Why Ford And VW Shut Down Their Multi-Billion Dollar Self-Driving Project 

As a robotics engineer, I see two very different technology stratagies. You can either focus on really good performance in a limited area, then expand that area to scale (what everyone else is doing), or focus on decent performance in any area, then improve that performance everywhere (what Tesla is doing.) Understanding this difference is critical to understanding why all the other L4/5 efforts are failing. They look great in their geofenced areas with custom high-res maps, but can't generalize. They didn't anticipate the "long-tail" of challenging circumstances encountered in real diving.

I recently had the opportunity to ride in a Waymo self-driving car for the first time in downtown Phoenix and I must say, it completely blew my mind. The ride arrived just three minutes after I ordered it and there was no driver or supervisor in the car, it was just me and the vehicle. The instructions were simple and clear - I used the app to open the door and start the ride. I also had the option to pull over at any time or change my destination if needed. At one point, the car had to make an unprotected left turn from a small street onto a busy one with a stop sign, and it did so with grace. I understand that Waymo uses Lidar and can only operate in certain environments, but nevertheless, the experience was truly mind-blowing. It felt like the future had finally arrived!

I instantly lost respect for this guy when he actually said Tesla's self driving is "a joke". And almost immediately after goes on to praise GM's CEO. I suspect something fishy going on here

If you ever felt useless, remember that this company was worth $12,400,000,000 dollars and never got anywhere.

My wife and I have a 2020 Tesla Model 3, dual-motor, long range with the full self driving option. Living in the Black Hills of South Dakota I use the FSD for all of my highway driving and frequently in town traffic. As you may know, I need to keep one hand on the steering wheel and always monitor the FSD's driving. It is not perfect, not yet, but for highway driving, it is great. Two summers ago I was returning from Deadwood, South Dakota on the north-south Highway 385. It was about two weeks prior to the Sturgis Rally. The Tesla was in cruise mode with full self driving engaged.. My speed was a steady 6o MPH and the Tesla was in the center of the lane. My right hand was on the steering wheel and the speed was a constant 60 MPH and I was in a long straight stretch About half-way between Deadwood and my turn at highway 44 I noticed a group of motorcycles coming toward me. As a former motorcyclist my thought was "boy, they are really bunched-up". When the motorcycles got very close, I was slammed into my driver's side door. I was startled and then noticed that the Tesla was now riding on the right white line. The car had moved over about six feet in the "blink of an eye". The only explanation I have is that the front camera saw a motorcycle cross the center line and took evasive action. Autodrive may have saved my life. So when "experts" say that Tesla's full self-driving is _____ (whatever), they do not have the real facts. In the future, hopefully your comments will be part of your embarrassment.

I have Tesla's FSD, and it's no joke. The rate that is it improving is stunning. Cannot wait for my next update coming this week.

I recognized all along that Tesla had this huge advantage because they are building the technology as an added feature to their core business of selling automobiles. And they are using their customers to train the system in real world conditions. Meanwhile those other companies are paying people to do all that.

This video will not age well.

Damn, the hate on Tesla by CNBC is real

looks like Tesal didn't send the check to that "guy", to call it a joke with elaborating why is a low blow. I wonder if he got offended by Musk lol

Dude actually said tesla FSD is a joke and its last amongst other companies. Apparently has not seen any of the vids and the astounding progress which will get solved soon. Yes im aware of musk optimism failed attempt for said dates however i own a model y with beta and progress is mind blowing. Its not if its a matter of time before tesla solves it

Tesla's tech is a joke?? That analyst is a joke

This video is incomplete. The legacy automakers tried full autonomous vehicles back in the late 80's early 90's. Of course the computers were laughably large but it did result in many of the safety features we use today like automatic breaking, adaptive cruise control and lane keep assist. I would say the only problem is that they recognized the tech wasn't going to be ready for years out, and stopped working on it just as Tesla came around.

The guy that said Tesla’s system was a joke clearly doesn’t own the car and hasn’t tried it. He also said at the end that he expected Argo to be there at the end of the decade. Dumb analysts barely know anything real.

