Best of LinkedIn: Next-Gen Vehicle Intelligence CW 11/ 12

00:00:00: Brought to you by Thomas Allgaier and Frenny's, this edition highlights key LinkedIn posts on next-gen vehicle intelligence in weeks eleven and twelve.

00:00:08: 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.

00:00:20: You can find more info.

00:00:20: the description

00:00:22: Imagine waking up Getting ready for work, walking out to your driveway and you just find that your car is completely bricked.

00:00:28: Oh no!

00:00:29: Yeah I mean the engine won't start The doors barely unlock And you are just stranded.

00:00:34: Here's a real kicker.

00:00:35: Your car didn't actually break down.

00:00:40: Wait, really?

00:00:40: Yeah.

00:00:41: A third party vendor that you have probably never even heard of got hacked overnight and last week That exact nightmare paralyzed like a hundred and fifty thousand drivers.

00:00:51: So what welcome to the deep dive.

00:00:53: we

00:00:53: are so glad You were joining us today.

00:00:55: We have basically come through The absolute top next-gen vehicle intelligence trends that Really dominated LinkedIn during calendar weeks eleven and twelve.

00:01:04: right just cutting Through all the noise.

00:01:05: Exactly we are skipping the fluff to give you the actual signal here.

00:01:09: Yeah, so we're gonna take a really fast-paced focused look at The massive shift toward open-source software platforms the rise of the AI driven vehicle just

00:01:21: huge right now

00:01:22: it really is and then we will get into the Incredibly gritty reality of continuous safety.

00:01:28: And finally, uh the hidden human bottlenecks of autonomy that you know?

00:01:32: The industry's desperately trying to ignore.

00:01:34: yeah

00:01:35: They really are.

00:01:35: let's just jump straight in as a foundation though because um The very architecture of modern mobility is totally shifting away from those proprietary walled off silos.

00:01:45: A massive story that was dominating conversations over the past two weeks, Google officially open sourcing Android Automotive OS for software defined vehicles.

00:01:54: You will usually see it referred to online as AOS-STV

00:01:58: Exactly

00:01:58: And executives like Matt Crowley and Sameer Samad have been extremely vocal about this on LinkedIn recently.

00:02:04: They're framing ecosystem shifting move.

00:02:08: Yeah, totally.

00:02:09: but I mean we really need to unpack why this matters so much to you listening at home because historically car makers have just been like terrified of software fragmentation.

00:02:20: oh

00:02:20: absolutely terrified.

00:02:22: and Samir Samad pointed out that this transition actually mirrors the start of the smartphone era.

00:02:28: by providing a common open foundation OEMs can finally just stop reinventing the wheel on basic connectivity in HVAC controls.

00:02:37: Right, and they can actually focus their engineering power on the user experience instead?

00:02:42: Exactly!

00:02:42: To really visualize mechanics of why this fragmentation has been such a nightmare for them look at the analogy.

00:02:49: Julian Emma shared his post.

00:02:50: I mean it is perfect... The

00:02:52: Lego

00:02:52: one?!

00:02:52: Yeah he described building car software like trying to assemble a giant Lego set, but all the pieces are sourced from dozens of completely different boxes.

00:03:01: Right which sounds awful!

00:03:03: It is because a breaking sensor from one supplier speaks totally different code language than like the infotainment screen form another supplier.

00:03:10: so because of that car makers are forced to spend massive amounts time and capital just integrating those pieces.

00:03:18: yeah you know writing these custom translation layers rather then actually innovating on the vehicle itself

00:03:24: which perfectly explains why Google's AAOS-SDV is being labeled across the whole industry as a Kpex killer.

00:03:32: Yeah, that makes sense!

00:03:33: Right because if you don't have to pay hundreds of engineers to write translation code for proprietary operating system You completely bypass those massive plumbing investments

00:03:42: right?

