Best of LinkedIn: Next-Gen Vehicle Intelligence CW 49/ 50

00:00:00: brought to you by Thomas Allgaier and Frennus.

00:00:02: This edition highlights key LinkedIn posts on next-gen vehicle intelligence in weeks forty-nine and fifty.

00:00:07: Frennus supports automotive enterprises and consultancies with market and competitive intelligence, decoding disruptive technologies, customer needs, regulatory change and competitive moves, so product teams and strategy leaders don't just react but shape the future of mobility.

00:00:23: Welcome to the Deep Tive.

00:00:25: Today we're cutting through all the noise to focus on what mobility pros we're really talking about on LinkedIn during calendar weeks, forty nine and fifty.

00:00:32: And you know what we saw, it really felt like an inflection point.

00:00:35: We've moved past those big visionary stories.

00:00:37: That's so true.

00:00:37: It's less about planning now and all about building.

00:00:40: The whole conversation has shifted to execution.

00:00:42: And when we boiled it all down, three big themes just jumped out.

00:00:46: First, this really rapid move from the software defined vehicle, the SDV, to what people are now calling the AI defined vehicle.

00:00:52: Right.

00:00:53: And then second, the really tough part, the, you know, the often painful process of actually hardening that SDV stack, making it work at scale.

00:01:01: And third, and this is critical, is the intense focus on autonomy validation because none of this gets on the road safely without it.

00:01:09: And all of this is happening with some pretty challenging data economics and really a major geopolitical reshaping of the whole supply chain.

00:01:18: So we've taken all those top insights from those two weeks and packed them into one deep dive.

00:01:22: Just for you.

00:01:23: Let's get right into that

00:01:25: first

00:01:25: big shift.

00:01:27: The AI-defined vehicle.

00:01:28: or ADV.

00:01:29: Yeah,

00:01:29: let's unpack that.

00:01:30: I mean, for years it was all SDV, right?

00:01:32: The cars run my code, but now someone like Renee Dee pointed out that ADs are really systems of systems.

00:01:37: It's not just about a simple algorithm anymore.

00:01:40: No.

00:01:40: It's about needing these sophisticated orchestration layers to connect all the tools and data pipelines for what he called self-evolving models.

00:01:47: And that idea of self-evolution, that takes us straight to agentic AI.

00:01:51: This is where things get really fascinating.

00:01:53: Shams Sundar basically framed agentic AI as the next huge disruption.

00:01:57: The vehicle goes from being a passive tool, just executing commands, to an active partner in mobility.

00:02:03: So

00:02:03: the car is actually reasoning and learning on its own.

00:02:06: Exactly.

00:02:06: Acting autonomously based on its environment and its goals.

00:02:09: It's a huge shift.

00:02:11: And it's already creating a real competitive divide.

00:02:13: Oh, absolutely.

00:02:14: Just look at Rivian's recent autonomy in AI Day.

00:02:18: Tenskip made it crystal clear they are all in on ADV and they're doing it with this deeply vertical approach.

00:02:23: They're owning the whole stack.

00:02:25: Right.

00:02:25: All the way down to the silicon.

00:02:26: They talked about their in-house RAP-One chip.

00:02:29: Built on a five nanometer process.

00:02:31: Yeah, a five nanometer process.

00:02:32: Yeah.

00:02:33: Capable of eight hundred TOPS.

00:02:34: Ever anyone listening?

00:02:35: that's tear operations per second.

00:02:38: That kind of power is what you need for complex, real-time AI, like their universal hands-free driving.

00:02:44: Exactly.

00:02:45: That vertical integration is what gives them speed and control.

00:02:48: But this agentic trend, as Stefano Marzani pointed out, it goes way beyond just the car.

00:02:54: He calls it the fourth disruptor for the whole enterprise.

00:02:56: It's optimizing everything, engineering, supply chains, manufacturing.

00:03:00: It's a massive efficiency play.

00:03:02: We saw a great example from Madie Alves Bizarre, I think it was.

00:03:05: He saw it.

00:03:05: He saw it, yeah.

00:03:06: They use an agenic AI approach and cut their software tests execution time by seventy-five percent.

