Best of LinkedIn: Next-Gen Vehicle Intelligence CW 45/ 46

00:00:00: Brought to you by Thomas Allgaier and Frenus, this edition highlights key LinkedIn posts on next-gen vehicle intelligence in weeks forty-five and forty-six.

00:00:07: Frenus supports automotive enterprises and consultancies with market and competitive intelligence, decoding disruptive technologies, customer needs, regulatory change, and competitive moves.

00:00:17: so product teams and strategy leaders don't just react but shape the future of mobility.

00:00:23: Welcome to the deep dive.

00:00:24: Today, our mission is pretty straightforward.

00:00:27: We're diving into the intelligence we've gathered from the industry's front lines, specifically top insights on next-gen vehicles from LinkedIn in weeks forty-five and forty-six.

00:00:36: And if there's one big message, it's that the whole conversation has really shifted.

00:00:41: It really

00:00:41: has.

00:00:41: It's moved from just vision to hard, measurable execution.

00:00:45: For SDVs, for autonomy, for everything.

00:00:47: Exactly.

00:00:48: We're seeing this maturation across the whole ecosystem.

00:00:50: It's no longer a debate about if we should have software-defined architectures.

00:00:54: Now it's all about how.

00:00:55: How do you deploy them reliably, cost-effectively, and at scale?

00:00:59: And we saw so much content confirming that these semiconductor-centric designs are now just the operating reality.

00:01:06: So we've clustered everything into the five big transformation themes we saw.

00:01:11: OK, let's jump in.

00:01:11: Theme one.

00:01:13: SDV architectures.

00:01:15: This is really the core engine, isn't it?

00:01:17: Shifting value away from hardware refresh cycles.

00:01:19: To continuously updated software and services.

00:01:22: Yeah.

00:01:22: The car as an evolving platform.

00:01:24: And

00:01:25: what's so interesting is how that vision of a single, monolithic supercomputer has changed.

00:01:30: Oh, it's completely evolved.

00:01:31: It's become.

00:01:32: this much more practical, federated reality.

00:01:35: If Henry C. Manguino booked this down really well, you need these small, hyper-reliable, deterministic safety controllers.

00:01:42: The ASL-C or D stuff.

00:01:44: The ASL-C, ASL-D stuff, totally isolated.

00:01:46: Then you have your big Linux or QNX high-performance computers doing the heavy lifting for AI and data fusion.

00:01:53: And then, you know, Android or something similar, just for the UX.

00:01:55: So it's not just one chip doing everything.

00:01:57: It's about strategic segregation.

00:01:59: You have to keep those safety islands separate.

00:02:01: Precisely.

00:02:02: And the Silicon guys are designing for this.

00:02:04: You see chips like Qualcomm, Snapdragon, RideFlex, or Armzina, CSS.

00:02:09: They keep those dedicated safety islands, often running good old, auto-SAR classic, physically and logically separate from Linux or Android, even on the same piece of silicon.

00:02:20: It's the only way to do it, right?

00:02:21: To balance safety with flexibility.

00:02:23: It's

00:02:23: the necessary compromise.

00:02:24: But

00:02:25: that level of complexity, I mean, managing multiple OEs, continuous updates, that brings us to the big challenge.

00:02:31: Yeah.

00:02:32: How do you stop costs just spiraling out of control?

00:02:34: That's

00:02:35: the million dollar question for OEMs.

00:02:37: And it's why the news from Remco Timmer and Mike Nefkins was so important.

00:02:41: This new SDV maturity framework from here and Omnia.

00:02:45: The first of its kind.

00:02:46: Yeah, and it's vital.

00:02:47: It gives OEMs a structured way to actually benchmark their progress so they can figure out where to put those massive investments.

00:02:53: But the architecture itself doesn't sell cars.

00:02:56: I love the perspective from Gregor Stanglin on this.

00:02:58: He said the industry has to stop selling architecture and start solving problems.

00:03:02: Absolutely.

00:03:03: I mean, think about the end user.

00:03:05: A fleet operator doesn't care about a zonal architecture.

00:03:08: They care about savings per mile.

00:03:10: Or better route planning.

00:03:11: Exactly.

00:03:12: Or a driver just wants to find a parking spot instantly.

00:03:15: If the complexity is invisible and the benefit is obvious, then the architecture is doing its job.

00:03:20: Which leads to this interesting paradox Dr.

00:03:22: Gabriel Cyberth brought up.

00:03:24: By abstracting the hardware to speed things up, you actually create a new bottleneck.

00:03:30: Time.

00:03:30: It's a huge strategic tension.

00:03:33: If you standardize your hardware too early to enable your software, that software's innovation is then stuck.

00:03:39: It's limited by the pace of the hardware.

00:03:41: You're trapped by your own standard.

00:03:43: It's a speed trap, yeah.

00:03:44: You're waiting for the next hardware generation before your software can take the next leap.

00:03:48: All that architectural complexity, it means nothing if people don't trust the system.