Whoever your guy is saying that Tesla’s tech is a joke doesn’t have an understanding about the AI that will win this race. Tesla is doing this completely different and their gamble has paid off with their huge internal investment in compute and the now millions of miles each day that are training their supercomputer. You might want to get an AI engineer to comment next time. This comment won’t age well when Tesla wins this 🤦🏼‍♂️

Marry Bara was also committed to Nikola motors( Trevor meltdown) and fraudstown motors

The issue is, most companies are interested in self driving just to make sure they can cut cost on labor cost and maximize the highest possible profit with the lowest labor cost. That why they all push for self driving

People said Tesla FSD is a joke is the same People who said SpaceX was a joke for attempting to recover an orbital rocket; And 5 years after SpaceX landed its first rocket still no company in the world can do that.

Teslas technology is not self drive, yet the most useful because the car doesn’t crap out and complain that it has “exited the supported area”. These geo-restricted self driving are only really useful to fleet operations that only drive specific routes. Teslas approach is a lot more useful to a normal human being that’ll choose destination based on convenience not support maps.

Looking forward to seeing a Mercedes cruising at 40pmh on the Autobahn. Such amazing software!

CNBC hates Tesla bash them every time they get the chance

As a ride shared driver it has been always so obvious that ride share companies cannot evolve to own the fleets, they just profit from someone else owning the car and doing the maintenance for free. It was so absurd those companies invested in self driving. It would make more sense they invested in self cleaning , self repairs , self fueling , and in parking space in the first place.

Tesla isn't a joke, that is a popular comment from haters to seem like they know their stuff. Mercedes system and most other systems coming can only engage in specific situations, like on a freeway under 60 miles per hour or something like that and on pre mapped roads. Tesla is going for it all, they probably could tweak their system now to work level 3 on some pre mapped roads and only certain conditions.

I don't know who that dude is that said Tesla's tech is a joke, but I know for sure one person who well respected and who knows what he is talking about and that is Munro. He is fascinated about Tesla's full self-driving tech. So, again, CNBC doing what CNBC does.

Looks like financial analyst never even looked at or understood the underlying technologies and issues like what type of sensors to use, data collection or AI and the requirement for data collection on a truly massive scale and the capability to exploit that for training.

The video gives a good overview on the different players/approaches on self-driving, except regarding Tesla. Calling Tesla FSD a joke is not the problem, since its still brittle. The problem is that Tesla has a different technical/business approach to self-driving and that wasnt even mentioned.

Tesla Is So far ahead of the Game its not even funny.

They don’t at all go into the differences of fully geo fenced pre mapped zones (whate everyone does) vs only using off the shelf video cameras (what tesla does). Neither do they talk about using a neural network and how much data it takes to train that.

2:15 the chart shows Ford having sold 6,600,000 and Tesla having sold 76,230, however the narration describes Ford as having sold only 10x as many units, when the factor is closer to 100x as many units

The real joke is companies whose vehicles are geo-fenced in the least complicated few city blocks and can only operate within those limited spaces in less than half a dozen cities. According to the SAE International chart displayed in this same article defining the levels of driving autonomy Level 3 still requires someone to be in the driver seat ready to take over when prompted by the ADAS system. That’s a joke for some companies but not for others if I heard right.

How did Uber lose 9 billion when they rip of and cheat their drivers .

Woah, self-driving technology is hard!?! Who coulda guessed 🤨

solving real world AI self driving is primarily a data issue, (if you don't understand why plz don't comment) and Te$la are many orders of magnitude ahead of the game in data harvesting + application, they r the horse I'd bet on

"Tesla, who isn't a leader" "Tesla's technology is kind of a joke" How can you diminishes that freely Tesla's effort? I mean even if you don't agree with the strategy and think it will not work at the end, you can't pretend they are not serious about it. They designed their own chips, their own training computers, they already make and sell millions of data acquiring vehicles. They even presented some of the work they've done. Denying that is kind of unbelievable... "It's decent like blue cruise" I think the joke is on you. To accept publishing such thing tells a lot about your editorial standards...

