Expert Tells Sandy Munro: Tesla's MASSIVE Auto Hardware Lead is Due... To Software!

At CES 23, Sandy Munro talked with reverse engineering guru, Thomas Mueller of Wipro about the HUGE advantages that Tesla has in their hardware business--in particular ECUs or Electronic Control Units--due to a more modern, multi-pronged software approach to making cars. Needless to say, legacy auto like Ford, GM, VW, and Toyota, are not doing this yet and it matters!
**To become part of our Patreon team, help support the channel, and get awesome perks, check out our Patreon site here: / drknowitallknows . Thanks for your support!
Get The Elon Musk Mission (I've got two chapters in it) here:
Paperback: amzn.to/3TQXV9g
Kindle: amzn.to/3U7f7Hr!
**Want some awesome Dr. Know-it-all merch, including the AI STUDENT DRIVER Bumper Sticker? Check out our awesome Merch store: drknowitall.itemorder.com/sale
For a limited time, use the code "Knows2021" to get 20% off your entire order!
**Lift pucks for your Tesla: amzn.to/3vJF3iB
**Emergency tire fill and repair kit: amzn.to/3vMkL8d
**If you're interested in a CO2 Monitor, you can check it out here: amzn.to/3PsQRh2
**Check out Artimatic: www.artimatic.io
**For 30% off the amazing OneCarStereo instrument display for Tesla Model 3/Y, use this link: bit.ly/3vd7NRv
**Get 20% off an awesome Ekster wallet and/or wallet locator(!), check out my affiliate link here: shop.ekster.com/DrKnowitall
**Want to get in on WeBull's awesome Crypto and stock fun? Check out this link, and get started trading stock and Crypto!
a.webull.com/i/DrKnow-it-allK...
**If you are looking to purchase a new Tesla solar roof, solar tiles or PowerWall, just click this link to get up to $500 off! www.tesla.com/referral/john11286. Thank you!
**You can help support this channel with one click! We have an Amazon Affiliate link in several countries. If you click the link for your country, anything you buy from Amazon in the next several hours gives us a small commission, and costs you nothing. Thank you!
* USA: amzn.to/39n5mPH
* Germany: amzn.to/2XbdxJi
* United Kingdom: amzn.to/3hGlzTR
* France: amzn.to/2KRAwXh
* Spain: amzn.to/3hJYYFV
**What do we use to shoot our videos?
-Sony alpha a7 III: amzn.to/3czV2XJ
--and lens: amzn.to/3aujOqE
-Feelworld portable field monitor: amzn.to/38yf2ah
-Neewer compact desk tripod: amzn.to/3l8yrUk
-Glidegear teleprompter: amzn.to/3rJeFkP
-Neewer dimmable LED lights: amzn.to/3qAg3oF
-Rode Wireless Go II Lavalier microphones: amzn.to/3eC9jUZ
-Rode NT-USB+ Microphone: amzn.to/3H2FmK2
-Focusrite Scarlette 2i2 audio interface: amzn.to/3l8vqDu
-Studio soundproofing tiles: amzn.to/3rFUtQU
-Sony MDR-7506 Professional Headphones: amzn.to/2OoDdBd
-Apple M1 MacBook Pro: amzn.to/3wPYV1D
-Docking Station for MacBook: amzn.to/3yIhc1S
-Philips Brilliance 4K Docking Monitor: amzn.to/3xwSKAb
-Sabrent 8TB SSD drive: amzn.to/3rhSxQM
-DJI Mavic Mini Drone: amzn.to/2OnHCEw
-GoPro Hero 9 Black action camera: amzn.to/3vgVMrH
-Suction car mount for camera: amzn.to/3tcUfRK
-Extender Rod for car mount camera: amzn.to/3wHQXsw
**Here are a few products we've found really fun and/or useful:
* NeoCharge Dryer/EV charger splitter: amzn.to/39UcKWx
* AI Superpowers, by Kai-Fu Lee: amzn.to/3dgRSs3
* A camping mattress for your Tesla model S/3/X/Y: amzn.to/3chSRrQ
**Music by Zenlee. Check out his amazing music on instagram -@zenlee_music
or KZbin - / @zenlee_music
Tesla Stock: TSLA
**JOWUA:
*For an awesome wireless charging phone holder mount, check out: bit.ly/3tRMSQR
use the Promotion Code: DRKNOW5
*Tesla USB Hub: bit.ly/2PXHPit
use the Promotion Code: DRKNOW5
*Powerful wireless iPhone/Android charger for your Tesla Model 3 or Model Y, check this link: bit.ly/3b5fHSW
use the Promotion Code: DRKNOW5
**EVANNEX
Check out the Evannex web site: evannex.com/
If you use my discount code, KnowsEVs, you get $10 off any order over $100!
**For business inquiries, please email me here: [e-mail]
Twitter: / drknowitall16
Also on Twitter: @Tesla_UnPR: / tesla_un
Instagram: @drknowitallknows
**Want some outdoorsy videos? Check out Whole Nuts and Donuts: / @wholenutsanddonuts5741
Munro Video: • 100 Million Lines of C...

