Best of LinkedIn: Next-Gen Vehicle Intelligence CW 37/ 38

00:00:00: Brought to you by Thomas Allgeier and Franis, this edition highlights key LinkedIn posts on next-gen vehicle intelligence in weeks thirty-seven and thirty-eight.

00:00:08: Frenis supports automotive enterprises and consultancies with market and competitive intelligence decoding disruptive technologies, customer needs, regulatory change, and competitive moves.

00:00:18: so product teams and strategy leaders don't just react but shape the future of mobility.

00:00:24: Welcome back.

00:00:24: Today, we're doing a deep dive into what's really moving the needle in next-gen vehicle intelligence.

00:00:29: We've scanned everything from calendar weeks, thirty-seven and thirty-eight, and our goal here is simple.

00:00:35: Cut through the noise.

00:00:36: We want to pull up the key insights on software-defined mobility, what's actually happening with autonomy rollouts, and, you know, how AI is changing the game in development.

00:00:44: And what struck me looking through the sources is how the conversation has shifted.

00:00:47: It's really moved beyond just theory now.

00:00:49: We're seeing posts from major OEMs, key tier ones, and they're all execution, things like operating models, actual framework specific partnerships needed to make this stuff happen safely efficiently.

00:01:00: Okay, good.

00:01:01: So let's unpack this then.

00:01:03: Maybe starting right at the core.

00:01:04: Our first big theme seems to be SDV architecture and middleware, the essential plumbing, if you like.

00:01:12: Yeah, you could call it that.

00:01:13: And it's, well, it's non-negotiable now.

00:01:15: Multiple sources confirm this, like burned hardone.

00:01:18: who basically said middleware is the decisive layer.

00:01:22: Decisive how?

00:01:23: Well, if you can't scale your software across all the different EE domains in the vehicle.

00:01:30: You know, the whole architecture, you're kind of stuck.

00:01:32: Millware provides that crucial interoperability.

00:01:35: It's what enables faster, more manageable software updates and cycles.

00:01:38: Okay, so faster cycles, more manageable.

00:01:41: But even as the industry grapples with getting SDVs right, some are already looking past it.

00:01:45: Sachin Luande made the point that maybe SDV is becoming... baseline.

00:01:49: Exactly.

00:01:50: He argued we're moving beyond just software-defined to what he called the intelligence-defined vehicle.

00:01:55: Intelligence-defined.

00:01:56: That sounds like a big leap.

00:01:57: It is.

00:01:57: It's a huge step change.

00:01:58: Yeah.

00:01:59: If SDVs are, as Lana put it, table stakes now, then this intelligence-defined platform, that's the real competitive edge.

00:02:06: It's about building intelligence, like context-aware, predictive models using continuous data into the vehicle from the absolute beginning.

00:02:14: Think anticipation, not just reaction.

00:02:17: That

00:02:17: sounds incredibly complex, and it implies needing serious speed and development, right?

00:02:23: Yeah.

00:02:23: Which brings us to the operational side.

00:02:26: Hadeat Pakhtian shared a really interesting case study from Mercedes-Benz.

00:02:30: Ah, yes, the DevSecOps transformation.

00:02:32: Exactly.

00:02:33: Showing how they use tools like GitLab and AWS to slash deployment complexity.

00:02:38: They went from, what, months down to weeks?

00:02:40: Weeks, yeah.

00:02:41: Which is impressive.

00:02:42: But it's more than just adopting tools, isn't it?

00:02:44: It's a deep cultural shift.

00:02:46: Moving away from that old V model development to continuous DevOps.

00:02:50: True.

00:02:50: But chasing speed like that, it must introduce risks.

00:02:54: Which leads to a pretty important warning we saw.

00:02:56: Definitely.

00:02:57: Robert Faye, you know, Mr.

00:02:59: Broken Testing, he flagged a real issue with this whole shift left idea.

00:03:02: He called it the copy-paste trap.

00:03:04: Basically teams taking their old hardware in the loop or Hilo test cases.

00:03:08: The ones designed for physical hardware.

00:03:10: just dumping them onto virtual ECUs, VECUs, earlier in the process.

00:03:15: Right.

00:03:15: Sounds efficient on the surface.

00:03:17: But Fay argues it's an illusion.

00:03:18: You're not actually validating the functionality properly early on.

00:03:22: You're just copying late-stage tests.

00:03:24: He says true shift less needs what he calls semantic contracts.

00:03:28: Semantic contracts.

00:03:30: What does that mean for, you know, the engineers doing the work?

00:03:33: It means defining the precise expected behavior of an application before you even code it.

00:03:38: Validating the logic?

00:03:40: The semantics?

00:03:41: Early.

00:03:42: That's where you find issues cheaply, not by just running old hardware tests on a simulator.

00:03:46: Got it.

00:03:46: So, smarter testing, not just earlier testing.

00:03:49: Okay, so we need speed, new architecture, smarter testing.

00:03:51: And the tech underpinning a lot of this, or second theme, has to be AI, right?

00:03:56: If middleware is the foundation.