I was once the Director of Research and Development for Breed Technologies, the company that developed the first commercial crash detection sensors used to deploy airbags for occupant protection in vehicles back in 1989. As remote sensing technology became of interest for prevention of crashes back in the late 1990s, I was always concerned about where the remote vision sensors were located on the vehicle. A typical reason for lawsuits involving injury to vehicle occupants injured in a crash is the degradation over time of the crash detection or airbag deployment system due to poor vehicle maintenance and damage incurred from things like parking lot collisions or rough road abuse. The physical location of the sensors that are looking for what is happening outside the vehicle is in close proximity or on the vehicle exterior as opposed to responsive accelerometers located within the vehicle is a very important distinction to make. In a vehicle equipped with a driver assistance system, that little dent in the bumper or side of the vehicle could be the difference in life or death if the vehicle is not inspected for damage and repaired very quickly which could prove to be very expensive over the lifetime of the vehicle. In summary, the practical use of sensors to prevent road accidents is a far more daunting problem than detecting that a crash is happening and taking appropriate safety measures.

I was a part of Argo AI in Munich - sad for everybody who lose their jobs

All the obstacles Tesla's competitors have is not an issue for Tesla. They make their own cars at a fraction of what it costs the ride sharing companies so can charge $1 per ride and still make a huge income.

In summary, self-driving is technically possible but it's such a money furnace and there is no clear road to profitability so that only giga corps like Google and Amazon can afford to prolong development.

Woa. Nice video piece. Thanks.

Has anyone (in the video or the comments) actually used Tesla’s latest FSD 11.3 software? It doesn’t seem like anyone actually has any real world experience with it…

“Tesla’s technology is kind of a joke” lol, that schlub’s gonna eat his words.

If they can’t design a self-driving train (an infinitely simpler problem), then… there’s no way they’ll ever have a self driving car. Also, notice how crisp and clear all the lane divider road paint is in these demos. That’s a huge contrast to most of the faded paint on the majority of real world roads. Many people have reported “lane keep assist” horror stories in areas with less than perfect lane paint.

Sometimes I wonder if large news companies are afraid to say anything positive about Tesla, as in do they fear loosing add revenue from legacy auto makers. I swear the car commercials never stop and Tesla doesn't do traditional commercials.

How to go back..... Thanks CNBC, you made me laugh as never, you recall me the first time cars was aimed to replace horses...

So what you're saying is that Tesla is the only one that's figured out a business model, believes in the technology, and is furthest along to developing it? Awesome!

Meta! The guy saying Tesla’s technology is a joke is the joke himself. Tesla is the leader (period)

I dont have tsla or any self driving car... But to call tsla's tech a joke? It seems like a big statement

You should have been around for the fight between Intel, AMD, Zilog, Fairchild, and the rest. Tube giant Sylvania made semiconductors (TTL). Everybody wanted to be a microprocessor company.

Tesla self driving is a joke? I don’t think this guy ever tested any of these cars…. I’m a car enthusiast in Taiwan, and I can CONFIRM after driving several cars by different companies, Tesla’s self driving feature is second to none.

I drive a 2022 Model 3 Performance with v11 FSD beta and it's good. Like really good. Like so good, when it comes to directions and navigating new areas, I let Tesla software take the reins. I do 10,000% prefer a battery powered car with this level capability, and Tesla is the only one that has it, that can sell you a car today, and can service that car. The others just can't. My car takes me from door step to door step. On and off highways, highway interchanges, roundabouts, stop light and signs, over takes slow cars, parks itself, can (kind of lOl) pick me up in a parking (sometimes).... its sooo dope. Like so dope. I And it is silent and faster than 99% off cars on the road. Like wtf. What's not to love?

Wait, WTF? "Tesla's AI is a joke..." says the company that put an athletic director in charge of Autonomous? Tesla is further than ANY of these companies!

CNBS last in the truth Sam is a joke

I watch CNBC for the comedy! That line about Teslas technology being a joke made it worthwhile!

"Tesla's technology is kind of a joke" - LOL

always remember the key point not that the tech itself fail but more of business model still not found. basically autonomous driving by removing driver means adding problems directly to the company which usually handle by the driver

Just like the content, am intrested in the name of software used to edit the video @CNBC

I love how CNBC trashes Tesla. It’s fine. When Tesla solves autonomous driving, these same experts are gonna be claiming nobody could have seen this coming and try to jump on the bandwagon.

