E9: Should Tesla Be Scared Of Nvidia's Self-Driving AI Chips?

[@TickerSymbolYOU]

Do you think we'll see self-driving cars by 2025 and if so, who will be the first to crack the code?

[@RemoteSpeed007]

Tesla 2025-6 FSD 12.3.4 is already drives itself, while getting better every couple of weeks now.

[@SamuelMM_Mitosis]

@@RemoteSpeed007agreed, he’s saying how hard Urban environments are, and they can hopefully try it in the future. Tesla already pretty much solved it.

[@CarstenHensch]

I just tried Teslas FSD 12.3.3 for 2.000 miles in Texas with much city traffic. They basically cracked it, even when they still need some updates (which happen fast).

[@RemoteSpeed007]

@@CarstenHensch Good job doing 2000 miles - wow. I only did about 4 hrs of driving and it's working great so far in Florida.

[@RemoteSpeed007]

@@CarstenHensch You seem to be the person to do a coast-to-coast test for no interventions on FSD - would make you famous.

[@JulianSnow]

This content you're reporting on is so much grander than stock market gains. Really cool what this channel has become.

[@TickerSymbolYOU]

I appreciate that!

[@antalszabolcsdesign5189]

I agree 100%

[@goldnutter412]

Digital twin for the win. Understanding where uncertainty grows lets us leverage asymmetry. As we dialed up fidelity of measuring the universe we found a limit, things start to approach infinity or uncertainty of precision without measuring.. and what Wolfram calls computational irreducibility. This path was inevitable..

[@paelnever]

Amazing that when this channel trow shit against nvidia calling it "AI bubble" their stocks rocket up to 970 and when he kisses the nvidia a** the stock drops down to 760. This fact doesn't talk bad about the company but everybody should consider things said in this channel.

[@JulianSnow]

@@paelneverI'm here for the product reporting, not the markets

[@teendrivethrive796]

I bought Tesla stock in 2013 and Nvidia stock in 2014 as a hedge against self driving making auto insurance obsolete. I own an insurance agency. The hedge has worked incredibly well and self driving is still not here. I also own a Tesla and have been testing Tesla FSD 12.3.3 in some extremely challenging driving scenarios is Seattle. Is it perfect, no but sometimes I have 45 minute drives that make me feel like I hired a chauffeur to drive me around town. I believe Tesla’s switch to a neural network approach to training FSD is the most important leap forward in self driving. This could not have been accomplished with Tesla’s Dojo supercomputer alone. It wasn’t until Elon recently purchased enough chips from Nvidia that he declared we are “no longer compute constrained”. Tesla also has millions of hours of driving to feed into its Nvidia AI supercomputer and let machine learning work its magic. No other company has this massive real driving data and could be the advantage Tesla has over all other companies. Will Nvidia’s approach of a supercomputer in a car with multiple sensors beat Tesla’s neural network approach? I don’t really care, even though my insurance business will suffer, human drivers each year cause 1 million deaths, even more injuries, and billions of dollars in property damage. Let’s just solve FSD already. Lastly, if you haven’t tried Tesla’s FSD 12.3.3 you really don’t know anything about Tesla’s current FSD capabilities.

[@MrBurwoodman]

It’s nothing like FSD. It’s an assistance system which is small utility until level 4. Every OEM will have level 4 by 2027. Tesla is redundant like their workforce. The Musk effect

[@250txc]

Who gives a F on what U say U did? lol ... U are probably a BOT. lol

[@teendrivethrive796]

Why does it seem like passionate self driving fans are either Team Tesla or Team Nvidia. I’m Team Tesvidia. The real problem is human drivers are terrible, let’s just work together and solve the problem.

[@KC-uw6ph]

To correct you they have a billion miles of data. I think waymo who is 2nd has 40 mil. IMO Tesla will reach FSD first. But with ai and nvidia omniverse who can basically train it on synthetic data will come in 2nd. Of course real world data is always better, but you could probably simulate synthetic data for specialized scenarios. Also how easy would it be to create the modeling in Omniverse to train their cars in? On the other hand when ai is able to create long enough and accurate videos of people driving Tesla could use these to train their neural net data. Interesting time indeed. I do hope they partner up, win win for everyone.

