5:52stack that enables this. I mean, this platform needs the
- 5:55highest level of computing power and safety certification
- 5:58available. It does, and the critical
- 6:00enabling component here is the NVIDIA Dr. AGX Hyperion 10
- 6:05Autonomous Vehicle Architecture. And we really have to understand
- 6:08why this specific choice matters so much for level 4.
- 6:11We do. L4 requires a level of safety
- 6:14and functional redundancy that is just.
- 6:16It's far beyond assisted driving.
- 6:18This architecture has multiple high performance chips, often
- 6:21triple or even quadruple redundancy.
- 6:23So multiple chips. You're checking each other's
- 6:25calculations constantly. Instantly.
- 6:27It's necessary because if a human driver fails, they lose
- 6:30control. If the computer fails in a
- 6:32driverless car, the entire service fails, and potentially
- 6:37dangerously right. So the Hyperion 10 is the core
- 6:40computer. It runs the safety certified
- 6:43NVIDIA Drive OS and the full stack NVIDIA Drive AV software.
- 6:47This software handles everything.
- 6:49Perception, planning, execution. All of it with L4 parking and L4
- 6:53driving capabilities. It's what allows the vehicle to
- 6:55be truly autonomous within its designated area.
- 6:58It makes sense why NVIDIA CEO Jensen Huang describes the Level
- 7:024 vehicle as becoming a robot, one that sees, perceives, plans
- 7:06and drives with superhuman precision.
- 7:09It's no longer a car that sometimes helps you drive, it is
- 7:12the driver, and this technological robustness
- 7:15connects directly back to the business side.
- 7:17While the upfront cost of the Hyperion system is high, the
- 7:20architecture optimizes for one critical business metric, the
- 7:24total cost of ownership. TCO.
- 7:26That's a term we hear a lot, but why is it the single most
- 7:29important metric right here? For a robotaxi service to be
- 7:31profitable, the primary cost isn't the car, it's maintenance,
- 7:34diagnostics, and most importantly, downtime.
- 7:37Right, lost revenue. Every hour a robotaxi is offline
- 7:40is an hour of lost revenue. By integrating a safety
- 7:43critical, highly redundant system like Hyperion 10, they
- 7:46minimize those maintenance events and maximize operational
- 7:49uptime. It's what makes the service
- 7:51viable and scalable by that 2028 target.
- 7:54Okay. That addresses the global scale
- 7:56and the platform. Let's pivot back to that second
- 7:58track though, the deep R&D for those complex urban
- 8:01environments. Where is Delantis focusing their
- 8:04research in Europe? That focus is centered in the
- 8:07High Drive Project, which is Europe's flagship initiative for
- 8:10advancing vehicle automation. It's designed to rigorously test
- 8:14and enable higher levels of automated driving in really
- 8:17diverse, challenging real world conditions, far beyond your
- 8:21typical highway test. This research feeds directly
- 8:24into that Pony dot AI collaboration.
- 8:27SVP of Technology Research and Lalleron called Hydrive an
- 8:31example of open innovation. Why is that kind of open
- 8:34collaboration so necessary for level 4?
- 8:37Because you can't perfect L4 in isolation.
- 8:40Europe has distinct traffic patterns, signaling
- 8:43infrastructure. Working with leading European
- 8:46academic and industry partners through High Drive allows
- 8:49Stalantis to well to share the burden of development.
- 8:52And test solutions against different regulations.
- 8:54Right, and deliver practical and reliable solutions faster than
- 8:58they could alone. It ensures the tech isn't just
- 9:01scalable, but adaptable. To make this R&D more tangible,
- 9:05Celantis just showcased 2 Living Lab vehicles in Brussels.
- 9:09Incredible vehicle. There's actually data vacuums on
- 9:11wheels. Let's start with the Peugeot E
- 9:132008. Yeah, the Peugeot is a true
- 9:16mobile the laboratory. It's built to just gather
- 9:18massive amounts of data to train and validate all four systems in
- 9:21dense urban areas. You can tell by the sheer sensor
- 9:24count. What?
- 9:24Are we talking about here? It has this huge rooftop module
- 9:27with eight lidars, 9 cameras and four radars. 21 primary sensors.