Nearly 10 times as much cars? Maybe I'm bad at math or something but I could be certain that 6.6 mil is over 80 times more than 76230.

i never understood the confidence about full self-driving at the same level of safety and success we have now with human drivers.

Self-driving isn't possible when it rains.

AAAAaaaand that's why the cold war model for disruptive R&D was superior. Doing high risk, long term and high cost R&D based on market forces and promises of profits when the uses for said future technology in development are not fully sussed out either result in slower development, or hype driven development that results in abandonware. In contrast a publicly funded R&D initiative like the ones we had in the cold war is what gave today's material technology, aeronautics, networking, batteries, etc. for example the invention of the microchip would be IMPOSSIBLE without 50 years of prior publicly funded research into material technology that at the time had no obvious use. If you want the US to be the leader in tech we need to re-introduce the cold war R&D model, the time we 'd set a goal that seemed unrealistic at the time, such as reaching the moon in a decade, which had no immediate monetary profit and yet we put 5% of the GDP into acheiving that goal. Decades later the tech that was developed to get us to the moon is the backbone of hundreds of industries that are bolstering the US economy, paying us back in entire orders of magnitude. If we want to remain a world leader we need to stop relying on market forces to fund disruptive R&D and initiate a new era of scientific "moonshots". Personally I believe the next frontiers are biotechnology (especially medical), AI, nanotechnology, advanced manifacturing technology, material technology, quantum computing and cybernetics. If we don't get off our current tragectory and don't invest the resources and funds only a superpower can muster to be the ones to develop this tech first then I suggest you prepare learnging chinese. Not only will their technology surpass the US and result in it gaining an economic upper hand but technological fields like medical biotechnology, medical nanotechnology and cybernetics are the ultimate soft power card, because this technologies are the key to ending disease, disability and aging. And there's no stronger card you can play than being a "life broker" It's extremely hard to beat "come with me and you'll live for centuries without aging and no disability"

They totally under estimated how difficult it would be

Very good

*Phoenix AZ was spelled wrong on the map in your video* ( at the 13:00 mark )

Saying Tesla is at the bottom is a joke. That guy needs to report on something else like dog toys and adult toys.

Ford should first focus on making cars that don’t have silly recalls and reliability issues

It’s not just the tech. It’s the business model. Once the tech is there, it’s going to be small boutique firms that offer luxury/exclusive products that can charge a premium who will win the competition. Think equinox of Uber. Ideal area for this business model would be cities like Los Angeles.

Im thinking about progressive automatization. Similar to airplane development. It seems reasonable that a driver will always be required, but with less and less intervention up to a point where there will not be required to pay attention/focus, unless something unusual or difficult comes along.

If you've ever been in a taxi in India or China, then you can understand just how excellent Tesla's FSD software is. I rate it to be 50x less stress and 100x safer than my Asian taxi rides.

13:05 So this is why TSLAQ thinks Tesla ranks dead last in autonomy. Hopefully they keeps using GuideHouse Insights "research"...

Tom Narayan gave excellent insights. 👏

c/c++ 1 buggy 2 malware vulnerable software issue too? How many UPDATES?

This sounds like a straight hit piece on Tesla. Having been on FSD since July I can safely say it is a lot further ahead in terms of ability of said companies on this video 😂

Comparing Teslas FSD Beta (free range) to all the other geofenced systems is apples and oranges. Amazing how some so called "experts" can't/won't see what's right in front of them. I'll take FSD Beta over any other system hands down. Bonus, the maker of FSD also makes the vehicles. Again, comparing apples and oranges.

As this video discusses the business side of self riving cars i see poor judgment when they dissmiss tesla FSD. FSD approach is transferable more to robots like optimus than the geofencing appoach. When you develop a difficult technology for multiple markets sectors seems that business analysts cannot follow such technology raodmaps and are prime targets be taken over by AI themselves so people can access less biased information as displayed by this video content.

Just passed one of these, on way home from family & thanksgiving…. 101 freeway in Scottsdale AZ.

I love how he says “Tesla’s technology is a joke.” Yeah how’s it working out for Ford and VW? Oh that’s right the company went bankrupt 🤣

Coworker has a Tesla with FSD. It’s pretty amazing although not perfect. I’d say Tesla are 80% of the way there, of course if the 80/20 rule holds the last 20% is going to take 80% of the total effort.

