Best of LinkedIn: Autonomous & Mobility Markets CW 33/ 34

00:00:00: provided by Thomas Allgaier and Frennus, based on the most relevant LinkedIn posts about autonomous mobility markets in calendar weeks thirty-three and thirty four.

00:00:09: 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.

00:00:21: you can find more info.

00:00:22: Imagine a prankster shutting down an entire city block using fifty empty robotexes as like, a coordinated physical DDoS attack.

00:00:32: Wow!

00:00:32: Right and that is not a scene from some cyberpunk novel That actually happened in San Francisco just this week.

00:00:38: Yeah so today we're looking at the top autonomous and mobility markets trends across LinkedIn And really unpacking what these raw real-world deployments are you know telling us about The future of how he actually moves.

00:00:49: Okay let's unpack.

00:00:51: Our mission for today's deep dive is to analyze four major industry shifts.

00:00:55: So first, we're looking at the brutal operational reality as Robotex has transitioned from these you know cute pilot programs into actual paid commercial services.

00:01:04: second will examine the global regulatory chess game that kind of shaping where this tech actually deploys.

00:01:10: Third, we need to dig into the unit economics and street level friction of micromobility.

00:01:16: Right!

00:01:16: The scooters and bikes

00:01:18: Exactly.

00:01:19: And finally We'll look at how human centric design is finally replacing a century of driver-first vehicle architecture.

00:01:26: Yeah, and the timing here?

00:01:28: it's just critical.

00:01:29: The industry is currently crossing this commercial threshold And as it does...the whole nature of the competition is fundamentally shifting.

00:01:36: I mean..The war no longer about who can build the most advanced self driving car in a vacuum.

00:01:41: It's a war to own an overarching mobility network itself.

00:01:45: Samarth Bali actually laid out really good framework for this transition on LinkedIn.

00:01:49: He pointed out that the robo taxi war has fractured into three distinct strategic pillars.

00:01:54: Yeah, so you have Waymo whose primary goal is to own the intelligence Like the autonomous driver itself right?

00:02:00: The brain

00:02:01: exactly.

00:02:02: and they are already executing roughly five hundred thousand weekly paid rides Which is huge

00:02:07: yeah massive.

00:02:08: then You have zoos And their betting heavily on owning the machine by building this bespoke ground-up robot.

00:02:14: and finally you have uber whose entire play is about owning the network and that customer relationship.

00:02:21: I'd actually take that analysis a step further because if you look at the long-term unit economics, the autonomous vehicle hardware will likely become commoditized over the next decade.

00:02:33: You think so?

00:02:34: Oh absolutely!

00:02:35: The real scarcity in the mobility market won't be vehicles it'll

00:02:43: Interesting.

00:02:43: Yeah, so Uber's bet is that passengers well they won't want to open five different apps To find a Waymo or a Zooks or whatever.

00:02:50: They'll just open the dominant aggregator app hit ride and the dispatch AI will Just deploy Whatever vehicle makes The most economic sense for That specific route

00:02:59: I see

00:02:59: yeah?

00:02:59: They just Wanted be the operational layer For all physical mobility.

00:03:03: that Makes the hardware strategy Incredibly high stakes then especially for A company like zoops Because according to some recent updates from Edward Ludlow and Ming Ma, Zooks is officially launching paid commercial services in Las Vegas.

00:03:18: Oh wow!

00:03:18: Vegas okay?

00:03:19: Yeah but the critical detail here Is that they received The very first NHTSA Commercial Exemption for a purpose-built vehicle Lacking a steering wheel or pedals.

00:03:29: No steering wheel at all

00:03:31: Nothing.

00:03:31: I mean it's like the industry is finally moving From the science fair project phase into running A real commercial airline.

00:03:37: It is a massive architectural leap because when you physically remove the steering wheel, You're eliminating the fallback mechanism entirely.

00:03:45: Right?