[@tashi282]

but that also doesn't mean what this guy said is final-it's all speculation now!

[@AP19327]

My biggest question is what is the decision-making process when they are getting conflicting data from radar/ultrasonics/cameras and how do they determine which is correct? Tesla ditched radar/ultrasonics due to this and it would be interesting to know if Nvidia cracked the code or if they are also dealing with these conflicts

[@neototem4110]

Using a multi-sensor suite for spatial resolution is what submarines and aircraft have been using for nearly 70yrs to my knowledge unless the Germans have some undiscovered WW2 underground base full of vehicles that already achieved this level of utilization prior. Tesla did not ditch everything from their sensor suite except cameras because of "conflicting data" lol No, they ditched it because they don't know how to make it work the same way they didn't know how to deliver the Cybertruck two years ago as initially promised. This thing where they came out Steve-Jobs-for-fake-signal-bars style and actually got their customer base to believe that "cameras" was all an algorithm needed to navigate "autopilot" functions was completely insane. That Tesla slammed right into a parked fire truck with lights flashing and plenty of size to obstruct background view. Driver killed. Cameras.

[@HepCatJack]

There was a plane that crashed some time ago because an instrument had been taped over during maintenance and the tape had not been removed. It provided data that contradicted what other instruments were saying. I suppose one algorithm could decide to go with the majority of the instruments and disregard the malfunctioning one under the reasoning that it was more likely that a single instrument was malfunctioning instead of the majority.

[@pandoorapirat8644]

more data the better - me data scientist

[@meggi8048]

@@pandoorapirat8644 only data necessary to make decision is necessary... more data = more noise = more latency = potential danger

[@meggi8048]

@@neototem4110 you realize that air and water have less action going on, thats not even comparable. germans have a undiscovered ww2 underground base? how would you know if its undiscovered? they ditched it because more data than necessary never works. musks philosophy is to remove parts, he does it everywhere. less parts = less failure possible. and nature does the same or why do only a few species have ultra sound? and when they do why do they have poor vision? why not have just everything super,... super hearing, super vision, super ultra sound, super muscle...? because everything super has costs and tradeoffs. only the minimum to make it work will prevail. tesla knows that, and everyone else will too, because even when other car makers will have similiar capabilities like FSD has now... they will cost much more. tesla will be the much cheaper option and work. humans drive with only 2 cameras for decades and are so far the benchmark...

[@katherinehicock9452]

That was a great tour. Thank you

[@TickerSymbolYOU]

Glad you enjoyed it!

[@anskey]

Great interview. Loving your content!!

[@TickerSymbolYOU]

I’m glad you enjoyed it!!

[@bjarnesegaard5701]

THanks for showing how Nvidia and others are working on autonomous driving.

[@TickerSymbolYOU]

My pleasure

[@pubwik]

i love how locked in and focused you are : great interview !

[@TickerSymbolYOU]

Glad you enjoyed it! Thanks for taking the time

[@natpainter8185]

@@TickerSymbolYOU roboreply

[@janicenagao7409]

EXCELLENT VIDEO THANKS

[@miikkasalavuo3015]

Thank you, great interview! I assume Nvidia Drive is using Soundhound for the voice assist part? Did you talk about other 3rd party companies involved?

[@natpainter8185]

lota profits to split and nvdia is greedy

[@cooljamesmom]

Thank you for sharing this info. You are appreciated.

[@TickerSymbolYOU]

Thanks for watching!

[@WaldoTheWombat]

2:40 so it has radar, ladar and cameras? or are two of these terms the same thing?

[@TickerSymbolYOU]

radar, LIDAR, and cameras are all separate kinds of sensors

[@stephanepires]

What do u think about ui path

[@RemoteSpeed007]

Good video - Did he give you an indication of when L3 could be expected?