[@ArendJanV]

Thanks for sharing. The great thing about following Tesla is that it gets better all the time.

[@craigmccracken3104]

Thank you, didn't realize so many of these things Tesla was doing, great news!

[@bardigan1]

Fantastically informative, many thanks for the hard work.

[@Tomm9y]

Happy Birthday! Thank you for this analysis of the issue, interesting and relevant.

[@richf3654]

This interview is the first time I've seen Sandy basically speechless 😁

[@jpl0202]

IMO Sandy is an old-school engineer and was not raised on the computer side of things. Thomas knows the software so Sandy is listening with both ears.

[@zdme4864]

I think 🤔 you are right. Sandy knows his stuff… Recognizes genius and insightfully can direct the shortcomings/ opportunities of improving mechanical things . This other engineer had his respect for his own abilities and insights as to software and technical awareness on a different areas of expertise worthy of Sandy’s silence and attention!

[@eyesuckle]

Is it just me, or does Sandy look as if he's considering punching Thomas?

[@philipdamask2279]

I think he is trying to understand what the guest is saying to see if he can make sense of it, Most of this stuff is not in Sandy's wheelhouse of knowledge.

[@hungly9776]

Edge processing is what Octopus does. Their tentacles have brain cells.

[@mikelally97]

Holy crap! Mind blown! This is 4D chess that has trillion dollar potential. I believe at some point Tesla will sell their vehicles at cost and make most of their vehicle profit through FSD, subscriptions, royalties, ride sharing, and supercharging. A Tesla car is just the Trojan horse.

[@cjjuszczak]

Re Tesla doing compute at the edge, prefiltering and transmitting a fraction of what the car senses to HQ, reminds of how Google used to do "Google Assistant" in the early days. When you asked your phone "Hey Google, how old is Johnny Depp?", your phone would record your "request" as a mono audio file, send it to Google, the audio would be decoded on huge data centers using the language AI models to understand "what is the question", then the answer arrived, and delivered to your phone to be spoken using text-to-speech (or text only if you chose that option). This was the same regardless if you asked the phone "hey google, turn on WiFi", or "Hey Google set alarm for 7am", the language decoding was all done on Google's HQ. Later Google would condense the language AI model enough to fit in your phone, decoding the request in real-time using your phone's tiny processor, and send the "already decoded question" in a form Google HQ can already understand, and give the answer much faster to your phone to be delivered to your ears/eyes as before. Now we have image to text decoding too using Google Lens, etc all done on-device at the edge. It's amazing the level of inferencing done on tiny SoC's :)

[@carl-Sp]

Great insights. Thanks for *harvesting* this info us doctor. 🙏

[@markkoons7488]

Gotta love what you bring to the understanding of ignorant enthusiasts like myself. Thank you.

[@brentftaylor]

nice, clear, explanation

[@buyanmandakhbattulga5708]

In-depth analysis. Thanks for sharing!

[@thatscottishengineerguy9606]

Great episode! I watched the original interview when Munroe posted it and I was fascinated. Your additional inputs and explanations only enhance my interest. I'm in the mechanical discipline, so software isn't exactly in my wheelhouse. However, Munro's video and your video explain the concepts and their implications with just the right amount of detail. Nicely done.