00:03:46: The software can never hand control back to human in the cabin

00:03:49: Because there's no way for them take over

00:03:51: Exactly so.

00:03:51: that means entire system architecture like from break by wire redundancy To localized power nets.

00:03:58: All of it has be completely self-reliant From start to finish.

00:04:01: But I mean let me push back on operational readiness.

00:04:05: Are these fleets genuinely prepared for the wild unpredictability of the real world?

00:04:10: Because, at that moment you push them into these edge cases.

00:04:13: The limitations on current machine learning models become pretty glaringly obvious.

00:04:17: Ben Hubbard highlighted a perfect example of this.

00:04:20: Oh!

00:04:20: The recall...

00:04:21: Yes.

00:04:21: Waymo just had to recall nearly four thousand vehicles.

00:04:25: and

00:04:25: why?!

00:04:26: because their multi-million dollar sensor suites were driving straight into flooded roadways, like a hazard that the teenager with fresh license knows to avoid.

00:04:35: Yeah it really highlights massive gap in sensor fusion versus semantic understanding.

00:04:40: What do you mean?

00:04:41: So human driver implicitly understands context of a flooded curb right.

00:04:46: they see water rippling and cars diverting.

00:04:49: just apply heuristic turn around.

00:04:52: don't drown.

00:04:52: Right exactly.

00:04:53: But the autonomous system's lidar, it simply registers the surface of water and the neural net.

00:04:59: despite all its massive training data.

00:05:01: It just failed to assign a negative weight to that standing water.

00:05:05: so it just proceeded blindly right into the hazard.

00:05:08: That is crazy!

00:05:09: And I saw another edge case perfectly illustrates this gap from Steven Snyder.

00:05:15: A robotaxi in Las Vegas drove directly onto an active fire scene.

00:05:20: Oh no

00:05:21: because the perception stack didn't properly classify the smoke.

00:05:24: So it just braked hard and immobilized itself right in the middle of the emergency zone, wow!

00:05:30: And the critical failure wasn't just the sensors missing the smoke... It was the complete lack of an operational protocol Like a remote operator had to literally connect and back the car out of the scene.

00:05:41: first responders need To know how to stop an empty vehicle.

00:05:44: And you know who'd call

00:05:45: see.

00:05:45: from a municipal perspective That is an unacceptable operational bottleneck.

00:05:50: Oh, totally.

00:05:50: I mean if you're a fire captain arriving at a chaotic multi-alarm fire You do not have the time to figure out How to stop in empty two tonne vehicle?

00:05:58: No or to hunt down some one eight hundred number for fleet dispatch

00:06:02: exactly.

00:06:03: Snyder's point Is that the industry so hyper focused on on the vehicle's perception software, but they're lagging years behind on the emergency HMI.

00:06:10: You know?

00:06:11: The human machine interface for the actual first responders

00:06:15: which brings us back to that physical DDoS attack you mentioned at the top of deep type.

00:06:18: right

00:06:19: yeah

00:06:19: Dan Lorman shared a prankster in San Francisco managed order fifty Waymos onto the exact same dead end street.

00:06:26: it

00:06:27: just wild think

00:06:28: about It really is created this massive pile up and forced company to disable rides until next morning.

00:06:35: I mean, how does a multi-billion dollar routing AI allow that to even happen?

00:06:40: Well

00:06:40: the system failed because the dispatch AI likely relies on localized density metrics.

00:06:46: That were just easily spoofed okay?

00:06:48: so The bad actor generated simultaneous highly clustered ride requests and that just overwhelmed the fleet's multi agent load balancing protocol.

00:06:56: Ah see yeah it completely exposed the vulnerabilities in How these centralized fleets authenticate and manage coordinated routing demands.

00:07:04: So athletes routinely driving into floods, getting stuck at fire scenes and being clustered by pranksters.

00:07:10: How many humans are actually sitting behind monitors bailing these algorithms out right now?

00:07:15: Well the ratio is actually much better than the headline suggests but the workflow is what really matters here.