[@jimtexas68]

Costs to scale this will be epic. It will not see the light of day on a grand scale.

[@cgsschaefer]

great. Shapiro really took time to explain ! nice

[@TickerSymbolYOU]

Yeah, he was super generous with his time!

[@Johnslist]

Hi neighbor, I'm in San Jose and was signed up to go to GTC but just caught parts on the web. I saw Danny present before and was already interested in Nvidia for autonomy, cars, and simulations. Everyone is only talking about AI, but that's a small subset.

[@bille2298]

Its funny on here your very serious and direct... BUT at the show you were very animated with your responses and almost over enthusiastic!! lol Great info regardless !!

[@DavidMorris-uy5jb]

Great video! A couple of years ago Mercedes Benz mentioned Brainchip as being intrinsically involved within their EQXX concept car. Do you have any knowledge of this continuing on? I believe the NVIDIA chip within the MB needs water cooling so I guess there are further wattage efficiencies that need to be overcome.

[@johnpottinger7645]

I bought NVIDIA back in August 2015 solely because of their work in self-driving chips. I profited from them, Mobileye and XILINX. The shares grew exponentially over the period and I reduced my holding significantly. With the benefit of foresight I would have held all 775 shares until now. As it stands I now own 124, bought for £1,613 in 2015, now worth £87,210 (5,304% growth). Good companies like NVIDIA and Tesla are not one trip ponies and if held over the longer term as part of a wider portfolio, are likely to bear fruit on a number of fronts.

[@clavo3352]

Very good video. Much relevant information. Autopilot variability per manufacturer is very market value important.

[@marcelfekonja6354]

Great video. I realy enyoyed. I recomend you since you was really intrested on NURO ti thriwing an eye on ARM also.

[@TickerSymbolYOU]

I'm glad you enjoyed it and thanks for recommending me.

[@dragonballz2560]

Anyone know if these systems recognize different cars?

[@aaronscottmatthews7883]

12:58 interior mapping is definitely going to be monetized

[@pandoorapirat8644]

This could be free. Buy only our expensive THOR GPU thats enough. Like the gaming market. Or borrow the THOR...

[@girander]

Will AV's - on top of the sensorinformation you talk about - take into account dynamic information on other traffic(position, speed, direction etc.), obstructions, traffic lights(redundant..) etc.? Ultimately vehicles should flow and adapt uninterrupted between other traffic etc.

[@Youtuber_YusukeFromGermany]

What Ive learned now: Buy Tesla and Nvidia and call it a day

[@Johnslist]

You're getting them on sale tomorrow.

[@joechai6077]

Time to buy

[@WaldoTheWombat]

So Nvidia provides a simulation and the autonomous vehicle developers use it? How do they apply their own methods and algorithms on those simulations?

[@andywong1602]

Thank you 🙏 Alex

[@JumpDiffusion]

Good stuff 💪🏻

[@TickerSymbolYOU]

Glad you enjoyed!

[@WanderingExistence]

YOU are a big value add!

[@davidlemieux615]

Re comment on not being dependent on 5G comms… that’s smart but also bad news for GM’s effort cruise which heavily relies on this and will come to a halt in dead zones.

[@blackhorseteck8381]

Nice interview, glad AV/ADAS tech is finally moving with the rebirth of Machine Learning.

[@charleswilliams-bu4ro]

If an EP pulse knocks out the electronics, and say, the batteries driving the electronics are conked out for some reason, BUT the car battery is OK, can I still drive the car in the normal way?

[@brian7android985]

Of course not. That would be sensible. Got to have chips running everything!

[@keithcook3908]

Very interesting 🤔

[@SuperAlijoon]

There many video in your channel that “company .. Will change everything. “ The real change is rare and happens gradually. It’s better to be selective and not amplifying every single tech news. Cheers

[@natpainter8185]

thank you

[@Horatio1886build]

Yes, exactly. And the idea that Tesla( Elon is electric/AI Jesus) also worries me when I get to enthusiastic about Elon- but thanks to him for supporting free speech) But I try to not be to much of Tesla fan boy. He is not infallible.

