Best of LinkedIn: Next-Gen Vehicle Intelligence CW 27/ 28

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

00:00:07: Frenness supports tier one automotive suppliers with early stage market validation for their R&D efforts By combining secondary research direct OEM expert interviews And facilitated customer meetings.

00:00:18: You can find more info In the description.

00:00:21: Welcome everyone

00:00:21: Yeah welcome So.

00:00:23: we are so glad your joining us For This Deep Dive Today.

00:00:26: our mission is To distill absolute top next-gen vehicle intelligence trends that we've been seeing across LinkedIn over the past two weeks.

00:00:34: Right, and there is a lot to cover.

00:00:36: I mean really unpacking some massive core themes.

00:00:38: today We're seeing this rapid consolidation of software defined vehicle architectures or SDVs plus all the geopolitical and regulatory hurdles That are basically rewriting the rulebook right now.

00:00:49: Yeah And of course The industry's critical leap from just software define mobility To actual AI Define Mobility.

00:00:55: It's honestly a wild time to be tracking this stuff.

00:00:58: I mean, the mobility industry is moving so fast right now that an architecture can become completely outdated before the car even hits the dealership lot.

00:01:06: Oh absolutely it's-it's moving at breakneck speed!

00:01:09: So let's just jump right in.

00:01:10: i want to start with that fundamental shift happening beneath The Hood.

00:01:13: uh Jaywant DeMahajan posted this really great observation recently.

00:01:17: he noted that for years cars have just accumulated ECUs.

00:01:21: You know electronic control units right and he compared it to like a messy closet just accumulating junk over the decades.

00:01:29: you want A new feature.

00:01:31: You'd, just toss another ecu in there which

00:01:33: is Just a nightmare for engineering

00:01:35: exactly so jay watts point was that The true sdv winners aren't going To be the ones you know, adding the most lines of code.

00:01:42: The winners are the ones who were aggressively using zonal architecture to remove wiring cut weight and just strip out that massive hardware complexity.

00:01:50: I totally agree.

00:01:51: And To build on That Yinesh and i do share this really striking observation about Volkswagen in China.

00:01:58: Oh

00:01:58: yeah!

00:01:58: I saw that one.

00:01:59: Right

00:02:00: Because while everyone else was obsessing over VW's overall sales numbers In that market Yineshe Was looking at their engineering speed which is the metric that actually matters right now.

00:02:10: Right, how fast can they iterate?

00:02:12: Exactly!

00:02:13: So VW actually stripped out thirty percent of their ECUs.

00:02:16: Wow... Thirty percent.

00:02:19: Yep, thirty percent.

00:02:20: and by doing that They cut their development costs up to fifty percent.

00:02:30: That is insane for Legacy Auto.

00:02:32: Right,

00:02:33: and the new electronic architecture that CEA... Yeah!

00:02:36: ...that was done.

00:02:36: in eighteen months they basically halved their legacy timeline.

00:02:40: Okay I have to push back a little bit here though because I mean stripping out of hardware sounds great on paper right?

00:02:45: It's like upgrading from room full bulky filing cabinets just single smartphone.

00:02:50: yeah it's good analogy

00:02:51: but consolidating all that raw computing power into few central nodes i means must create massive heat.

00:02:58: Oh you are spot-on.

00:03:00: That is the huge physical roadblock right now.

00:03:03: And it's exactly what Matt Damocino brought up in his insights, when you have centralized processors running sensor fusion plus The AI Plus basic vehicle controls all at once It turns thermal management from just a basic component level problem into a critical system-level challenge

00:03:21: because Everything relies on that one node.

00:03:23: Exactly, if it overheats It impacts everything from your EV battery range all the way up to your AI capability.

00:03:29: So we're essentially replacing a wiring problem with a thermodynamics problem.

00:03:33: pretty much Yeah.

00:03:35: speaking of that centralized architecture We have to talk about the software That's actually running on.

00:03:39: let's do it

00:03:40: so such a broader.

00:03:41: Proudhon pointed out this massive shift happening in the operating system battleground.

00:03:46: For a long time we've thought of Android Automotive OS or AAOS as just the impotement layer,

00:03:51: right?

00:03:52: Yeah like maps and music.