00:03:53: Let's shift to theme two.

00:03:54: Autonomous functions, safety, and trust.

00:03:58: Feliz Fortuno made this great point that AI is the answer to building that trust at scale.

00:04:03: For everything.

00:04:04: For EVs, you know, fighting range anxiety, and for AVs, guaranteeing reliable decisions.

00:04:10: So how are they building that trust?

00:04:12: It's about completely rethinking validation.

00:04:15: Helen Yu detailed Nvidia's approach, which they call halers.

00:04:18: It's this three pillar system for safety.

00:04:22: You use DGX for the heavy AI development.

00:04:25: Then you use Omniverse and Cosmos for photorealistic simulation of billions of edge cases.

00:04:31: And finally, you deploy on the vehicle with the modular redundancy of their Drive AGX hardware.

00:04:37: And Laura Major from Motional had a great take on this.

00:04:40: She framed their training curriculum like coaching athletes.

00:04:44: It's a beautiful metaphor.

00:04:45: It's not about having the most data.

00:04:47: It's about having a rigorous curriculum to find and use the right data to train your models effectively.

00:04:52: We're also seeing these features just becoming mainstream now.

00:04:55: Yeah.

00:04:55: Jim Charlie announced Ford BlueCruise is expanding across Europe.

00:04:58: That's huge.

00:04:59: Hands-free highway driving is becoming a standard expectation.

00:05:02: And it's not just for high-end DVs anymore.

00:05:05: Federal Ribeiro highlighted how Bosch is bringing high-performance navigate on autopilot to icy vehicles, like the G-Elide Lincoln Co.

00:05:12: O-three.

00:05:13: That's a key point.

00:05:14: These high-level features are now powertrain agnostic.

00:05:17: That's the path to mass adoption.

00:05:19: And the economic scaling is really happening.

00:05:21: in logistics first.

00:05:22: For

00:05:22: sure.

00:05:23: Hugo Rout shared that Einride went public via SPAC at a billion dollar valuation.

00:05:29: That shows real commercial correction.

00:05:31: And we also saw that level four autonomous delivery pilot in Germany with REWE and LOXO, which Dominique Peer-Losher posted about.

00:05:39: Okay, let's move on.

00:05:40: Theme three.

00:05:41: In CAROS and the software stack, Prisanth Gorvahala called this a major strategic battleground.

00:05:48: It really is.

00:05:49: The split is between OEMs building their own thing, like Mercedes-Benz with MBOS.

00:05:54: Which Marcus Schaefer described.

00:05:55: Exactly.

00:05:56: A huge.

00:05:58: And then everyone else who was standardizing on Android Automotive.

00:06:01: And these OEs are driving some wild innovation in the digital experience.

00:06:05: Oh, absolutely.

00:06:05: Katherine Renz shared how BMW is integrating Amazon's next-gen Alexa, which is using over seventy large language models.

00:06:13: This isn't just a voice assistant anymore.

00:06:15: It's trying to be a genuine, context-aware, conversational partner.

00:06:18: And we saw a fantastic example of the underlying STV architecture, enabling a really cool feature, Google Maps getting live lane guidance on the Polestar IV.

00:06:27: Yeah, Nucle Sigal and Jargon Barons explain this.

00:06:30: It's so clever.

00:06:31: The car uses its existing ADES camera and local AI to analyze the lane markings in real time.

00:06:37: And feeds that video right back to the map interface.

00:06:39: Seamlessly.

00:06:40: It's a perfect example of sensor fusion for the user experience.

00:06:43: But

00:06:43: that kind of integration only works because the deep engineering stack allows it.

00:06:47: That's the crucial point MLKDHERI made.

00:06:51: The success of an SDV still depends more on deep software engineering, you know, managing auto SAR, Linux, Yachto than on abstract AI.

00:07:00: The embedded software defines the limits of the whole

00:07:02: system.

00:07:03: So how does that foundational engineering actually connect to these high-level features.

00:07:07: It's all about abstraction.

00:07:09: Ahmed Munir broke down why those AutoSAR layers matter.

00:07:12: You've got the MCAL abstracting the chip, the ECL abstracting the ECU hardware,

00:07:17: and the service layer abstracting the system complexity.

00:07:20: Right.

00:07:21: Those layers are what let the Google Maps feature grab sensor data without needing to know the specific details of the camera hardware it's talking to.

00:07:28: And that need for shared infrastructure is driving huge collaboration.

00:07:32: It is.

00:07:33: The news from Dr.

00:07:34: Moritz Neukirchner and Santil Kumar Sivanyanam that Qualcomm joined the Eclipse escort community is significant.

00:07:42: It's a shared, safety-aligned middleware, so everyone doesn't have to build the same foundational plumbing from scratch.

00:07:48: That takes us directly into Theme Four.

00:07:50: Semiconductors and EE architecture.