[@patharrington1375]

Great interview Alex thank you.

[@TickerSymbolYOU]

cheers! 🍻

[@jt-vj4sl]

Another autonomous driving hype now Tesla can’t sell vehicles. It will be there eventually but not in times to save Tesla in short term. 10% Tesla employee cut was just first step they took yesterday. The next step is to have better leader at the company. Musk has been a distraction for this company. Both China and Russia needs him

[@davidlemieux615]

Interestingly, should you wish to investigate, you will find an increasing number of advanced ADAS developers have announced they were dropping LIDAR and just using LIDAR to calibrate their camera systems… which is exactly what Tesla does (there are videos on KZbin showing just this). This avoids the “noise” they are finding when trying to merge several sources. It should be remembered that cameras have very high resolution relative to human eyes (other animals have better eyes than us). A nice demo was once done showing what we saw at late dusk relative to the Tesla cameras… it’s amazing how much more they see… add AI, as Google is great at doing, and night vision can be greatly improved. Just check out a Google Phones camera. If Tesla should add an additional sensor, it MIGHT be an IR sensor… but even that is hard to justify. It should be noted a great many animals have stunning image detection, just check out predators which hunt at night and things like owls and hawks.

[@NikiK57]

Thank you very much for the interview! This is very important stuff for people to understand. I can see all the time people shouting Tesla, Tesla, Tesla, when anyone even mentions self driving. But in reality - there A LOT of companies going for AI in cars. A lot of car makers are at level 2 autonomy. Does Tesla have any significant advantage over others? Why do people keep on thinking Tesla will be the top dog and lets say VW group, Mercedes, Ford, GM, Korean, Japaneese, Chineese car makers etc. will all fall behind heavily?

[@carloorelli3538]

Why do you even ask? Every single Tesla on the road is equipped with FSD chips and 8 cameras. 4 million veichles that have spent a decade sending data to be used to train Tesla AI. This is a gigantic advantage over anyone else. The closest competitor has a tiny fraction of the number of cars Tesla has

[@NikiK57]

@@carloorelli3538 and still Tesla is at level 2 and experts are saying its not the best 2nd level out there. Mercedes is tipping its toes into towards level 3 for example.

[@chrisfpv7615]

@@NikiK57 Even the chinese have much better technology than tesla but none mentions it

[@carloorelli3538]

@@NikiK57 it all depends who the experts are. Building a FSD system relying on cameras only will allow the syste to be used on every car. 10K LIDAR can be good for FSD but means it is going to remain an élite extra only that goes against the idea that one day every car should be able to drive itself to avodi as many road deaths as possible

[@GlobalMan-nr3hq]

@@carloorelli3538 Tesla quietly went the Transformer architecture from v12 only. They are basically starting from square 1 more or less like the rest. Yes, their very real advantage is gobs of road data but Nvidia uses simulation for infinite combination scenarios and with Lidar for better night vision. If done right with real world data and fist class FSD LLM and Tesla remains static without any improvements, Nvidia can overtake Tesla. That's why Musk is showing the urgency and is eager to show the "lead" in Robotaxis by Aug (I wont hold my breadth though). That doesn't mean Nvidia will take lead but Nvidia now belongs in the same conversation.

[@hunterdale3186]

Thanks for the great videos!

[@NexGenTech2024]

Thank you for the insight into detailed-oriented interview questions. I would like to ask if NVIDIA will accelerate computing by 1000 times, what would be the fate of quantum computing?In light of NVIDIA's potential to enhance computing speeds significantly, what implications might this have for the future landscape of quantum computing?

[@FARBOLUOS]

How much is one of this FSD delivery vehicles???

[@davidlemieux615]

FSD is improving by leaps and bounds… when will NVIDIA be able to do something similar? I hope soon (I’m an investor in NVIDIA and Tesla).

[@joechai6077]

I actually tried to do some research on when will nvidia bring the system to actual roads but ended up not finding anything. I think it will take a few years, the nvidia guy said it will first start on highway driving and they will gradually take it to urban areas.