00:03:53: Right

00:03:54: but Satya Prada noted that as of March twenty-twenty six it is expanding way beyond screens.

00:04:00: It's moving into safety relevant controllers.

00:04:02: wait really Like what kind of controllers?

00:04:04: things like seat actuators to climate systems And even the instrument clusters.

00:04:08: okay Wait hold on Open source Android handling functional safety.

00:04:12: Yeah, I mean how does an OEM manage ISO to six two six two compliance when you have Google pushing it upstream update?

00:04:18: Like what if an update breaks your instrument cluster while you're driving?

00:04:21: It

00:04:21: sounds terrifying right but that is the exact problem The industry is solving right now.

00:04:27: Pedro Vero and messiahs boinegg highlighted a new white paper from ETS in Bosch That tackles this head on.

00:04:33: oh okay What's the work around?

00:04:36: the solution they are pushing is a platform of platforms approach.

00:04:40: So basically you pair Google's AAOS which handles all those dynamic cloud native features with something like Eclipse S-Core and Eclipse handles the highly deterministic safety critical workloads.

00:04:52: Ah, I see.

00:04:52: so You basically sandboxed the fun stuff from the critical stuff

00:04:55: exactly.

00:04:56: they run side by side but strictly isolated.

00:04:58: That makes total sense.

00:04:59: And, you know watching the software unification happen it's just moving so incredibly fast especially in the commercial space.

00:05:05: Oh!

00:05:05: The trucking side is fascinating

00:05:07: right now.

00:05:07: Right Robert O had this great post about Trot and OneOS.

00:05:11: In just twelve months...twelve months they onboarded two hundred engineers and built a single software platform.

00:05:19: Wow And it runs trucks across four major distinct brands.

00:05:23: You've got Scania, MAN International and VW Truck & Bus all running on the same unified OS.

00:05:30: That level of alignment across four legacy brands in one year is just phenomenal

00:05:34: It really is.

00:05:35: But you know while the engineers are out there brilliantly solving this software puzzle The geopolitics and regulators are throwing up massive roadblocks.

00:05:44: Oh yeah...the compliance side

00:05:47: Brutal.

00:05:47: It's getting so complicated.

00:05:49: Ambrose Conroy had the stark warning about The new US connected vehicle rule specifically fifteen CFR part seven ninety one.

00:05:57: right under this Rule poll star is officially banned from the u.s.. Market starting with model year twenty-twenty-seven

00:06:03: Which sent shockwaves through?

00:06:04: The industry.

00:06:05: and if you look at the context provided by Jeffrey Gertz And Brandon Barry it really clarifies why This happened.

00:06:10: because it wasn't About where the car was physically built, right

00:06:12: I mean pole star builds cars in South Carolina

00:06:15: Exactly.

00:06:15: It was strictly about the Chinese ultimate corporate ownership, Geely owns them and it was their inability to fully decouple from software that were sourced by an adversarial nation.

00:06:25: But wait Volvo is also owned by Geely.

00:06:28: Yes Why do they get a pass?

00:06:29: Well

00:06:30: its not like they got a pass Its that Volvo complied?

00:06:33: Yeah Because there US marketer so huge it actually justified massive cost of ripping out and replacing that software For Polestar.

00:06:44: The compliance math just didn't work out.

00:06:46: Wow, so the code literally decides if a car can exist in market?

00:06:50: Exactly!

00:06:51: And it's not just international trade rules.

00:06:52: right.

00:06:53: regulation is hitting the actual dashboard design now.

00:06:55: Yeah

00:06:56: regulators getting very hands-on.

00:06:57: Qing Yi Chen highlighted the UK Automated Vehicles Act, this one blew my mind.

00:07:02: It makes it a literal criminal offense in the

00:07:05: U.K.,

00:07:05: to use terms like self-driving or autonomous...to market L zero to L two at ES

00:07:10: vehicles.".

00:07:11: So no more misleading marketing for advanced cruise control?

00:07:15: None!

00:07:15: It's a criminal offence.

00:07:16: and then over in China Adam Bernard noted this draft law that is mandating physical buttons those completely screen-only controls.

00:07:29: I think a lot of drivers will be very happy about that, but you know this regulatory tightening connects directly back to data privacy too.