00:03:42: The plumbing has already done

00:03:43: exactly And Izer Erdem actually noted that Renault is already leveraging this platform for their upcoming vehicles and Qualcomm is actively helping to scale it by delivering a pre-integrated stack.

00:03:55: Wow!

00:03:56: Already?

00:03:57: Yeah, they're moving fast...and the mechanism here's crucial because this enables much faster over-the-air or OTA updates for the core car functions.

00:04:06: Right, so we are not just talking about updating a Spotify app on your dashboard?

00:04:09: No exactly!

00:04:10: We were talking about upgrading the actual mechanical and comfort functions of the car seamlessly.

00:04:15: I want to pause and challenge The long term reality Of that smartphone analogy you mentioned earlier though

00:04:19: Okay late on me

00:04:20: Because if you think back to the early smartphone wars, The hardware manufacturers thing Motorola or HTC who eagerly gave up their OS layer To a tech giant they largely just became commoditized.

00:04:34: Right that is true.

00:04:35: So If legacy OEMs Just hand over the foundational os layer to A Tech Giant like Google?

00:04:40: Just to save on integration costs do They risk becoming well Nothing more than the hardware shell for someone else's software ecosystem.

00:04:48: I mean, that is the existential dread keeping legacy automaker executives awake at night like

00:04:53: it has to be

00:04:54: because if The plumbing belongs to Google what?

00:04:57: Is your differentiator?

00:04:59: you basically have to compete entirely on the intelligence sitting On top of that plumbing.

00:05:03: right?

00:05:03: Because if an open-source OS Solves the plumbing of a software defined vehicle the industry Has already realized That just having good plumbing isn't enough.

00:05:13: We have to talk about the brain.

00:05:14: Yeah, we do and Sean Ceri pointed out an incredibly sobering timeline regarding this.

00:05:19: he noted that The entire software defined vehicle era which was really just About basic architecture and being able To update features after a car ships.

00:05:27: yeah He says it effectively ended in January.

00:05:30: wait Really?

00:05:31: Just like

00:05:32: that The industry language has already moved on.

00:05:35: Now it is all about the AI-defined vehicle or AAV.

00:05:39: Okay, let's clarify the difference there for the listener just so we are all in the same page.

00:05:43: Yeah

00:05:43: good idea.

00:05:44: So an SDV update improves a static feature right maybe?

00:05:47: It just changes the layout of your digital dashboard Sure But an AI update actually improves how the vehicle makes decisions and AD can reason It can adapt to its environment and it genuinely gets smarter after leaving the factory.

00:06:01: Exactly,

00:06:01: like...it learns that a specific intersection is always icy in the morning And just adjusts its torque distribution dynamically.

00:06:08: Right!

00:06:08: That's huge leap!

00:06:09: It IS!

00:06:10: Honestly nobody illustrated The aggressive power grab In this new ADV space better than NVIDIA at their recent GTC twenty-twenty six event.

00:06:19: Oh I saw some posts about that.

00:06:20: Yeah Augustine Friedl broke down Nvidia strategy Honestly terrified traditional tier one suppliers because NVIDIA is no longer just selling silicon ships, right?

00:06:31: They're doing so much more now.

00:06:32: Yeah they are providing the entire underlying Substrate for cars.

00:06:36: I mean there offering the Hyperion ten hardware The Alpameo One point five AI brain and the Halo safety OS.

00:06:44: Wow the whole package

00:06:45: the whole Package And Philip rash Mac this out using a brilliant concept.

00:06:48: he calls the AV sandwich

00:06:54: of the new value chain.

00:06:55: It really is!

00:06:56: So for the listener, in this sandwich a company like Uber controls the customer access there are the cost layer okay and then NVIDIA controls the actual driving technology on AI grain.

00:07:07: so they're at bottom right.

00:07:09: where does that leave a legacy OEM like Mercedes or Stalantis?

00:07:14: I'm stuck in middle

00:07:15: exactly their stuck right in the Middle.