[@dreamphoenix]

Thank you.

[@2024Miracle]

Would you please make a video on AMD?

[@nenickvu8807]

If Tesla doesn't have anything that can compete with the digital twin environment where millions of fully automated situations result in accidents and mistakes, it's going to get leapfrogged. In order to learn, the AI must make mistakes and learn from them. In fact, making mistakes may be just as important than learning to drive correctly and between the lines. Also digital twinning can also be used as a regulatory model, where AI is tested before it is certified as safe. It'd be up to the regulators to come up with the tests, but it's a great way to simulate the complexity of the real world and see how the AI responds. Lastly digital twinning might be a potential black box method where forensics can use the data to reconstruct the failure points and fault in accidents involving self driving cars. To be able to verify that the automated system is or isn't at fault is critical to developing better regulations and statistics regarding the safety of the technology.

[@teendrivethrive796]

Will a digital twin outperform a neural network trained by billions of hours of real driving data? How does the digital twin learn about all the edge cases?

[@nenickvu8807]

@@teendrivethrive796 billions of hours of driving data, most of which is normal driving and cruising down the highway, doesn't teach an AI how to respond to a complex situation like a cop or construction worker directing traffic. Only by "experiencing" a million different scenarios in digital twinning can give and AI insight on how to not only how to take direction from a worker correctly, but also how to do it wrong and avoid going down that path. AI needs to also know how it can make mistakes, create dynamics where accidents can happen, and prevent and defend against situations where the probabilities of accidents go up. The training data you're talking about where everything is fine 99% of the time is inferior to training data where all possibilities and probabilities are accounted for, good and bad.

[@teendrivethrive796]

Perhaps to solve FSD, a combination of real driving data and synthetic data will be the key. I think Tesla and Nvidia should team up like Marvel superheroes. 😊

[@dannyy7654]

@@teendrivethrive796Simulations will prove itself as far more useful for reaching level 5.

[@Bham67]

I'm unclear on this intermediate stage of "shared driving responsibility". For example, you still need to be in the driver's seat and fully alert to apply emergency breaking if needed. Seems like by the time you realize the car is not stopping at the stop sign or red light it is too late. No way I'm getting in that car. I either want to fully control the drive or be 100% confident that the car can fully control the drive.

[@mjcapinto]

Synthetic data isn't as good as real-world data, and sensors galore will increase costs significantly. Its great to see some competition between Nvidia and Tesla.

[@Tj-wc4hp]

You guys are describing the new ev nio car. Already comes with all these gadgets.

[@chrisfpv7615]

They won't simply because it is Chinese.

[@olivermiller2013]

To be honest: We have self-driving cars, when the car does not has a steering wheel anymore and this is the standard (lvl 4). We are talking about self-driving cars for years and it is still something you have to assist in certain moments - if you can. It is your responsibility. The technology is going further, but we do not have 100% self-driving cars yet. Let´s talk about assisting cars. This is the state of the technology today (lvl 2). If we are lucky, we see some prototypes with lvl 3 in 2025, but I´m not convinced to see it 100%. What is also concerning: A Blackwell costs more than an average car today. This has to change, otherwise you have not enough customers 😁.

[@paulschaaf8880]

That's my thinking as well pretty much. When I can buy a car without windows or a steering wheel then self driving is a thing.

[@ThomasTomiczek]

Nice - but as Tesla does not use it, it will be a fringe hope for tier 2 providers.

[@raghuhiriyur]

It is for other car companies

[@rossmcintyre6548]

Woah! 😂 Dude … this is like Bill and Teds Excellent Adventure

[@rossmcintyre6548]

Making fun of it … but actually a great episode … Good job

[@hohoh0]

Would have liked to see you challenge him and ask for his thoughts on Tesla vision only strategy-why does Nvidia seemingly disagree and the rationale behind it. Food for thought!

[@TickerSymbolYOU]

Sensor redundancy

[@randomsitisee7113]

If going to theatres Is the use case then I am worried. Why not AI that actually tells driver about that preemptive repairs like mileage on the tires, AC filter, transmission health , etc provide real value rather than bs like finding movie theatres.