[@patmcdaniel2016]

ok, I subscribed finally. You and Lars from Best in Tesla are the best at what you are doing.

[@eugeniustheodidactus8890]

Thanks for breaking this down!

[@helionfilm1992]

Happy birthday to you 🎉 And thank you for the great and insightful videos .

[@johnreese3762]

Very well done! Happy Birthday!

[@rlshultz5841]

Good analysis. So it’s really not limited to FSD. Tesla could license a standard vehicle software to control the basic or standard operations that could be separate from hardware. That’s makes sense.

[@turbodewd1]

lol FSD doesnt work!

[@johnpoldo8817]

John, your report tripled my understanding of Mueller / Munro video. Thank you !

[@DisruptiveWealthCreation]

Nice breakdown and analysis.

[@bru512]

Exactly. Serious software developers at Tesla design their own hardware so they have the cheapest, most integrated solution. At Tesla, ECUs (manufactured on older chip dies) are being eliminated as quickly as they can manage. As Tesla scales they can afford to invest in replacing more and more of them. or put another way... The best ECU is no ECU. This is also why legacy OEM's have no chance of following because part suppliers like bosch don't have any off the shelf components to sell them, because Tesla never commissioned bosch to make them.

[@zhchbob]

NVIDIA and INTEL can help them.

[@bru512]

@@zhchbob How?

[@user-ll4lf5go1m]

And yet their build quality is terrible and the constantly miss new product entry dates by YEARS.

[@markpaterson5568]

Great vid 👌

[@Indecisive7337]

Another great video, a lot to take in but thank you all the same.

[@kstaxman2]

Always to great to hear truly knowledgeable people talk about the way Tesla does things so differently and why that is a good thing. In every way Tesla is ahead in the software that cars need today. It's simply astonishing to hear and thing about how different everything is in a Tesla over other cars. And the public and the media are basically blind to what the differences are. No wonder companies like VW have failed to produce comparable software and features.

[@alfredogonzalez1280]

Excellent comments !

[@noleftturns]

Blame the MBAs for this mess Legacy Auto is in. MBAs are generally not engineers, and each looks at a car differently. The engineer is interested in efficiency, safety, reliability, etc The MBA is consumed by saving a penny and could care less about anything else. Legacy Auto listened to the MBAs and broke up the car into hundreds/thousands of parts that went out for bid around the world, with China getting most of the bids - the cheapest bidder. Throw in a month-long supply chain, and the current new car being sold is using computer chips 5-10 years old, and none of the parts talk to each other - a hot mess. That's how parts don't talk to each other, and the efficiency of the car went to hell so did reliability - a car made up of hundreds of companies that submitted the cheapest bid is a sure way to sell crap. Elon took the opposite approach, and the result is far superior and even cheaper. Throw in the fact that changes can be made 100X faster than the MBA way of building a car, and you have a car that is safe, reliable, and able to use new technology overnight.

[@bru512]

Exactly. Elon took this approach because they had to, (ie suppliers like bosch was not selling what they needed) Now that Tesla is successful and have sufficient scale, they are finding that this is a huge advantage. Tesla can make changes in minutes to their own software, and are not reliant on the supplier to make changes.

[@nigeljohnwilson4549]

Happy Birthday🎂 Great job! 👏

[@aerohk]

Too good, mind blown 🤯

[@paulmichaelfreedman8334]

Shadow mode is something I would do if i was to design such a device. I love efficiency in electronics. Don't let cycles go to waste if you can let them do work.

[@cgruita]

This is a great video, very informative, way beyond the typical Tesla articles we're seeing everywhere! Do you or anyone else know where to get a primer on the software Tesla is using? (technology, trends) I'm a software developer and I'd like to know, maybe even get into this, who knows!? Thanks!

[@mxbreed20]

Just brilliant. Brilliant. and happy birthday

[@richardkule9384]

7:28 Mr. Mueller said the ECU is under the passenger seat, not the Driver seat.

[@Tanstaaflitis]

Regarding the edge computing comments, Tesla talked about this advantage at one of the AI days. Specifically, they can serve requests to the fleet to gather specific examples for further training. Some made up examples: "send clips and telemetry for intersections with occluded opposing traffic", "send views of a specific location from different positions and lighting conditions", "send down mountain driving at night in a snow storm"..... So Shadow Mode isn't just monitoring how the stack performs compared to a human driver, it allows the fleet to be a queryable resource.