00:03:11: Seventy-five percent.

00:03:12: I mean, that's the kind of gain that just fundamentally changes your development costs.

00:03:16: It is, but there's a big warning that comes with all this speed.

00:03:20: Thomas Dickhoff offered a very necessary dose of caution.

00:03:24: Right.

00:03:24: He pointed out that a lot of these AI solutions, they're brilliant on their own, but they risk falling into the post-C trap.

00:03:31: The proof of concept trap.

00:03:32: So for anyone who hasn't run into that, what does that mean in an automotive context?

00:03:37: It means you build this amazing prototype that works perfectly in a lab, but it has no horizontal integration.

00:03:44: It can't be deployed across the whole vehicle or across different engineering teams.

00:03:48: So

00:03:48: it's stuck in a silo.

00:03:49: Exactly.

00:03:50: Dick Hof's warning is that the foundational work, the scalability and governance is getting skipped over for the flashy demos.

00:03:57: And you can't commercialize something that lives in a silo.

00:04:01: That's a perfect pivot to our second theme then, that critical, really hard work of hardening the software-defined vehicle stack and its architecture.

00:04:09: If AI is the brain, this is the nervous system.

00:04:12: And it looks like a blueprint for that.

00:04:14: nervous system is really starting to emerge.

00:04:17: Matt Damascino highlighted that zonal architectures and service-oriented platforms

00:04:23: are

00:04:25: pretty much the dominant design now.

00:04:26: Yeah, that zonal approach basically flips everything on its head.

00:04:30: It organizes the electronics by physical location, these zones, and funnels it all through centralized high-performance computers.

00:04:39: HPCs.

00:04:40: It's a huge move away from the old ECU heavy system.

00:04:43: And it's all driven by hardware innovation.

00:04:46: Michael Budd explained how chiplet technology is making this possible.

00:04:49: Think of them like Lego bricks for silicon.

00:04:51: Right, exactly.

00:04:52: Standard modular blocks.

00:04:54: And they're a game changer for EVs because they reduce complexity, they speed up development, and maybe most importantly, they save a ton of power.

00:05:00: Which means better range.

00:05:01: Better range.

00:05:02: It's so critical that Bosch is co-leading this chassis initiative to set up open standards for it all.

00:05:08: So the hardware foundation is there, but Justine Lito Koonthanam pointed out that the real nightmare is the software integration.

00:05:15: The huge challenge is bridging autosar classic.

00:05:17: Which is your hard real-time safety, like... breaking.

00:05:20: Right, with Autosar Adaptive, which is all the dynamic stuff infotainment, OTA updates.

00:05:25: That is such a critical point.

00:05:27: Classic is all about safety and predictability, running on smaller MCUs.

00:05:31: Adaptive is about flexibility, running on those powerful HPCs.

00:05:35: I'm trying to get them to talk to each other seamlessly.

00:05:37: Is integration hell, especially for the legacy OEMs?

00:05:41: Well, Raul Arradondo said they struggle because of their Hardware-first culture, right?

00:05:46: Right.

00:05:46: These slow seven-year development cycles and always prioritizing costs today over specifying the right hardware for tomorrow.

00:05:54: They're just not built for this kind of change.

00:05:57: And Venetius Tadeuzein argued it's also an organizational problem.

00:06:01: It's Conway's law in action.

00:06:04: The system architecture ends up mirroring a broken org chart.

00:06:07: I've definitely seen that play out.

00:06:08: Lucas Tim mentioned the challenges with Volvo's EX-Ninety were because of internal issues.

00:06:13: He said agility was outpacing systems discipline.

00:06:15: You just can't rush the foundations.

00:06:17: It really highlights the need for standards.

00:06:19: Marcus Rettstat made a great point that interoperability doesn't always have to mean open source code.

00:06:25: It starts with open specifications.

00:06:27: Exactly.

00:06:28: Things like SOMI-P and DDS.

00:06:31: as everyone's tech can play nicely together, even if the underlying code is proprietary.

00:06:36: That need for discipline brings us to our third cluster.