00:07:17: They literally just supplying metal and rubber for robotexes Basically getting squeezed out of the high margin software revenue by these tech giants on both sides.

00:07:28: Wow, that is a highly precarious position to be in.

00:07:31: it really is.

00:07:32: and

00:07:32: at that same event Nvidia's CEO Jensen Huang Stood on stage.

00:07:37: And essentially called this The chat GPT moment for self-driving cars.

00:07:41: Oh

00:07:41: boy

00:07:42: Yeah.

00:07:42: his premise Is that AI models can now generalize and reason through complex driving tasks rather than just relying on like millions of lines have hard-coded, if then rules that human engineers created.

00:07:55: I mean...I have to throw a massive red flag on the hype to be honest.

00:07:58: Absolutely!

00:07:59: And H-Main Wino posted really fantastic takedown for that exact phrase on LinkedIn.

00:08:04: calling it chat GPT moment is like super seductive for investors but incredibly dangerous in reality.

00:08:10: Yeah..the stakes are just different.

00:08:12: Right Let's look at mechanics of why ChatGPT works because If large language model hallucinates a wrong answer.

00:08:21: The consequence is trivial, maybe you get a really weird recipe for chocolate chip cookies right.

00:08:26: no big deal

00:08:26: but a hallucination at seventy miles per hour on a crowded highway.

00:08:30: that is fatal.

00:08:31: yeah that is terrifying.

00:08:33: and mangrino asks a critical question we all need to consider here are we confusing ai's incredible perception abilities with actual validated safety?

00:08:43: Man, that tension really exposes the fundamental difference between probability and determinism.

00:08:47: Exactly!

00:08:48: Because neural networks... I mean they are extraordinary at perceiving their environment.

00:08:52: They can look a blurry pixelated shape in the fog And guess with like ninety-nine percent probability That it is pedestrian.

00:08:58: But you cannot just be probably right when you're merging into fast lane of traffic.

00:09:03: You need determined guaranteed safety.

00:09:05: Yeah we do.

00:09:06: And that lethal difference between Probably Right and Guaranteed Safe brings us to the absolute nightmare of how governments are actually supposed to regulate these evolving AI machines.

00:09:15: Yeah, just think about the car sitting in your driveway right now... How does a government regulator even begin to certify a vehicle that can literally change its underlying decision-making code overnight?

00:09:27: Ugh short answer!

00:09:28: They don't know HOW.

00:09:29: Right.

00:09:30: And Ozychorectomere highlighted this structural paradox perfectly, he points out that OTA updates have completely broken the traditional concept of an approved system.

00:09:40: Oh!

00:09:40: That makes sense.

00:09:41: Right.

00:09:42: Historically regulators validated a fixed physical architecture you crash test to car You check breaks and approve it then just never changes.

00:09:50: Yeah It was simple.

00:09:51: But now they are forced to approve dynamic systems where features are activated on the fly, and the configurations evolve across millions of vehicles simultaneously.

00:10:00: So regulators are no longer validating a stable finished product—they're basically forced to validate mere moments in

00:10:07: time.".

00:10:07: Wow!

00:10:08: Which brings us right back.

00:10:11: Because the threat to these dynamic systems isn't just a rogue AI making a bad driving decision.

00:10:17: It's the vulnerability of the entire interconnected supply chain, because Chaudhari shared details about that hack and it honestly sounds like Hollywood thriller!

00:10:27: So an aftermarket breathalyzer company called Intoxalock suffered back-end cyber attack.

00:10:34: And because those ignition interlock devices are digitally connected to the vehicle's core systems, that single breach of a third-party server physically disabled one hundred and fifty thousand cars across forty six states for eight straight days.

00:10:48: Eight days?

00:10:49: I mean just Processed.

00:10:50: the cascading impact of it is huge.