[@NickMackenzieMD]

I like your new music.

[@cicero5451]

Happy birthday Sir

[@ggrthemostgodless8713]

There is no link to the original video in the description, but a ton of merchandise links.

[@Rovinman]

Happy Birthday, Doc. ! Very interesting subject, it may be better, eventually, if there is a standard layout for cameras on the cars, which will make it better for both Tesla, AND the car companies, licensing the software !

[@balrogger]

Basically, Tesla becomes Microsoft for all transport applications

[@chrisbraid2907]

Nope Unix. M S would crash too often …

[@sergiosergio12345678]

No, Microsoft is unreliable, becomes Linux

[@Dularr]

Tesla will be Apple. The question is who will be Microsoft Google Amazon Apple

[@galvinstanley3235]

More then that he has completely changed the world.

[@freddybell8328]

@@Dularr no apple is a closed ecosystem

[@joebushnell143]

Happy Birthday 🎂 🥳 🎉 🎈

[@kevinchampagne6661]

Just watched a podcast showing ChatGPT can write, evaluate and correct code in seconds. Amazing!

[@joergplatzen9620]

Happy happy Birthday!!

[@ossiepenn2154]

Happy Birthday!

[@harrynoren7150]

Happy Birthday!!!

[@michaellatta]

A counter example to the assumption about using “standard” software, is that Tesla built their own ERP rather than usjng something like SAP.

[@bru512]

Exactly. Any company using SAP will NEVER be agile.

[@aeromtb2468]

best thing not to use SAP. id like to see their ERP.

[@JosephWolfgram]

As a CTO, this is the point that made me pay attention to Tesla. They are super-smart in all aspects of their business. 👍🏼

[@austinpowers1999]

So no mention of the PPP?

[@bru512]

@@aeromtb2468 That would be fascinating. I spent early part of my career writing ERP solutions.

[@brianbeasley7270]

Happy Birthday!🥳

[@cmosack]

Happy birthday!

[@TheGaussFan]

My guess is that a lot of the chip shortage improvisation was using various programmable logic devices to temporarily replace inexpensive devices that were not programmable.

[@jamesdaleo3643]

The C programming language is still very popular for embedded systems.

[@HermanWillems]

Yes C is a bit of a "standard" because of its ABI, slooooowly Rust is getting used more. But not as much yet. C is still the default.

[@alanjameson8664]

My first car was a 1952 Buick Roadmaster. This video makes it look awful good.

[@whowhy9023]

Tesla does not use SAP or Oracle they created their own unique system long ago & its integrated with their AI and runs all their factories , sales etc.

[@fredhearty1762]

The 'Harvester' is perfect name for the functionality that finds the needle-in-a-haystack and sends it back to the mother ship. This IS a software company.

[@rameshvenkat6907]

Happy Birthday.

[@Pete35vail]

Happy Birthday

[@HerrJarvinen]

I have a feeling that Dan O'Dowd pioneered C in automotive industry ECUs, and never looked forward again.

[@allanbartram1849]

It amazes that no one has ever approached Tesla for a licence??? Great breakdown sincere thanks and regards from South Africa

[@allanbartram1849]

Here we go Sandy- if you have Elons ear- try teasing him on why he doesn’t set a huge profit trap with an FSD auction - offer the first legacy or other taker / best offer for an FSD licence with sole usage with Tesla with a minimum take number for one year- conditions are that after one year another taker can get a similar deal and so on? Laggards may see this as a only way out of something they cannot do alone Regards from South Africa.

[@danielhull9079]

1:01, the code and ecu situation will improve with the High Performance Compute silicon coming with some Blackberry QNX muon kernel tech imo.

[@danielhull9079]

The underlying OS could be Blackberry QNX Real Time Operating System imo.

[@ralphwagenet852]

But Tesla wrote its own ERP software ... buying off-the-shelf software gets generic solutions. Custom software, if you're good at it, can be much more powerful because it focuses exclusively on your needs and isn't a general purpose software that tries to meet every possible need of every customer.