- 9:3121 sensors, not even counting the connectivity hardware.
- 9:34That's just It sounds like overkill, but it speaks to the
- 9:37complexity of navigating A dense city intersection without a
- 9:40driver. You need 360 3 awareness
- 9:44constantly with multiple redundant views.
- 9:47And second vehicle, the Maserati Levante, it demonstrates the
- 9:50role of connectivity. Exactly.
- 9:52Specifically V2X or vehicle to everything communication.
- 9:57This is less about what the car sees and more about what the car
- 10:00knows from the network around it.
- 10:02How does that actually work in practice?
- 10:04The Levante system uses V2X and high def mapping to communicate
- 10:08with digital road infrastructure and potentially other vehicles.
- 10:11OK so this lets get real time hazard alerts, say a sudden lane
- 10:15closure around a blind corner or dynamically recognized signage
- 10:19and construction changes faster than a stand alone vision system
- 10:22could. But if they're relying so
- 10:23heavily on V2X and 5G, doesn't that make deployment hostage to
- 10:27infrastructure build out? What happens if the connectivity
- 10:29drops? That's a great question and it's
- 10:32why the system is fundamentally designed around redundancy.
- 10:35Connectivity provides crucial pre emptive information, but the
- 10:39on board NVIDIA AI still has to be able to navigate safely if
- 10:42V2X goes down. So the sensors are the fall
- 10:45back. The advanced sensors and the
- 10:47computing power are the necessary fall back.
- 10:49The R&D focus is on making the system robust enough to handle
- 10:53those inevitable connectivity gaps.
- 10:55So synthesizing this entire strategy, we see 2 clear
- 10:58benefits. First, the immediate goal is
- 11:01level 4 robo taxes by 2028. Second, the fundamental research
- 11:05from Hydrive is feeding back into their existing ADAS
- 11:08portfolio. It's a virtuous cycle.
- 11:10Stalantis is implementing lessons from sensor fusion and
- 11:13connectivity testing in programs like Hydrive directly into
- 11:17consumer features like enhancing Level 2 plus hands free highway
- 11:20driving that's available today. So they're refining the consumer
- 11:23product while aiming for the future driverless service
- 11:25markets. If we connect this to the bigger
- 11:27picture, the dual strategy is it's a masterstroke of hedging.
- 11:31They use R&D to refine the core tech, the sensors V2X5G to make
- 11:36sure the foundation is ultra reliable for complex European
- 11:39traffic. And while that refinement
- 11:42happens, they're simultaneously locking down the massive
- 11:45commercial partnerships needed to deliver a Level 4 service
- 11:48globally at scale by 2028. Their commitment feels total.
- 11:51It is. They're positioning themselves
- 11:53to capitalize on the economic potential.
- 11:55Their CEO suggested that autonomous mobility opens the
- 11:58door to new, more affordable transportation choices.
- 12:02The efficiency from the optimized TCO of a robot fleet
- 12:05is what makes that 2028 target not just a tech aspiration, but
- 12:09a major business transformation. And you just have to consider
- 12:12what Uber's CEO Darko swishaw. He said that NVIDIA is the
- 12:15backbone of the AI era, and now they're unleashing L4 autonomy
- 12:18in an enormous scale. When you think about the vehicle
- 12:21truly becoming a robot, how will our relationship with
- 12:24transportation change when the responsibility of driving is
- 12:27just gone? Will we ever truly trust a
- 12:31machine with our lives the way we trust a human driver?
- 12:35That's something worth exploring far beyond these sources.
- 12:38It raises fascinating questions about urban planning, too.
- 12:41I mean, if fleets are optimized, what happens to parking
- 12:43structures? We are watching the auto
- 12:45industry transform into a robotics industry in real time.
- 12:49We hope this deep dive into Stalantis's autonomous strategy
- 12:53gave you exactly the targeted information you needed to
- 12:55understand this critical moment. We love digging into these
- 12:58complex topics with clarity and interest, extracting the most
- 13:01important insights. If you enjoy this analysis, we
- 13:04invite you to follow the channel to ensure you never miss a deep
- 13:07dive. Thank you for listening.
- 13:09We'll catch you next time.