00:03:57: Then AI is becoming the vehicle's central nervous system.

00:04:01: That was a phrase Mate Al's Bezera used, and it captures it well.

00:04:04: A unified, intelligent mobility platform.

00:04:07: And you see the money following that vision.

00:04:08: Volkswagen committing up to a billion euros to AI by twenty

00:04:12: thirty.

00:04:13: Yep.

00:04:13: Sabine Schunert and Steve Greenfield highlighted that.

00:04:15: And it's not just R&D for its own sake.

00:04:17: The goal is specific.

00:04:19: Cut the product development cycle.

00:04:21: Get it down to thirty six months or less.

00:04:23: Wow.

00:04:24: So AI is directly linked to core engineering efficiency.

00:04:28: Yeah.

00:04:28: It's not some futuristic maybe.

00:04:30: And we see that efficiency showing up right in the cabin, too.

00:04:33: Magnus Ostberg showed off the Mercedes MBUX virtual assistant.

00:04:38: Using Google's automotive AI agent, right?

00:04:40: Yeah,

00:04:40: and it's clearly more than just better voice commands.

00:04:43: It's aiming for like context-aware conversation, a proper co-pilot feel.

00:04:48: And beyond convenience, AI's boosting accessibility.

00:04:52: Tim Scannell pointed out a fascinating potential application in car, real-time sign language interpretation for deaf drivers.

00:04:59: That's incredible, using intelligence for inclusion.

00:05:02: Exactly, enhancing safety and independence for everyone.

00:05:04: Okay, but we need to inject some caution here.

00:05:07: There was a strong counterpoint raised forcing some critical thinking.

00:05:10: Steve Greenfield cited Missy Cummings.

00:05:12: Ah, yes.

00:05:14: From her talk at the autonomous event, she really didn't mince words about the fundamental challenges with AI and self-driving.

00:05:20: What was her main concern?

00:05:21: She hammered the problem of hallucinations.

00:05:24: These are basically statistical errors, inferencing mistakes that can lead to really dangerous things like symptom breaking.

00:05:30: Phantom

00:05:31: breaking, right.

00:05:32: We've heard about that.

00:05:33: Her

00:05:33: take was pretty blunt.

00:05:34: She said neural nets do not know anything.

00:05:37: They just spot correlations in data.

00:05:39: They don't understand context or causation in a human way.

00:05:43: If that's the case, that these complex systems have inherent flaws.

00:05:48: How does the industry square that with massive investments like VW's billion euros?

00:05:53: Is it solvable with better code or is it a deeper limitation needing, well, careful handling?

00:05:59: That's the crucial question, isn't it?

00:06:01: And it leads perfectly into our third theme, the need for disciplined execution and autonomy and AD gas.

00:06:07: The industry isn't stopping, but the approach is changing.

00:06:09: How

00:06:10: so?

00:06:10: Volodymar Seminition, among others, stressed that the best way to handle those rare, tricky edge cases safely is simulation-first development.

00:06:18: Using digital twins, synthetic data, that sort of thing.

00:06:21: Precisely.

00:06:22: Instead of driving millions of real-world miles hoping to encounter a one-in-a-million event, you create and test it digitally.

00:06:30: over and over.

00:06:31: We saw Vallejo and Capgemini setting a really ambitious target here.

00:06:34: Oh yeah.

00:06:35: They're aiming for eighty percent virtual testing for ATIS development.

00:06:38: Eighty percent.

00:06:39: That's a massive shift with huge implications for cost and time.

00:06:43: And looking at how this plays out in the real world, especially for trucking, the hype seems to be fading.

00:06:49: Completely.

00:06:50: Woshik Siwick noted that autonomous trucking is all about disciplined gradual scaling now.

00:06:56: Focus on early deployments and lower-risk scenarios, like defined hub-to-hub routes on highways, slow, steady, proven execution wins.

00:07:03: Makes sense.

00:07:04: But city pilots are still pushing forward, aren't they?

00:07:06: We saw news from Singapore.

00:07:08: Right.

00:07:08: Jacqueline Paul and Lee Yauhee reported on their first autonomous shuttles rolling out in Punggol.

00:07:13: In London.

00:07:14: Daniel Abreu-Marquez suggested it's becoming a real battleground, potentially Waymo versus Wave Uber versus Babylift.

00:07:20: A three-way showdown, interesting.

00:07:22: And that competition is pushing the hardware side too, especially for enabling that disciplined level three rollout.

00:07:28: Swaraj Postel pointed out improved sensors are key.

00:07:31: Like the new Vallejo litter.

00:07:33: Exactly, the Scala three unit.

00:07:35: It can detect objects up to three hundred meters away.

00:07:37: Okay, three hundred meters.

00:07:38: Why is that specific distance such a big deal for someone listening?

00:07:42: Because if you want safe eyes off level three autonomy at highway speeds, you need that kind of foresight.

00:07:47: Three hundred meters gives the system enough time to detect, analyze, and react safely.

00:07:52: The hardware has to provide that buffer.

00:07:54: Got it.