[@timomuller7345]

Great video. But basically everything regarding the second AI chip and also shadowing has been explained on detail on Tesla autonomy day in 2019. :)

[@walthooker1434]

Happy Birthday weekend, John. Celebrate it with Miss Information.

[@Ask-a-Rocket-Scientist]

That other auto companies are dysfunctional with regards to SW does mean Tesla has an insurmountable lead. That their processor is centralized vs distributed is also not insurmountable.

[@garycarson3128]

John, you started to mixup the EV operating system software that controls ECUs, sensors, the physical components of the EV with FSD software. Tesla has solved the problem of developing the EV operating system software. They are currently working on solving the FSD software solution. The legacy automakers have to solve the EV operating system problem first. I suspect that Apple and Google are working on an general purpose EV operating system that they can license to the legacy automakers. This software will include over-the-air updates, the user operation interface, entertainment system, and integration with the dozens of ECU suppliers. This is similar to how Google created Android for all the other smartphone companies, each with different hardware, cameras, screens, buttons, microphones, etc. After the legacy automakers get the EV operating system, they could then look to Tesla to license FSD.

[@Chainyanker007]

Also software engineers from Tesla and SpaceX are allowed to transfer to the other firm. I think some of the best software engineers in the world work at those places. This acts as a recruitment tool to attract the best that’s out there, including the best graduating students. The other students end up at legacy auto.

[@Random_Innovation]

Yo, Happy birthday!

[@universeisundernoobligatio3283]

He talks about SAP, my company switched to SAP, SAP is pure evil, just a huge money maker for SAP and hour eater for users.

[@SurfectedGermany]

Great insights well done!! Thank you John! 🙋🏻‍♂️😊

[@macrumpton]

Considering the unused processing power out in the world, it is amazing that companies are not having idle devices mining bitcoin or some other $ generating activity.

[@roberthotaling4303]

Love this video. Thx for taking the time to pull that out. I love the AI chip part of tesla.

[@russadams3008]

I do monitor upload traffic for my m3. After doing a day trip to Fredericksburg VA from Alexandria VA, the m3 uploaded 1.08GB of data. Seems large if it's filtering out what it sends back. Also, what about MobileEye as a licensing source?

[@davidbeppler3032]

You have FSD. It collects far more information from you than others. In an hour trip it collected 1.08GB to send back, but think about how much it collected that it did not send. Also, this shows either you are an excellent driver, or a very bad one. lol

[@russadams3008]

@@davidbeppler3032 I had a score of 98 when I got beta several years ago. 😁

[@mikemccarthy1638]

The “Al-Gore-Rhythm” wishes you a happy birthday 🎉 😊

[@marcusnichols5595]

! suspect that some cars are sending raw unfiltered data - likely those vehicles who trip multiple edge cases.

[@_PatrickO]

Very few would do that. Tesla designed the system so they can send search parameters to the cars. They can then only upload data that matches the search parameters or if a user reports something that they are interested in. That is the genius of the system. The filtering of the data happens on the car, so they only upload what they are looking for. This greatly reduces bandwidth and storage in their data centers. Their cars are essentially a distributed super computer filtering tons of video and telemetry data on the car instead of in a data center. Filters can be anything like objects, user flags, gps coordinates, weather conditions, time of day, etc.

[@ianollmann9393]

Tesla can do what it does in software because it is embedded in Silicon Valley with Apple, Google, etc. and can get great engineers. If you have your engineering team in Detroit, this is going to be harder.

[@nmoritz]

Mind-blowing. Why is it Tesla is doing it before anyone? In my opinion, the answer is SpaceX. That’s the kind of hardware / software combo you have to have in the space industry.

[@neiliewheeliebin]

Because hes taking a risk, its not exactly redundant if its actively being used

[@vinayr.6015]

Now this makes me wonder, some of the Chinese Auto companies, Nio,Xpeng, sounds like their EV is ground up EV design or heavily modified/copied version of ICE vehicles ?