We all know self driving cars will fall under our AI overlord ChatGPT.

These self driving vehicles need a lot more work to Iron out detections software. Just last week I almost collided with a Waymo vehicle in Tempe on Broadway Rd as it made a left in front of me. I was driving a City Bus; you would think the Engineers would ensure the sensors would pick up a large Commercial vehicle easily!

Sam literally lost all credibility when he called Tesla's system "a joke". Just because other companies have loss leading ride hailing services approved doesn't mean it's ahead of Tesla's. A profitable approach with minimal sensor suite while being able to scale through software overnight is the masterplan. Very very few understand this.

Self driving technology should target easier though still useful projects. Such as low speed shuttles that run in downtowns closed to private cars on defined routes. The shuttle can learn a single route, run at lower speeds, not have to deal with personal cars. Or autonomous farm tractors.

LOL. I was about to comment, after watching this video presentation, on the complete misunderstanding of the TWO major approaches to Full Self Driving: The Tesla version using cameras and the Geo-fenced version via lidar etc. However, I read the comments and many explanations expose the "Real" situation better than I could explain. One thing: That closed-minded guy, Sam Abuelsamid, will regret making such a dismissive statement about Tesla's capabilities re FSD. Tesla had started down the Lidar track but realised the limiting problems associated with that approach. Then they headed off at a tangent with their own system. History will tell us which "system" will win out. I know who I've got my money on. Cheers.

Realistically good level 3 is what the market wants.

I was recently driven to an appointment by an 80 year old woman. After having watched numerous Tesla FSD videos, there is no doubt in my mind that the Tesla is a better driver than the woman I rode with. Tesla may technically be at level 2, but it most likely is somewhere around level 2.8. It has made great strides over the last year. If it keeps improving at the current rate, level 3 is just around the corner. Level 4 easily within 3 years.

I was lucky to sell my Ford and GM stocks on time.

Cut off executive bonus and they’ll have plenty of capital to fund these projects.

He said it's joke me watching a video of man going to his job without touching anything 😂😂😂

"tEsLa Is A jOkE". 😂😂😂 TESLA is eating everybody's lunch.

They could have just asked any normal software developers. When it’s someone else’s money anything is possible until the suckers catch up to reality.

Autonomous vehicles feel like the most insane way to try to avoid just investing in public transit. We invented a way to get on a vehicle and get to where you need to go ages ago, trains and buses, invest and build out the networks and we literally don't need this crap at all.

We don't need autonomous private vehicles. We need mass transit.

Cant believe they said “Tesla is a joke and not even a contender” that man clearly did not test Tesla’s fsd beta

13:06 You forgot to mention that EVERY SINGLE AV company listed there, except Tesla, require extensive mapping data and extremely expensive and high maintenance LIDAR. Tesla cars can drive themselves in most circumstances. It's basically level 4.

First I don’t believe that level 4 or level 5 self driving autonomy is achievable : there will always be some rare outlier cases that, at time, will require a human conciousness to take over and make a (risky) decision. Waymo strategy, with some remote intervention at time, or in situ if needed, seems more a more plausible path to most of the time self-driving. However many of those self driving technological efforts to be really profitable may actually require some technological advancement / disruption in semiconductors like Non-Volatile Memory / Storage Class Memory (SCM), and memristor technologies to create a more energy efficient and less costly system.

I had a coworker quit to go work for them last year, we're in an adjacent field to waymo. Long story short, they're back with us now.

By gum, that was increadibly dismissive of the current development state of Teslas offering in this potential marketplace.. that seems to undermine the credibility of this analysis in my view.. have I become a Tesla fan boy or is there an underlying bias here that we seem to be increasingly seeing as the other incumbents in this field start to feel the real heat of mounting commercial threats to their traditional dominance? 😮

This Analyst obviously has not spent much time actually testing out these technologies. There's only one automaker that can map and execute a route with zero human intervention and it's not Mercedes.

Waymo & Cruise are the market leaders in autonomous driving. Tesla is not as close.

When everyone else fails, you should have reasonable doubts for the only guy who claims he can do it.