00:07:21: Daniel Ebrey Marquet shared that a healthy remote assistance ratio today is between one to twenty and one to fifty.

00:07:27: Wait...really?

00:07:28: How was one person managing fifty autonomous vehicles?

00:07:32: What happens when a storm hits and ten cars encounter a flooded road at the exact same time?

00:07:37: So what's fascinating here is it's crucial to understand that remote operators are not driving the cars with the joystick.

00:07:44: Okay, It functions much more like an IT help desk ticket system.

00:07:47: Oh interesting

00:07:48: Yeah.

00:07:48: The onboard compute handles all the continuous real-time driving.

00:07:54: However, if the vehicle's trajectory confidence level drops below a safety threshold say it encounters a complex instruction zone.

00:08:01: It just can't map right?

00:08:02: It pauses safely and generates a task.

00:08:05: A remote human looks at the telemetry for that one specific vehicle For that moment only draws illegal bounding box or path around The obstacle closes the ticket in disconnects.

00:08:15: Oh wow Yeah.

00:08:15: So the one to fifty ratio is really a testament To how infrequently the operational design domain has actually breached.

00:08:21: Okay, that makes perfect sense.

00:08:23: The human is providing localized logic not localized control Exactly!

00:08:27: But because these localized logic failures have real-world consequences on city streets we're seeing a massive shift in how different global regions regulate deployment right?

00:08:38: Like operational friction is naturally mapping to regulatory friction.

00:08:41: Oh absolutely which is why the rollout in Europe looks nothing like the initial ride hailing boom a decade ago Right...the

00:08:48: whole move fast and break things era.

00:08:50: Exactly Timothy Papandrea noted that Waymo is executing a highly structured, almost diplomatic European entity strategy.

00:08:59: They aren't just dropping cars on streets.

00:09:01: they're setting up R&D centers in Switzerland their integrating deeply with German manufacturing and homologation standards And there heavily navigating EU policy in Brussels well before launch.

00:09:11: Yeah, I mean they have to play that game considering Europe's regulatory framework really prioritizes those structured safety cases over moving fast.

00:09:19: But while Waymo plays the long game in Europe there is a fascinating supply chain dynamic playing out on US soil.

00:09:25: Oh yeah!

00:09:25: The Zika imports.

00:09:27: Yes Dr.

00:09:28: Volker Kies and Ken Kwong pointed out that Waymo is currently importing thousands of Chinese-built Zikr EVs for its U.S.

00:09:35: Robotaxi fleet.

00:09:37: Yeah, the economics are just staggering.

00:09:40: Waymo has shipped over three thousand two hundred of these stripped zikr gliders through the port of Long Beach.

00:09:48: But because they're Chinese built They are subject to a massive one hundred twenty seven point five percent US import tariff.

00:09:55: Wait,

00:09:56: really?

00:09:56: Yes.

00:09:56: Waymo is taking a thirty-nine thousand dollar base vehicle and turning it into an eighty nine thousand dollar chassis before they even install a single LiDAR sensor or compute node.

00:10:06: But why would way mo willingly eat a massive one hundred twenty seven percent penalty on their hardware pipeline just to use these chases?

00:10:14: Is it time to market play Or the hardware's that much better?

00:10:18: Are there truly no domestic alternatives?

00:10:21: It comes down to architecture of Glider itself.

00:10:23: Waymo isn't buying finished cars.

00:10:25: They're buying stripped chassis with deep unencrypted access to the vehicles can bus, power distribution networks and thermal management systems.

00:10:33: Yeah because retrofitting a standard consumer vehicle With massive rooftop sensor pod And trunk mounted supercomputers is incredibly inefficient.

00:10:43: So paying tariff allows Waymo To deploy purpose built easily integrated platform today Rather than waiting three-five years to co develop domestic alternative.

00:10:53: Time to market is just trumping unit costs right now.