00:07:35: Oh

00:07:35: for sure.

00:07:36: Matt Scaffo posted about the GM OnStar litigation recently which is perfect example.

00:07:42: GM actually paid a twelve point seven five million dollar penalty in California

00:07:46: just for data

00:07:47: sharing.

00:07:48: yeah for allegedly sharing connected vehicle driving data.

00:07:51: so things like hard braking and speeding they were sending to third-party data brokers.

00:07:57: Who were then selling it to insurance companies?

00:07:59: Exactly,

00:08:00: and all of this was happening without clear explicit consent from the drivers.

00:08:05: That is so sketchy.

00:08:07: imagine your insurance premium spiking because you had to break hard for a squirrel And you have no idea how they even know about.

00:08:13: It's exactly why data governance Is becoming a massive liability issue.

00:08:17: Yeah

00:08:18: So okay let's connect the dots here.

00:08:19: we've got consolidated hardware.

00:08:20: We've got unified software platforms.

00:08:26: What is the final form of all

00:08:33: this?

00:08:36: SDVs, the software-defined vehicles to ADV's.

00:08:39: AI defined vehicles?

00:08:40: Exactly where AI isn't just an add on it actually drives The continuous learning and the core perception of a car

00:08:47: knuckle de gall from qualcomm had A great phrase for this.

00:08:49: he called at the era Of physical ai.

00:08:52: oh i like that because It's moving off of our screens And into the real physical world

00:08:56: right and you see the big players preparing For this in a bayou.

00:09:00: a foe noted That vw's carry out Just opened This massive one-thousand person campus in Berlin.

00:09:06: A

00:09:06: thousand people,

00:09:07: a thousand people and it is dedicated entirely to AI powered mobility.

00:09:11: that is huge investment.

00:09:13: but let's do a bit of reality check here on full autonomy because we've been promised RoboTaxis for awhile.

00:09:18: True!

00:09:19: Jamie Bennett pointed out there are actually nearly fifteen hundred commercially registered autonomous vehicles in Texas right now mostly Waymo and Vareed.

00:09:28: That'a lot metal on the road

00:09:30: It is.

00:09:30: And for hot Babakan even published this crazy field guide, he took over a hundred and seventy Robotaxi rides just to analyze the different tech stacks in the real world

00:09:40: which has fantastic research.

00:09:42: but we have to temper that excitement with some hard data.

00:09:45: Yeah Junko Yoshida brought up some really sobering stats from Austin.

00:09:48: okay despite all the millions of miles driven Austin recorded forty seven separate incidents In just three months where Robotaxis interfered with emergency responders.

00:09:59: Oh

00:09:59: man Forty-seven in three months.

00:10:01: Yeah,

00:10:02: and it raises this really critical almost philosophical question The AV companies expect first responders to learn how to deal with their cars.

00:10:10: But when will the AVs learned too seamlessly interact?

00:10:13: With human emergency crews?

00:10:15: right because an AI doesn't understand a firefighter waving It away with their hands

00:10:18: exactly.

00:10:19: you can't read Human context.

00:10:20: yet

00:10:21: that is such A fascinating bottleneck.

00:10:23: You know I want to leave you the listener.

00:10:24: what's final thought tomorrow?

00:10:26: over building on all

00:10:27: of us go for

00:10:28: Think back to Yanesh and I do's point about Volkswagen earlier.

00:10:31: The

00:10:31: twenty-four month development cycle?

00:10:33: Yeah, they solved the software defined vehicle to completely cut down their development time.

00:10:38: but physical assembly machines still take eighteen or twenty four months of design and commission.

00:10:44: Oh right!

00:10:46: So car got faster in designing but factory didn't...the next massive frontier isn't just a software defined car.

00:10:54: it is a software designed assembly

00:10:56: line.

00:10:57: Yeah, that is the real bottleneck.

00:10:58: Exactly!

00:11:00: If you enjoyed this episode new episodes drop every two banks.

00:11:03: also check out our other editions on electrification and battery technology future mobility in market evolution and commercial fleet insights.

00:11:10: Thank You so much for hanging out with us For This Deep Dive.

00:11:12: don't forget to hit subscribe.

00:11:14: Thanks for listening.

00:11:14: everyone see y'all next one.