00:10:52: people couldn't get their kids to school They couldn't Get to their shifts at the hospital all because a vendor they probably didn't even know was in Their car software stack went down

00:11:01: exactly.

00:11:02: It fundamentally shatters The illusion of vehicle ownership, it proves that In a highly connected ecosystem your two-ton vehicle Is Honestly, only as safe the weakest link in your digital supply chain.

00:11:15: Which is exactly why Abdul Aliyo emphasized that compliance standards aren't just some expensive bureaucratic chore engineers love to complain about... He argues that espice is essential engineering infrastructure.

00:11:31: He actually uses a really great analogy for this.

00:11:33: You don't build a massive high-speed highway and then skimp on the asphalt just to save a few bucks.

00:11:38: No, you definitely don't

00:11:39: right.

00:11:40: in a software defined vehicle The compliance frameworks are the asphalt.

00:11:43: they dictate the rigorous testing And documentation required before a single line of code ever touches a moving vehicle.

00:11:50: Wow

00:11:50: Yeah

00:11:51: because if there are potholes in your compliance Your beautifully designed AI software It's just gonna crash when it hits the real

00:11:58: world.

00:11:58: Yeah, that is so true.

00:12:00: But we also cannot get so lost in the cloud architecture and the software compliance That we forget these are still multi-ton kinetic objects operating in the real physical world.

00:12:10: right physics Still apply

00:12:11: exactly despite all this talk about AI brains Physical hardware and sensors still rule the road Partie.

00:12:17: Goyleman and Sony Andrews Jober-Doss both reminded the industry of a very sobering physical reality.

00:12:24: No algorithm, no matter how incredibly advanced it is can fix a blind spot that's caused by a clump of mud on camera lens

00:12:31: Man!

00:12:31: That is such vital pivot because if an AI hallucination is dangerous The AI basically needs flawless eyes.

00:12:40: Software cannot correct poor physics

00:12:42: Precisely.

00:12:43: When you are driving in zero visibility rain or heavy fog, optical cameras are essentially useless.

00:12:49: Right It is the radar sensors which are processing upwards of a hundred million data points per second by The way that take over one

00:12:56: hundred million.

00:12:57: Yeah

00:12:58: They are the unsung heroes doing the quiet unglamorous work of tracking targets and keeping drivers alive when human eyes fail.

00:13:06: Great software amplifies good data but it absolutely cannot rescue bad data.

00:13:11: So this creates a really massive friction point then, doesn't it?

00:13:14: Oh

00:13:14: definitely.

00:13:14: Because if the software is infinitely updateable in the cloud but the physical hardware like the cameras and radar or compute limits are permanently fixed at factory our OEMs just creating multi-constraint bottleneck.

00:13:26: Yeah that's the big question.

00:13:27: Vinicius Tadduzain brought us up.

00:13:29: He noted that entire sdv development pipeline is stalling out right now And not because of one single bottleneck Like bad code.

00:13:38: But due to these misaligned constraints You have agile software ambitions constantly slamming into physical testing limits and rigid legacy supplier ecosystems.

00:13:48: And that exact tension between Agile Software and Rigid Legacy Infrastructure, That perfectly explains the massive execution gap we are currently seeing between European and Chinese automakers.

00:14:00: So it really does.

00:14:01: It is fascinating to watch this play out on a global scale because its not that European engineers lack talent you know?

00:14:09: purely a gap in execution and tooling.

00:14:11: Youchaluo hit the nail on the head regarding this cap.

00:14:14: if you're an engineer listening to This, you will definitely feel this in your bones.

00:14:17: Oh

00:14:17: yeah prepare yourself.

00:14:18: He pointed out that European tier one suppliers are hiring thousands of top-tier software developers To catch up in the AI race.

00:14:24: Okay

00:14:25: sounds good so far

00:14:26: Right, but then they force those brilliant engineers to spend sixty percent of their work week updating JIRA tickets and manually filling out Excel traceability matrices just to prove compliance.