[@davesradiorepairs6344]

I agree with everything you've said, except in one regard; Other auto companies must decide NOW, and not in 2-5 years, or it's too late... Tesla is doing so many wonderful things, full steam ahead, yet their are still so many detractors and antagonists against what Tesla and their successes.. Maybe it's the same people that pay huge political donations to have things their way maybe ? Just guessing of course..

[@dr-k1667]

The moment when Tesla's prowess as a company shifted in my mind was during the chip shortage. When I heard they bought the chips available and wrote the code they needed I understood that NO ONE else is playing the game like they are. Having some understanding of what that move meant... my conviction grew as did the lead Tesla was able to take and grow during the most time outside of a world war that we have just lived through. Headwinds seem to turbo charge Tesla because so far they just keep showing that they are not doing things like everyone else.

[@user-ll4lf5go1m]

So now they have a bunch of non standard cars running around.

[@whowhy9023]

Twitter also runs on antique code. Elon is in the process of moving to modern code language now.

[@davestagner]

Speaking as a long-time enterprise professional… it’s not about “language” (even the software on your phone is written in many languages, and I’ve nested different languages four layers deep in a piece of software), and it’s not really about “modern”. The biggest difference is code quality. Brand new code that is badly written will KILL you as you try to maintain it, but well designed and well maintained code can work for decades.

[@davidbeppler3032]

@@davestagner Are you suggesting that Legacy Auto code in C is well written and efficient?

[@davestagner]

@@davidbeppler3032 I’m suggesting that a: the choice of programming language has very little to do with code quality (and considering the entire Linux operating system that underpins most computers today is written in C, it’s not like you can avoid it), and b: the idea that Twitter or any other large system is a monolith written in a single language is WILDLY wrong. It is not at all unusual to find 2-3 different languages embedded even in a single program (like sql code or called shell scripts). And a system the scale of Twitter consists of thousands (or tens of thousands) of different distinct programs, some of which may date back to Twitter’s founding, and some of which were written within the past year. So saying something like “Elon is in the process of moving to modern code language now” just betrays an incredible ignorance of how large-scale software development and legacy code actually works. (And fwiw, the “modern code language” these days is largely NodeJS, which I consider a pox on code quality. It’s possible to write good NodeJS code, but it’s much harder imho.) And don’t knock C. A lot of what’s going on in a car is embedded systems - sensors, motor/device control, etc - that are often proprietary or custom-made. And C is widely used for such code, because of its fine-grained control over memory and processes, and the ability to easily jump out to assembly language if needed. You ain’t writing no LIDAR device controller in Javascript! Mostly because it’s impossible. So anyway, if you say “legacy auto code written in C is well written and efficient”, are you talking about the device controller for the automatic door locks? Or the operating system for the main board that sorts out the engine sensors? The fuel injector sequencing code? The back seat DVD player? Those are all different programs, written by different teams at different times for different needs. Some of it is probably good, most of it is probably bad, because frankly, most software is bad.

[@DanaVastman]

You DIDN'T post the link! 😡

[@billholder1330]

Hahaha - to hear C called "boring old spaghetti code" - and here I'm still coding in raw assembler! lol... Time to Retire, methinks...

[@JulioGarciaLx]

The use of dozens of ECUs is the great Achilles heel of OEMs. The absence of a coherent system prevents many things, the most critical being updating software like Tesla does. How to update a feature that is composition of parts of several ECUs? You can't and OEMs can't too.

[@keithdow8327]

You can't tell a whole lot by looking at a chip. You get a big picture, the process technology and not much else.

[@flaymil5373]

As long as fsd can figure out Manhattan New York them I’m sold.

[@royceyoung]

What love of you could interview him and get more insight into teslas software advantage.

[@ricinro]

the half dozen times my model Y has phantom braked the car never allowed me to inform Tesla that an serious error occurred. Tesla probably thinks the car was preventing collisions. Service people told me that autopilot (I don't have FSD) was still in beta. I can't anticipate or counteract phantom braking so I can't safely use something as simple as cruise control.

[@andrewpaulhart]

Havnt we known that Tesla vehicles are selective about what data is sent back to Tesla for a very long time? My understanding has always been that Tesla requests what specific types of data it wants from the fleet and that the cars only sends back what has been requested.

[@mcpounds4228]

Sandy's guest said the chip is under the passenger seat.