00:07:52: Yeah, August and Friedl positioned semiconductor strategy as the critical lever now for differentiation, cost, and performance.

00:08:00: And when you see vendors like NVIDIA with gross margins around seventy percent, it

00:08:04: changes the whole calculation for an OEM.

00:08:06: It

00:08:06: completely does.

00:08:07: It's driving players like Tesla, XPeng, and NIO toward pure in-house chip design to capture that value and control their own destiny.

00:08:15: But

00:08:16: isn't that a massive risk?

00:08:17: Aren't you just replacing a vendor bottleneck with an internal R&D bottleneck that cost billions.

00:08:22: It is a colossal risk, but they see it as necessary to get that absolute performance edge.

00:08:27: And Sean Seehe explained why.

00:08:29: The new bottleneck isn't just raw compute power, it's memory bandwidth.

00:08:33: An autonomous development card generates terabytes of data.

00:08:36: Moving that data fast enough for safe real-time decisions.

00:08:40: Using things like Micron's LPDDR-VX DRAM.

00:08:44: That is the new horsepower.

00:08:46: And you need custom silicon to solve that.

00:08:48: So we're seeing this rapid hardware convergence.

00:08:51: Dr.

00:08:51: Mathias Lash detailed how Infineon is focusing on open architectures like RIS-CV and making automotive ethernet the backbone.

00:09:00: And supply chain resilience is also driving strategy.

00:09:02: We heard from Dr.

00:09:03: Ramoon Ajawe about Samsung's push for triplets.

00:09:07: Like Lego blocks for chips.

00:09:08: Exactly.

00:09:09: It's a modular architecture.

00:09:10: It's transformative because it allows for cost-effective customization, faster time to market, and it makes the supply chain way more resilient than these big monolithic designs.

00:09:19: Okay, finally, let's get to theme five.

00:09:21: Strategy, geopolitics, and the China factor.

00:09:24: Patrick Caller's overview was a stark reminder of China's scale.

00:09:28: Oh, it's a global template now.

00:09:29: Top vehicle exporter, largest NEV producer and Chinese brands have fifty-five percent of their own domestic market.

00:09:35: That scale is reshaping everything.

00:09:37: And it creates immediate geopolitical tension around connectivity.

00:09:41: A huge amount.

00:09:42: Gonzalo Martinez-Dozagra and Michael Barnard raised concerns in Europe after those Chinese-made electric buses were found with remote access, including a remote

00:09:51: halt feature.

00:09:52: And as Evalentin Weber pointed out, remote access is pretty standard, but this is different.

00:09:56: It's

00:09:57: different because it forces Europe to confront a severe sovereignty risk.

00:10:02: What happens when your critical national infrastructure relies on a foreign software stack?

00:10:07: And China is no longer just catching up.

00:10:10: Puskal highlighted how their ADS ecosystem has gone from import-reliant to self-sustaining dominance.

00:10:16: Totally.

00:10:17: They're leading in core domains, like Lidar with Huawei and Hussai, and in advanced automated parking.

00:10:23: The whole definition of a tier one is being rewritten over there.

00:10:25: And this global complexity forces a really tough trade-off at the engineering level.

00:10:29: It does.

00:10:30: Karthik and Rajan talked about this conflict between functional safety, ISO, two, six, two, six, two, two, and cybersecurity, ISO, two, one, four, three, four.

00:10:38: Safety wants redundancies.

00:10:39: security wants a smaller attack surface.

00:10:41: With geopolitics in the mix, you have to make structured architectural trade-offs and not just patch things later.

00:10:46: So the takeaway for Europe from Kirsten Heinecke and Professor Kambis Viscini is basically a call for urgency and partnership.

00:10:54: That's it.

00:10:55: More cross-company cross-border collaboration to close these technology gaps before it's too late.

00:11:01: So if we look across all five themes, the message is clear.

00:11:04: It's all about execution.

00:11:06: The battles for supply chain, for performance, for sovereignty, they're being fought and won right now in the software layer and in the silicon strategy.

00:11:14: And that focus brings us to our final thought.

00:11:16: Marcus Rettstep put it so well.

00:11:19: The future of mobility, of safety, of driving pleasure, it's not being decided on an assembly line, it's being decided in a git repository.

00:11:26: Everything is code.

00:11:27: Everything

00:11:27: is code.

00:11:28: From architecture to compliance to AI models.

00:11:31: He says it's the new DNA of the industry, and it's really the only way to manage all this complexity we've just been talking about.

00:11:37: So the question for you, as you look at this new global benchmark, is this.

00:11:42: What critical processes in your own organization, be it architecture, testing, compliance, still rely on static documents, on manual steps, on human interpretation?

00:11:52: And maybe more importantly, how quickly can you shift all of that into deployable code to meet this new high velocity standard being set around the world?

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

00:12:04: Also check out our other additions on electrification of battery technology, future mobility and market evolution, and commercial fleet insights.

00:12:11: Thank you for joining us and remember to subscribe.