00:07:54: So the sensors enable the function.

00:07:56: And once these systems are out there, maintaining them becomes huge.

00:08:00: Bruno Moretti shared some eye-popping numbers.

00:08:02: Yeah, the economics of ADA servicing.

00:08:05: projected to save society something like thirty four billion dollars a year in accident costs

00:08:09: while also creating a fifty billion dollar aftermarket industry by twenty thirty exactly

00:08:14: a massive economic driver for getting the service and maintenance right.

00:08:18: okay so shifting from preventing costs to creating service revenue that requires fundamental changes in how companies operate which is our final theme.

00:08:26: right partnerships ecosystems strategy.

00:08:29: absolutely Dr.

00:08:30: Matthias Traub argued strongly that for Europe to stay competitive, it needs to fully embrace SDV through tech collaboration.

00:08:39: and innovation, not just incremental improvements.

00:08:41: But how companies collaborate seems to differ geographically.

00:08:45: Dr.

00:08:45: Jan Weinger shared some interesting research comparing approaches.

00:08:48: That was fascinating.

00:08:50: His work showed European firms often still view partnerships, mainly through the lens of cost optimization, risk reduction.

00:08:56: Standard supply chain thinking.

00:08:58: Kind

00:08:58: of, whereas US and Chinese companies, they're using partnerships as innovation catapults, actively co-creating, especially around AI and SDV.

00:09:06: And the numbers back that up.

00:09:07: Starkly.

00:09:08: Winger found seventy-three percent of U.S.

00:09:10: auto companies see partnerships as essential for survival.

00:09:14: It's a different strategic mindset.

00:09:15: Less about managing costs, more about betting on growth.

00:09:18: And we are seeing more open collaboration now, even in Europe.

00:09:21: BMW, Mercedes, Renault's Ampere, they're implementing open source development for actual series production.

00:09:27: Confirmed by Martin Schleicher and Markus Rettstadt, they see open source not as a risk, but as a genuine innovation accelerator, even for the big traditional players.

00:09:36: Okay, but... Let's balance that optimism.

00:09:39: There was some sobering news too, reflecting the economic pressures.

00:09:43: Lucas Tim highlighted the Bosch situation.

00:09:45: Right, the

00:09:45: eight thousand two hundred and fifty job cuts in their software and ADES units, that sent ripples through the industry.

00:09:52: And

00:09:53: the reason given.

00:09:54: that the auto software market is not developing as expected.

00:09:58: Yeah,

00:09:58: and that supplier software teams are often the first hit when OEM innovation budgets get tight or shift focus.

00:10:04: It's a tough market signal.

00:10:05: Definitely a wake-up call for suppliers.

00:10:07: You can't just offer generic software.

00:10:08: You need specialized high-value systems that OEMs see as indispensable.

00:10:12: Precisely.

00:10:14: Although, on a more positive partnership note, fleet management continues to be strong.

00:10:18: Lars Kraus detailed the global deal between connected cars AS and Volkswagen commercial vehicles.

00:10:23: Focused on uptime, predictive maintenance, core value for commercial fleets.

00:10:26: Exactly.

00:10:27: And circling back to the tech foundation, the Silicon and the cloud are key partnership areas.

00:10:33: Augustine Friedl noted Qualcomm's auto design wind pipeline hit forty-five billion dollars.

00:10:38: Forty-five

00:10:39: billion.

00:10:40: shows their big push into SDV and AI.

00:10:43: Huge numbers.

00:10:44: And the cloud partnerships are delivering efficiency games too.

00:10:47: Dr.

00:10:47: Henning Rudolph shared how Vallejo working with AWS on virtual testing environments.

00:10:52: A cloud-managed tester.

00:10:53: Yeah, it's cutting ECU software development cycles by up to forty percent.

00:10:56: Bringing that hill-like capability into the cloud really speeds things up.

00:11:01: Okay, so pulling all these threads together from the last couple of weeks, what's the main takeaway?

00:11:04: I think it's this clear shift.

00:11:06: The industry is moving beyond being just defined by software.

00:11:10: It's now aiming to be driven by intelligence.

00:11:12: But, and this is the crucial part, success hinges entirely on pragmatic execution, meaning simulation-first development, adopting these new open operating models, and frankly, having the financial and cultural stamina to navigate these major shifts.

00:11:26: Right.

00:11:27: Well, if you enjoy this deep dive, new additions drop every two weeks.

00:11:30: Also, check out our other additions on electrification and battery technology, future mobility and market evolution, and commercial fleet insights.

00:11:38: And before we wrap up, here's something to think about.

00:11:40: We see these massive AI investments, like VW's billion euros, happening at the same time as tough restructuring, like the Bosch cuts.

00:11:49: So is the ultimate success of this intelligence-defined vehicle going to be a sprint one by the absolute smartest tech?

00:11:55: Or is it more of a marathon, won by the companies with the deepest pockets and the most adaptable culture?

00:12:00: Something to ponder.

00:12:01: Thanks for diving deep with us today.

00:12:02: We'll catch you on the next one.