00:14:38: Sixty

00:14:38: percent?

00:14:39: That is insane!

00:14:41: It is...they are paying premium salaries for what's basically administrative data entry.

00:14:46: Meanwhile, Chinese EV makers like NIO and BYD have built their entire tool chain specifically for the SDV era.

00:14:55: They automate their compliance directly through their CICD pipelines?

00:14:59: Okay

00:14:59: let's explain how that works with a listener.

00:15:01: Sure

00:15:01: So continuous integration means that when a developer writes a line of code to adjust vehicles breaking logic The system automatically runs it thru thousands of simulated physics tests Automatically, automatically.

00:15:16: The compliance is proved by the machine not By a human filling out of spreadsheet.

00:15:20: so their engineers get to spend Their time actually innovating

00:15:23: and the real-world result Of that tooling difference is staggering.

00:15:27: Alex Elliott shared A striking example of what this Actually looks like in practice.

00:15:30: Oh I read This post.

00:15:31: it was wild.

00:15:32: yeah.

00:15:32: So elite motor C ten experienced a sharp unexpected Braking glitch while driving on the Autobahn In Germany.

00:15:40: okay

00:15:41: And the driver who actually happened to be an executive for the company, emailed The Engineering team back in China before stepping into an afternoon meeting.

00:15:49: Right!

00:15:49: By the time he walked out of that meeting just a few hours later... ...the Chinese engineers had isolated the issue written-the-cunged and beamed an OTA software fix directly to the car completely smoothing out the braking behavior

00:16:02: Man.. That is almost incomprehensible for a legacy automaker.

00:16:06: It

00:16:06: really is

00:16:07: A structural fix in HOURS that would traditionally take weeks, if not months of bureaucratic approvals.

00:16:13: They call it China speed and It is unequivocally the new global benchmark.

00:16:17: Wow!

00:16:18: But you know even If The Rest Of The World Perfects Their Software Speed And Matches That Automated CICD Pipeline There Is Still A Massive Hidden Void When It Comes To Scaling True Autonomy.

00:16:30: Yeah, this is where the conversation gets really counterintuitive because we constantly talk about driverless cars but we completely forget about humans operating behind the curtain.

00:16:40: Yes exactly

00:16:41: Juan Carlos Aguilera ran a math on his LinkedIn and numbers are mind-blowing if you look at recent partnership deals Uber aligning with Rivian Waymo scaling up their fleets bold expanding in Europe, the industry is actively targeting a deployed fleet of over one hundred fifty thousand autonomous vehicles.

00:17:00: In.

00:17:02: But the thing is, autonomous vehicles aren't truly autonomous.

00:17:05: Wait really?

00:17:06: No they require constant remote supervision.

00:17:09: if an AI gets confused by like a complex construction zone or uh traffic cop using hand signals it just stops and pings a human to take over

00:17:17: right so to actively run one hundred fifty thousand AVs.

00:17:21: Aguilera calculates that industry needs workforce of well over a hundred thousand humans.

00:17:25: A hundred thousand!

00:17:26: Yes We are talking about remote operators, sensor calibrators hub crews and specialized maintenance techs.

00:17:34: These are highly skilled jobs that require intense concentration you know managing long periods of absolute boredom.

00:17:40: they're just punctuated by seconds.

00:17:42: high stakes crisis when the AI throws its hands up.

00:17:46: Yeah!

00:17:46: That sounds incredibly stressful.

00:17:48: And a huge problem is those jobs barely exist today.

00:17:51: There's no pipeline for this talent.

00:17:54: right now

00:17:54: It is a massive scaling bottleneck that basically nobody is budgeting for.

00:17:59: And Aguilera actually compares this impending workforce crisis to the nineteen eighties offshore oil industry.

00:18:06: Oh,

00:18:06: it isn't interesting comparison.

00:18:07: Yeah

00:18:07: right because back then when deep water drilling exploded The oil companies didn't have a trained workforce either.