[@taijistar9052]

NVIDIA’s vision relies on Radar and LiDAR in addition to cameras. That technology is like Venmo uses and already level3. It is much more expensive than Tesla’s approach with only cameras, if Tesla can succeed.

[@goldnutter412]

Uncertainty and asymmetry 👌

[@Horatio1886build]

I personally would never drive ( ride) in full auto mode because as I get older I have noticed many humans go into “ nap mode” or if young into internet “ gab or game or sports or stock research mode” and there is always the “anomalous encounter”out there waiting no matter how much AI “learns” .augmentation too leads to similar result

[@Memyselfandi59057]

Car that is an AI agent = Knight Rider. We’ve had that tech since the 1980’s.

[@dragonballz2560]

I HAD to watch this twice. 😮

[@TickerSymbolYOU]

🤯 right?!

[@maguilla]

Wait for the raining day or snow day or very hot days any one of this elements will shut down the system. Second : this cars will be even more expensive, regular working class will never change if they don’t make this cars at $18,000 to $23,000 max.

[@syweiner434]

1. How many watts does Nvidia car need? 2. How much does it impact mileage,?

[@jimidaly0]

Calling it now: Cars and buses with Blackwell stacks will be jacked for black market supercomputers.What a world!

[@midwestcannabis]

I'm late! Now I listen 🎉🎉🎉🎉

[@catherinedavid2722]

Excellent 👌

[@squfucs]

nice video bro

[@giladkingsley]

Reinforcement learning in simulation is a superior AI training method than human imitation.

[@dancover3624]

For an AI ETF would you Peck QQQ m or do you have something better? I love your channel and you teach me a lot

[@TheInfectous]

I think self-driving cars are going to come out about 9 years earlier than fusion and fusion is "only 10 years away!"

[@pocketsfullofdynamite]

How does fusion function

[@BilichaGhebremuse]

Could i use AI technology in mining with the help of robots so that the robot could mine and produce end productwith out human intervention

[@purol7877]

We get this same self driving story for 10 years already

[@VeteranWS]

can NVDA give Tesla run for the money?

[@georgegale6084]

Great Stuff.

[@Bot101101]

I would like to see your take on the new Destiny Tech100 fund. It seems interesting, but the volatility and valuation is a concern. All hype?

[@nancygonzalez6788]

Very interesting

[@Vancouver_Island_Guy]

It's hard to comprehend how big nvidia will be in a year or 2. It will be surpassing apple and Microsoft in the near future.

[@TickerSymbolYOU]

Right?! 🤯

[@christopherdennis6785]

My wife asked me, “did you hear about the self-driving taxi that hit a pedestrian in Texas?” I replied, really, did you hear about the 500 vehicle-pedestrian accidents last month involving humans running on “Neural Networks?” When a human driven car hits another, it’s an accident. If it’s an FSD car, it’s a lawsuit!

[@silberlinie]

Does Danny Shapiro, NVIDIA's VP of Automotive, really believe that anyone in the auto industry will buy this bunch of sensor stuff from him? Where others are able to clearly show that cameras can make things easier, safer, cheaper and more universal? Danny Shapiro, NVIDIA's VP of Automotive, is putting his reputation and career on the line.

[@TickerSymbolYOU]

Well, considering hundreds of automakers are already buying Nvidia's technology... yes

[@silberlinie]

@@TickerSymbolYOU It would be great to see these sales numbers and which car companies they go to. I think that companies buy the Nvidea software to evaluate it against the 'real thing'.

[@donovan1333]

so what do you think? Should tesla be scared? Im looking at robo taxis. Is nvidia a player , leader? etc

[@tumangeorange8388]

Behind the Polstar is Geely China which they state Polstar is the key foothold for Chinese EV step in United States markets. Good tech, but always the thing is the safety issue on chinese EV.

[@TheMurrblake]

What does the car do when it gets hacked at 70 mph?