[@fractalelf7760]

I must have missed something - what happened to the cool intro music?

[@wotireckon]

You mean the really really irritating intro music? Good riddance - this is much better!

[@fredbloggs5902]

Nobody cares.

[@fractalelf7760]

@@fredbloggs5902 You cared enough to comment - dumb comment debunked.

[@fredbloggs5902]

@@fractalelf7760 I cared enough to comment on your inane comment. Cluelessness confirmed. Blocked.

[@fractalelf7760]

@@fredbloggs5902 I cannot believe this is a watcher of this channel. Wonder if he knows how proud the host was of the music and pushed the composer? Sad.

[@skinnymoonbob]

Tesla has some EV competition. Tesla has none autonomy competition.

[@fredbloggs5902]

True, but competition is good, it keeps everyone on their toes.

[@MrBroady02]

@Ben Dover There will come a point where tesla have saturated their own market with FSD. Competitors will wave their white flag and cheque book and then tesla will be able to sell autopilot to other companies. They aren't a car company, their a tech/energy company, and car manufacturers work together all the time.

[@zhchbob]

@Ben Dover You really know China.

[@skinnymoonbob]

1, at most 2 players from China, and Tesla. Like Android and IOS.

[@andrewsaint6581]

@@skinnymoonbob I think BYD, then SAIC, then Geely and maybe GWM Ora. I've tried the Ora Funky Cat and MG 4 in UK and they're astonishingly good.

[@johnmontgomery3169]

The best data is no data!

[@HughButler35]

Whenever one uses unused capital, cost is essentially zero. Think Airbnb. Idle assets. Software enables extra value. That's shadow mode. Tesla harvesting free data.

[@davidbeppler3032]

No Legacy Auto companies are building cars that can use Tesla FSD software. That would require them to add $2k to each car for free at the factory or spend more than $5k after market.

[@dickritchie2596]

What is the global forecast for nickel, cobalt, magnesium and lithium production and price for the next 3-5 years? Now that we have 10 years plus experience with EV’s, what is the cost of reclamation and disposal of EV batteries?

[@johndoe1909]

the c language is not a problem. still the very best language around for this type of application. but the hardware stack in current cars is a musmasch of various hardwares loosely interconnected with one another theres the real problem, ther is no overaching architecture to speak of, all coming from various suppliers with dubious testing and integrstion abilities.

[@fsbbs]

I‘m a fan boy, too, but…. The sensor data consolidation (7:00) is done by Mobileye in millions of OEMs cars for quite some time and builds their basis for what they call REM. In an interview they emphasize on this necessity to reduce data volume to be transmitted and give the number of approx 20 kbytes per mile IIRC. If you’re interested, search for „CES Mobileye Shashua“. It’s not only Tesla doing cool things. However, my feeling is, Tesla has a significant advantage in its OTA update ability.

[@MrRyanDevo]

The sensor data consolidation - yeah, I imagine mobile eye does that. But do they use the failure back-up chip as a shadow-mode processor to perform the consolidation?

[@fsbbs]

@@MrRyanDevo I suggest to find out. In my opinion, it would be wise to better understand other approaches. If it’s our attitude to think, Elon is Techno King anyway, then we would do, what Techno King explicitly not suggests: just thinking we are smarter than the others. Anyway, it’s my guess, that Tesla has the higher quality in their collected data for several reasons, and their machine learning power leads to the better generalized behavior independent from HD maps. - (answer: no, they do not have spare CPUs.)

[@NickMackenzieMD]

This supports my thesis that Tesla motors will end up making more profit from its software services and licensing than from selling hardware. You might dive into the value proposition of the Virtual Power Plant at scale.

[@Kenlwallace]

Yes, within only years, legacy auto must get FSD to survive. If only Tesla has it with affordable hardware they will virtually be forced to use Tesla’s software @$15k ish each. So essentially every car made by the ‘competition’ will be paying Tesla a $15k (ish) margin (about the same $ Tesla makes from selling their own cars). Ipso facto, if Toyota sells 10M cars in a year paying Tesla about $10k for software, Tesla’s essentially sold another 10M cars with a $10k margin (but no OH cost). Right now we think “that’s crazy, not all sales will have FSD” …. And right now we’d be right … maybe not so in five years imo.