00:10:57: Wow, but it's interesting because Chinese automotive companies aren't content Just being hardware suppliers for US fleets.

00:11:02: No definitely not.

00:11:03: they are expanding aggressively.

00:11:06: Yanis Daga highlighted that companies like pony dot AI by do we ride and momenta Are treating the European Market as their primary global deployment ground?

00:11:15: And that expansion is happening rapidly.

00:11:17: Athena Browning noted that Biders Apollo Go!

00:11:19: Is already actively testing its systems in London, running right alongside Waymo and Wave.

00:11:24: Really?

00:11:24: London?

00:11:25: Yeah...and Christian Eckert pointed out a fascinating ecosystem play in Zagreb where Uber and Pony.ai are scaling up operations in partnership with the local operator called Verne.

00:11:35: so they're decoupling the stack.

00:11:37: One company provides the autonomous tech, another provides the dispatch network and a third manages the physical fleet operations in charging.

00:11:44: That's a

00:11:45: smart way to do it yeah but with that rapid influx of foreign operated fleets into European cities Merrick Van Zura offered a highly critical impartial perspective.

00:11:55: that kind of shifts the conversation away from vehicle safety.

00:11:58: oh The data privacy angle exactly.

00:12:01: he argues that while we debate vehicle safety we're ignoring the reality that these vehicles are essentially two-ton corporate controlled surveillance drones.

00:12:10: It's a stark framing

00:12:12: it is but they are packed with high resolution cameras, LiDAR and radar right mapping public spaces continuously.

00:12:19: And he points out that they're often sending data to Chinese entities.

00:12:22: Right, and it just highlights the immense data footprint of a robotexie which can generate you know several terabytes of mapping in telemetry data per day.

00:12:30: Yeah raising massive questions about data localization and municipal privacy.

00:12:34: yeah...and Just to be clear we are neutrally reporting with concerns raised here but it perfectly illustrates why geopolitical scrutiny is just intensifying so much right now.

00:12:45: Definitely And you can really contrast that rapid European expansion with the U.S.

00:12:50: regulatory stance.

00:12:51: Steve Greenfield noted that Nevada just heavily restricted Tesla's RoboTaxi permit application.

00:12:57: Oh,

00:12:57: I saw this!

00:12:57: If Tesla requested a permit for five thousand vehicles state regulators granted them a maximum of ten... Just

00:13:04: him?

00:13:04: ...just ten vehicles captured forty-five miles per hour on a strict corridor along the Vegas Strip.

00:13:09: With heavy supervision requirements regulators are officially slamming on the brakes.

00:13:14: It is a definitive signal that having advanced driver assistance software in consumer cars does not automatically grant you their regulatory permission to operate an unsupervised commercial fleet.

00:13:24: No, it's completely different.

00:13:26: burden of proof Absolutely.

00:13:27: By the way if your finding this deep buy valuable make sure subscribe so don't miss our future editions covering these market shifts.

00:13:34: Awesome, well we've spent a lot of time on two-ton autonomous machines but the exact same operational friction.

00:13:41: where Silicon meets The Street is playing out in the micro mobility space too.

00:13:45: Oh

00:13:46: here's where it gets really interesting because We're seeing a massive maturation in micromobility economics But also rising friction on the streets.

00:13:55: We finally have hard data answering that long-standing question of whether shared scooters and e-bikes can survive without endless venture capital.

00:14:04: Right, we got our first share some incredible financial data recently.

00:14:08: the profitability Question has basically been answered.

00:14:10: Lyme just posted three hundred four million dollars in quarterly revenue with a massive twenty seven point seven percent margin.

00:14:17: Wow yeah Pervoy is hitting record profits.

00:14:20: Meanwhile, DOT is reportedly struggling significantly with debt.

00:14:24: You know a twenty-seven percent margin on heavily abused street deployed hardware Is an astonishing operational feat.

00:14:30: It really is.