[@Phtang-phtang]

Sounds very hypothetical and the hardware would be expensive

[@TickerSymbolYOU]

Hypothetical? It's literally already on the road.

[@isaachunt5799]

why i have invested in nvidia. make a ton of cash over the next 10 years.

[@omegaweapon1980]

Basically a wheeled AI robot you can ride in....

[@sergeyzelvenskiy5126]

Nvidia does not have the data to train the model. They only can run rules-based solutions, which would not work.

[@MARKXHWANG]

lol

[@mukamuka0]

In the gold rush, sell shovel. That's the Nvidia stance. They don't in it to make self driving car, they're in to make chip for self driving car

[@Lightsydephil]

Tesla 12.3.4 FSD is flawless, beats any lidar, radar, sonar that already costs a fortune.

[@TickerSymbolYOU]

Tesla FSD costs $12,000 so let's not pretend it's cheap. And it's far from flawless.

[@TreCayUltimateLife]

I want you to perform a thought experiment. Think about where this all goes, where it ends up. It ends up with us creating a simulation of reality that is wholly indistinguishable from reality. We have already done this. You think The Matrix was a fictional movie?

[@jamessolomon8465]

Relevant for robot taxies as this is too much cost added and carried around for regular customers. Tesla has the right idea. They will add sensors to robo taxies the vehicles need to work in all conditions not for regular consumer cars.

[@leomoval]

It's hard to be excited about Nvidia's tech when Tesla is so much more ahead. This is old news with Tesla

[@christian15213]

great video

[@TickerSymbolYOU]

Cheers

[@santiagoalmada4736]

Elon said: radar and lidar is a waste of time and and too expensive... Pay attention..

[@paulschaaf8880]

Personally, I think self driving is a lot farther away than most people think. We're close to, if not already at the point where self driving works under ideal conditions. The problem is for it to be safe, it has to also work in the worst possible conditions and I think we're years if not decades away from that.

[@pwells10]

You should sit in a Tesla and experience the latest V12 with your own eyes.

[@TickerSymbolYOU]

I think the worst conditions are exactly why regulators will end up demanding different kinds of sensors.

[@animation-recapped]

“Years” “decades” did you just leave your cave? 😂

[@brian7android985]

Yeah, I will get one when it can drive me back from the pub down a winding UK moors road. Oh.. and avoid the sheep!

[@aisac21]

​@@brian7android985no need to avoid the sheep then I guess :)

[@xef322]

Nvidia's market cap is bigger than the whole energy sector in the S&P 500.😮 Let that sink in. Way overdue for correction ?

[@mandolorian25]

Tesla made a design decision early designing their own chips as Nvidia chips were consuming too much energy. If you believe most cars are going to remain ice in the future Nvidia may be a good bet. If you think EV costs are going down and in the future most cars are electric they are way behind. Frankly biggest risk to Nvidia is companies will make ASIC chips for there particular needs and won’t need so much GPUs in the future

[@dialac1]

Never underestimate how wrong you can be

[@FARBOLUOS]

Self driving whether by Tesla or whoever is clearly expensive on top of the already expensive EV.

[@edwijaranakulaph.d.2010]

Why do I need my car to reserve a table at a restaurant for me? I drive an Audi A6 and there are already plenty of gadgets that I will never use…ever! 🙂🙂🙂

[@joewilder]

After watching many examples of Tesla's FSD, I would think it's going to take awhile for this system to be trained, and that's if you have good programmers to point it in the right direction.,

[@brucemcintyre295]

It's not so much the "great programmers" as it is the billions of miles traveled with the system gathering data. As noted by others a full net-to-net model requires scads of sample quality data, and not just from simulations. You need to capture as many edge cases where humans provide the appropriate (and inappropriate) responses.

[@Michael-il5wd]

Near

[@Gurubu]

Car manufacturers are going to be sucking off the teet of Nvidia for a long time lol. Also any kind of cloud services get ready for subscriptions

[@pandoorapirat8644]

Maybe i am a simulation in a THOR GPU...