[@NickMackenzieMD]

@@Kenlwallace I am with you. It is sort of like Microsoft Remote Desktop and Citrix client access licenses. My company has no real option but to pay for them to support our software-as-a-service product and these license costs now make up 66% of our data-center costs.

[@1farlan]

Thanks for these videos, they are really informative and thought provoking. The problem of the automobile industry IMHO is that their success stems from the fact that they were “mechanical” successes before the computer age, so they produced a business model based on solid mechanics and over the years added electronics in the MINIMUM as necessary. Lots of tiny pieces added to the mechanics. Now they are having difficulty re-conceiving themselves as computer companies building computers that transport themselves with payloads and adding peripherals as necessary in a minimal fashion. Tesla, and possibly Apple - AMD - Nvidia, who can design their own CPUs, GPUs and memories (and efficient ones at that) may spell the death of legacy automotive companies. Imagine a SOC which contains an “automobile, AI,” and connects to motors and sensors like Lego pieces and only requires a shell and other items like steering wheel, seats, doors, etc. to make the payload comfortable and protected. Air conditioning, radios, televisions, even windows could be optional/additional.

[@jpl0202]

Agree! The auto companies' upper management who are making decisions are very scared to change their business model which has worked for decades. Thus, a disruptor like Tesla steps in and survives the most challenging part of building a car company, the earliest years, producing a car that people really want and trying hard to make it afforadable. That is why Elon is such a genius. He has an overarching design of how everything will work together and has the balls and the money to implement his idea. I love how he is always giving credit to his employees. He knows he can't do it alone so he hires the best and brightest and lets them lose to maximize their productivity.

[@bgebbq314]

Add Google to that list. They are designing chips too. One can see an "automotive OS" for EVs in the future for the car's functions. ADAS and FSD will reside in another layer.

[@john_hind]

I found this interview quite puzzling. Sandy is constantly advocating vertical integration and even here he is saying doing their own software is a key to Tesla's success. Yet he agrees with Thomas that car companies should not go it alone with software but buy it in. Also, one of the reasons for multiple ECUs is to reduce the wiring, another thing Sandy has advocated. One central processor would mean all the sensors and actuators had to be individually wired back to a common point. Multiple ECUs allow multiplexing of signals onto, typically, a pair of 'bus' wires plus power and greatly reduce the total amount of wiring. Maybe is is just a matter of reaching a happy medium though as 100 million lines of code and north of 100 processors is getting a little excessive!

[@richardheck3794]

Legacy auto doesn’t have the luxury of development talent to develop a new system, or the free cash to create a system from scratch at this point. Cheaper to buy. Tesla has been building over time so doesn’t have to make any transition.

[@john_hind]

@@richardheck3794 Yea, I remember VW crowing about their 10,000 software engineers and thinking, well there's your problem right there, you've not read "The Mythical Man-month"!

[@jmattoxriskpro]

Because they will never catch up or keep up with Tesla. That would be like GM deciding to overtake OpenAI and DeepMind in the next 5 years. Literally impossible for a non software co

[@dmpease01]

Hi John, I'm of the mind that that the alternate chips that Tesla got weren't reprogrammable. I think they adapted their main car system to talk to the new vendor's protocol.

[@gordonlowe5405]

Just subscribed and Happy Birthday! Are you planning to go to the next Tesla takeover event in San Luis Obispo later this year?

[@michaellatta]

There was discussion on ai day 2 or something similar about splitting tasks between the two chips. My take on it is that they no longer use the second chip as a backup. I assume this is currently because the human is currently a backup, and the chips have been proven reliable enough to not need a backup. There was also discussion that the hardware 3 chips are not enough to do full FSD on a single chip.

[@john_hind]

The point is you have a 'failover' chip that is not being used for mission critical activities in normal operation and you use it for value-add but non-mission critical tasks, such as compressing training data to be sent back to Tesla. If the primary chip fails, you can ditch those value-add activities and use the failover chip for the mission critical activities.

[@charlesvanderhoog7056]

Imagine the smug directors of GM, none of them knows what this is all about. I really would love to overhear GM's executive meetings.