00:14:30: If

00:14:31: we connect this to the bigger picture, The divergence in these companies proves that profitability and shared mobility Is fundamentally an operations and logistics problem.

00:14:39: it's not a demand problem.

00:14:40: No!

00:14:40: The demand has proven.

00:14:42: I mean North American Shared trips just hit a record two hundred thirty seven million

00:14:46: Exactly...the operators who are surviving like Lyme They figured out the granular mechanics of margin expansion.

00:14:53: Like what?

00:14:54: Well They move to swappable batteries, meaning the scooter stays on street earning revenue while a technician in a van only swaps power cell.

00:15:02: Oh that makes sense!

00:15:03: Yeah.

00:15:03: and they utilize predictive repositioning to ensure vehicles are at transit hubs right as trains arrive.

00:15:09: The company's drowning-in debt is ones who just dumped hardware onto sidewalks And hoped utilization would magically outpace depreciation.

00:15:17: You can't hope for the best But managing two hundred and thirty seven million trips brings an enormous amount of friction regarding safety and user behavior.

00:15:27: Mark Williams shared data showing that emergency room visits for e-bike and scooter injuries have skyrocketed from under one thousand back in twenty seventeen to tens of thousands today.

00:15:37: And while

00:15:38: part of it is just a function scale, A researcher named Calvin Thigpen identified structural root cause in the business model itself.

00:15:46: Yeah he found that the paper minute pricing model, you know PAYG actively incentivizes unsafe writing behavior.

00:15:55: That makes total sense.

00:15:56: I mean it's exactly like washing a ticking taxi meter.

00:15:59: You are financially motivated to sprint through a yellow light or cut a corner just to save a few cents

00:16:06: Thick pens.

00:16:06: simulator studies proved exactly Writers on a paper-minute model will ride at higher average speeds and accept significantly more dangerous gaps in cross traffic than writers on privately owned bikes or subscription models.

00:16:19: Wow!

00:16:19: Yeah, the revenue structure of industry is actively generating risk on street

00:16:24: And municipalities are responding to that risk with brute force.

00:16:28: Colleen Heffernan noted that Toronto police just launched a massive enforcement campaign targeting illegal sidewalk riding after the city logged over five hundred collisions this year.

00:16:37: Good grief!

00:16:38: But she correctly points out, that enforcement only treats the symptoms of regulatory patchwork and an infrastructure mismatch.

00:16:45: I mean our street infrastructure simply wasn't built for vehicle that goes twenty miles per hour.

00:16:49: No you cannot enforce your way out of build environment mismatch

00:16:53: Exactly.

00:16:54: Which brings us to The Legislative Nightmare in New York City.

00:16:57: Dan Sarasi analyzed the situation, pointing out that NYC currently has seventeen different bills on the table trying to regulate e-bike safety.

00:17:05: Seventeen?

00:17:06: Seventeen.

00:17:07: But his core argument is its objective enforcement is impossible.

00:17:11: You can't expect a patrol officer To conduct a split second audit of a vehicle's motor wattage as it just zooms past them in a bike lane.

00:17:19: So what does this proposed mechanism for objective enforcement then?

00:17:22: It requires physical visual classification.

00:17:25: Sarasi argues for treating class one pedal-assisted bikes exactly like traditional bicycles, you know capping the motor assist at twenty miles per hour and fully legalizing them in bike lanes.

00:17:34: okay then The city needs to mandate distinct visual classes for low speed throttle vehicles And most importantly they must aggressively reclassify those high-speed emotos ones weighing like a hundred fifty pounds and hitting forty miles per hours as the mopeds.

00:17:49: They actually are right requiring registration and keeping them entirely out of the protected micromobility lanes.

00:17:55: It's all about aligning physical momentum in kinetic energy with vehicle, with correct infrastructure Yes And Thomas Brotton brought up a similar point regarding parking infrastructure.

00:18:06: Currently most micro mobility parking relies on GPS geofencing

00:18:11: Which is always so messy

00:18:12: It is!