00:06:39: Autonomous driving, safety, and validation.

00:06:43: The feeling from the sources is that we are definitely past the hype peak.

00:06:47: Oh, yeah.

00:06:49: It's not about big promises anymore.

00:06:50: It's about systematically, and I mean rigorously, handling those rare edge cases.

00:06:57: And that starts with the data.

00:06:58: Shansai brought up this thing called the multirate problem.

00:07:01: Right, because all your sensors cameras, LiDAR, they all run at different speeds.

00:07:06: So you're losing data in the translation.

00:07:07: He said up to fifteen percent can be lost.

00:07:10: It's a huge gap in perception.

00:07:11: A huge

00:07:12: potential failure point.

00:07:13: But you mentioned a research team at Kumamoto University found a really clever software solution to unify all those data streams without touching the hardware.

00:07:21: And the potential is a double-digit drop in prediction errors.

00:07:24: Yeah.

00:07:24: That's a game changer for sensor fusion.

00:07:26: Absolutely.

00:07:27: And that reliability is what underpins the two big and very different AD strategies in the US that Michael Sura contrasted.

00:07:34: You've got Waymo on one side.

00:07:36: The built-in suspenders approach.

00:07:37: Totally.

00:07:54: It's a

00:07:54: fascinating contrast, and while they figure that out, the global competition is not waiting.

00:08:00: Ozger Noret and Poo School highlighted how intense things are in China.

00:08:03: Right, with Cherry's Exceed Star Era ES.

00:08:06: Yeah, powered by WeRide and Bosch.

00:08:09: It won a really tough ADS competition with the only zero takeover run.

00:08:14: They credit a single stage end-to-end AI architecture that drives in a very human-like

00:08:19: way.

00:08:20: That puts a ton of pressure on Europe.

00:08:22: Philip Oliven noted that Europe's quote, obsession with safety.

00:08:25: while good could cause them to fall behind the U.S.

00:08:28: which favors faster experimentation.

00:08:30: And Pierre Francesco Moran was even more blunt.

00:08:32: He said Europe has to act by twenty twenty six or the gap with Asia and the U.S.

00:08:36: and AV tech could become unbridgeable.

00:08:38: So if safety is the bottleneck then validation becomes everything.

00:08:41: It becomes the core strategic asset.

00:08:44: Preeti Ranadeev explained that managing A.I.

00:08:47: risk means you need a mix of everything.

00:08:49: Deterministic rules.

00:08:50: Runtime monitoring.

00:08:52: ISO two six two six two and so too.

00:08:55: Can you just quickly unpack SOTIF for our listeners?

00:08:57: Yeah.

00:08:57: Why is it so important with AI?

00:09:00: Sure.

00:09:00: SOTIF stands for Safety of the Intended Functionality.

00:09:04: It's critical because AI isn't about a simple malfunction.

00:09:07: It's about the system performing as design, but still being unsafe because it misinterpreted something like a faded lane line.

00:09:14: Because the AI is statistical.

00:09:16: Exactly.

00:09:16: So you need SOTIF to manage those risks.

00:09:18: And that's why Andrea Leitner said that validation capacity, combining huge simulations with real-world testing, is now a core strategic

00:09:26: asset.

00:09:26: Okay, let's talk money.

00:09:27: Our fourth theme is data, connectivity and lifecycle economics.

00:09:32: How does all this tech actually change the business model?

00:09:34: Well, according to Wilhelm Hedberg, it changes the vehicle from just being depreciating metal into an investable asset with stable yield.

00:09:41: Because of the data.

00:09:42: Because the value is in the continuous data stream and the ability to monetize services over the vehicle's life.

00:09:48: We're even seeing that in things like fleet management.

00:09:50: Jeremy McLean showed how Senseems AI can pick the right pool vehicle for a trip based on purpose, sustainability, everything.

00:09:57: Just smart, data-driven allocation.

00:09:59: And the key to all of this lifecycle value is over-the-year updates.

00:10:02: OTA, it's mandatory now.

00:10:04: And it has a real ROI.

00:10:06: Balasubramani Sandy Prakash said OTA can boost ROI by twenty-five percent.