00:18:13: It was completely the Wild West.

00:18:15: Wow So these fiercely competing oil companies actually had to come together To create the OPI TOE standard.

00:18:21: it Was a unified industry-wide certification for offshore workers And Aguilera argues that AV companies absolutely must do the same thing.

00:18:29: That makes

00:18:29: total sense

00:18:30: Right, they need a shared transferable credential For remote operators or they simply will not have the humans required to monitor their multi-billion dollar rogotaxi fleets.

00:18:41: Let's

00:18:42: look at how this shift is already reflecting in the physical design of cars themselves, too.

00:18:46: Libya Montoya made a really brilliant observation about Tesla interior design roadmap.

00:18:51: Oh I saw that!

00:18:52: Yeah if you look at the physical progression They moved from a standard circular steering wheel model three To that highly controversial rectangular yoke In the Cybertruck.

00:19:03: Right

00:19:03: people had alot opinions on it.

00:19:05: they sure did, and now to the recently announced cyber cab which lacks any steering control at all.

00:19:11: Crazy!

00:19:12: It is not just an aesthetic design choice right?

00:19:14: it's a physical manifestation of capability graduation.

00:19:18: The physical driver is actively being phased out.

00:19:21: But to your point about the workforce, The human isn't disappearing at all.

00:19:25: They are just being moved into a remote control center halfway across

00:19:36: with the cognitive act of remote supervision.

00:19:38: We

00:19:38: really are,

00:19:39: but amidst all this hyper advanced tech you know?

00:19:42: The AI models reasoning like humans, the robo taxi fleets, the remote operators there is a really harsh reality check.

00:19:50: we need to mull over before we wrap up today.

00:19:52: okay

00:19:52: bring us back down to earth.

00:19:54: yeah what does the actual person buying a car think about all of this?

00:19:59: well

00:19:59: Vanessa Vu shared some incredibly grounding data from Deloitte.

00:20:04: When you actually survey consumers and ask them what they want, They are not asking for AI-defined vehicles.

00:20:09: Really?

00:20:10: Nope!

00:20:10: They still prioritize the absolute fundamentals safety affordability And The total cost of ownership.

00:20:16: I make sense

00:20:17: Yeah...and only about twenty five percent Of customers Are even willing to pay For software subscriptions in their cars.

00:20:22: So it really forces us To ask our automakers Becoming way too obsessed with building AI defined Vehicles as like a marketing buzzword for Wall Street Right when they should instead be using AI quietly behind the scenes to build customer-defined mobility that is simply cheaper, more reliable and ultimately safer.

00:20:41: That...is a phenomenal point!

00:20:43: Are we building tech for the sake of tech or are we actually solving the driver's problems?

00:20:48: Exactly.

00:20:49: And before we go here's one final thought for you to chew on.

00:20:51: that builds on everything we've discussed today.

00:20:54: We have talked endlessly about vehicles getting exponentially smarter learning to reason and updating their code from the cloud.

00:21:01: But what happens when physical infrastructure doesn't keep up?

00:21:05: If we are deploying trillion-dollar AI defined vehicles onto analog, twentieth century roads with crumbling asphalt fading painted lines in broken stoplights who is ultimately to blame if this system fails?

00:21:17: That's a great question!

00:21:19: Is it highly regulated software or is it bankrupt city planner that couldn't afford repaint crosswalks?

00:21:27: It is a huge collision between the future of software and past public infrastructure, And mobility industry will have to answer for it soon.

00:21:35: They absolutely well!

00:21:37: If you enjoyed this episode new episodes drop every two weeks.

00:21:40: Also check out our other additions on electrification & battery technology Future Mobility Market Evolution & Commercial Fleet Insights.

00:21:48: Thank You so much For Joining us On This Deep Dive.

00:21:50: Make Sure To Subscribe.

00:21:51: We Will See You Next Time.