00:18:13: In dense urban canyons, GPS multi-path errors cause the location data to drift.

00:18:19: So you end up with scooters digitally registering as parked properly while physically they are blocking a wheelchair ramp.

00:18:26: The software thinks it's compliant but the physical reality is literally a tripping hazard.

00:18:31: Precisely Broughton argues the industry must transition to physical infrastructure like spark park stations that mechanically verify The vehicle is locked in the correct orientation before the user's trip can even end.

00:18:42: Oh,

00:18:42: That's brilliant.

00:18:43: Yeah It just removes the guesswork from the algorithms completely.

00:18:46: and Aligning these physical vehicles with the built environment kind of brings us to our final theme the ultimate user.

00:18:53: Whether it's a robot taxi or an e-bike, the ultimate success of mobility relies on serving human needs not just flexing technology.

00:19:01: Shailen James Anto introduced this concept called The Passenger Defined Vehicle.

00:19:05: Okay I like that.

00:19:06: For one hundred and forty years automotive design was constrained by absolute physical requirements for a human driver.

00:19:12: You needed forward facing seats A dashboard A steering column Pedals All Of It.

00:19:18: But the moment you remove the driver, you removed the control center.

00:19:21: You can suddenly design the vehicle's interior volume from the passenger outward—the HVAC, the sight lines and seating configurations….

00:19:29: It all fundamentally changes when a vehicle becomes a rolling lounge or mobile workspace rather than just a driver's

00:19:35: cockpit.".

00:19:36: And Senator Phillips expanded on this beautifully in her TED Talk Insights... Challenging the industry's focus on affluent able-bodied urbanites!

00:19:44: She argues that the real question of mobility access isn't about luxury.

00:19:48: It's about closing the transport poverty gap, right?

00:19:51: She asked this room full of auto executives Would you let your daughter take the bus at an unsafe stop a ten

00:19:57: p.m.?

00:19:58: Can your elderly parent who can no longer see well enough to drive still maintain their independence and social connections?

00:20:04: And when you reframe autonomous mobility around access rather than just autonomy for its own sake The operational priorities completely change

00:20:13: also.

00:20:14: Well, Jay Shen noted that Europe is facing a rapidly aging demographic right.

00:20:18: Nearly one in three people will soon be over sixty-five.

00:20:22: In that context accessibility features like kneeling suspensions or advanced voice interfaces for visually impaired passengers.

00:20:30: They become a core operational requirement For fleet survival not just a niche side service.

00:20:35: Right

00:20:35: it's essential and we are seeing this demographic shift impact deployment strategies everywhere.

00:20:41: Adam Bulkher argued that autonomous shuttles represent the ultimate bridge for rural communities.

00:20:46: Oh, interesting!

00:20:47: Yeah everyone focuses on dense urban cores because utilization rates are higher but rural populations increasingly isolated from healthcare and jobs.

00:20:55: The operational design domain of a fixed rural route might actually be simpler to solve than the chaotic city center and the socio-economic impact would be life changing.

00:21:04: Absolutely, And this human centric approach it must extend to the infrastructure supporting the transition too.

00:21:10: What do you mean?

00:21:10: Well Eubring Grunwald pointed out a glaring flaw in our EV infrastructure rollout.

00:21:15: In Olympia Washington half the residents are renters.

00:21:18: Okay If the industry only subsidizes level two EV charging infrastructure for homeowners with private garages The entire electrification transition fails.

00:21:28: It leaves half the population behind purely based on their rosing status.

00:21:31: Wow, yeah that is a massive oversight.

00:21:34: So what does this all mean?

00:21:36: We are entering an era where the vehicle Is no longer defined by the driver But but in network the passenger and city infrastructure.

00:21:44: If we truly optimize cities around shared fleets And passenger experiences rather than drivers What happens to a century-old culture built entirely on the personal joy of driving and individual car ownership?

00:22:17: Something.