00:10:11: That's a massive incentive to get that software back end right.

00:10:14: And it's pushing OEMs to become what Magnus Esberg called original experience manufacturers.

00:10:20: He's talking about Mercedes-Benz and their MBOS.

00:10:22: one point two integrating partners like Google Maps and the Unity game engine.

00:10:27: The experience becomes the product.

00:10:28: And that experience starts before you even buy the car.

00:10:31: Guy Van Maal pointed out how Porsche is using an interactive three D configurator to elevate that initial customer touch point.

00:10:37: It's the new luxury front line.

00:10:39: Zooming out a bit.

00:10:40: Dando reminded us of the climate imperative.

00:10:42: Transport is fifteen percent of global emissions and the EV shift is being pushed not just by policy but by price parity thanks to falling battery costs and fierce Chinese competition.

00:10:53: Which is forcing a whole rethink of how we tax mobility.

00:10:56: Dr.

00:10:57: Odyssey from out proposed this mobility user charge or MOBC.

00:11:01: Yeah basically transparent.

00:11:03: impact-based pricing tied to your personal CO² footprint.

00:11:07: It's a big shift.

00:11:08: And Pete Dyson analyzed the UK's proposed three pence per mile charge for EVUs, which is trying to solve that same problem.

00:11:15: How do you replace falling gas tax revenue?

00:11:18: Which brings us to our final and maybe most urgent theme, geopolitics, competition, and market disruption.

00:11:25: And this one is a huge deal, especially for Europe.

00:11:27: Gabrielle Ivanoe gave this really stark warning about Europe's growing tech dependence on Chinese companies.

00:11:33: He

00:11:33: cited specific examples, right?

00:11:35: OPO and Audi for five G, RoboSense and FAW Toyota for Lidar.

00:11:39: Exactly.

00:11:40: And he explained this creates real security risks around espionage and potential sabotage because of China's national intelligence law.

00:11:47: Which is why Philip Brett argued that Europe's firm, twenty thirty five phase out of combustion engines is actually a vital industrial strategy.

00:11:54: He called it productive discomfort.

00:11:56: forcing the industry to reinvent itself.

00:11:58: He said Europe needs to stop curling parenting its auto industry.

00:12:01: But the speed difference is scary.

00:12:03: That Steve Greenfield report noted.

00:12:05: Chinese automakers already have massive overcapacity, which is fueling their global push.

00:12:10: True, but Yarsharim Lukas made a counterpoint that they still need to build trust and really localize the user experience for European consumers.

00:12:18: Although Mark Heiliger pointed to the Firefly, which was developed in Munich but built in China, it won Car of the Year in the Netherlands because its value was just off the charts.

00:12:27: Price and design are powerful levers.

00:12:29: And before we finish, we have to touch on the aftermarket.

00:12:32: Frank Turlip said the collision of all these forces, ADEs, AI, EVs, means collision shops have to become calibration hubs by twenty thirty.

00:12:41: Repairs becoming a software problem.

00:12:42: A huge one.

00:12:44: And Oliver Rusling pointed to the twenty twenty eight warranty cliff.

00:12:47: That's when the first big wave of EVs falls out of battery warranty.

00:12:51: It's a massive threat, but also a huge opportunity for the independent aftermarket.

00:12:56: If you enjoyed this deep dive, new additions drop every two weeks.

00:12:59: Also, check out our other additions on electrification and battery technology, future mobility and market evolution, and commercial fleet insights.

00:13:06: You know, when you look at all these sources together, the core challenge isn't just about the tech.

00:13:12: It's not just about eight hundred TOPS of silicon or solving the multifonari problem.

00:13:18: The real challenge you face as a mobility professional is organizational leadership and strategic sourcing.

00:13:24: Absolutely.

00:13:25: Considering the geopolitical warnings about supply chain dependency, the ultimate question we'll leave you with is this.

00:13:30: Yeah.

00:13:30: Who do you actually trust to build your critical architecture and who really owns the experience you're selling to your customer?

00:13:37: Thanks for joining us.

00:13:37: We'll talk soon.