Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

This site may earn commission on affiliate links.
I was thinking about the AI presentation and the fact that The CEO of a company like this has such a significant interest and understanding of something that most would consider an arcane topic. That got me thinking about what watching this must be like for CEO's of a traditional car manufacturer.

[presentation starts discussing discussing D1 computational capacity and bandwidth/latency performance]

CEO: "What is this? He not talking about anything to do with cars. I don't understand any of it... go get me my wise men technical advisors!"

Wise men [watching presentation]: "Ummm... we have no idea what he's talking about either..."

[Collective head scratching commences... completely oblivious to how screwed they are...]
 
Sounds like this person is as good at analyzing AI as Gordo is at analyzing TSLA. Makes sense they would hang out together.
20 sigma! that’s kinda a very big number, since..

“…a 7-sigma event corresponds to an expected occurrence of just once in a period approximately five times the length of time that has elapsed since multicellular life first evolved on this planet.”

 
Now that is impressive.

Clickbait though - the water level came up to maybe middle of driver door. The picture is a little deceptive because it was taken at a level low enough that the front wake of the water wave made it appear as if the car was much deeper in the water.
 
  • Like
Reactions: Drumheller
For anyone curious about wafer scale chips that Dojo is, here is an interesting article from New Yorker about Cerebras, a start up trying to build such chips. Talks about the same packaging and cooling challenges that Tesla touched on.

I would be interested in seeing how this compares to Dojo of course. (Cc: @Singuy)


Dojo isn’t a wafer scale chip. They make a load of D1 chips from a wafer, then assemble them into tiles. There maybe similar challenges cooling a tile as cooling a Cerebras chip, but there are loads of differences too. I don’t think the quality issues that prevented wafer size chips from being done before have gone away, and a Dojo tile won’t have to work around dead chips inside.

On the performance comparison, consider this quote from the article:

Branson found that, in the DNA-sequencing work, the CS-1 was about eighty times faster than the sixteen-node cluster of G.P.U.s he’d been using.​

80x16 = 1280. Then remember that Dojo will be a success if Tesla turns off their ~10,000 GPU cluster. So a Cerebras CS-1 may outperform a Dojo tile, but from the article there’s no indication that multiple CS-1s can be efficiently added together for greater performance (like tiles into an exapod) or that there’s a CS-1 configuration with ten times the internal compute available.
 
I was thinking about the AI presentation and the fact that The CEO of a company like this has such a significant interest and understanding of something that most would consider an arcane topic. That got me thinking about what watching this must be like for CEO's of a traditional car manufacturer.

[presentation starts discussing discussing D1 computational capacity and bandwidth/latency performance]

CEO: "What is this? He not talking about anything to do with cars. I don't understand any of it... go get me my wise men technical advisors!"

Wise men [watching presentation]: "Ummm... we have no idea what he's talking about either..."

[Collective head scratching commences... completely oblivious to how screwed they are...]

Probably true of most US CEOs and others who do not want to see the reality of their upcoming doom ( "If I don't see it it doesn't exist" )

Some of the smarter ones (many Chinese ones I iwould imagine) are salivating and sending new directives to their R&D's. Knowing that something can be done and has been done is very valuable information

Others are keeping mum, not wanting to add to what amounts to free advertising and are in secret planning their moves post bankruptcy. That step would get rid of pension obligations, and clear the path to reincarnating (or selling) their companies for what is their most valuable assets: brand recognition and style, body design and assembly, sales, maintenance and marketing.

Now that Elon has clearly indicated that they would not refuse licensing their FSD tech, car manufacturers could do reasonable EV's subcontracting most everything if not all - like to Magna who assembles the iPace Jaguar etc. Question is would they be willing/ able to invest massively to match Tesla's economies of scale and manufacture

Third party companies like Magna or Bosch (who recently dissolved their driving assist R&D alliance with Mercedes, because they realized it was too hard hah ha) are probably scratching their heads .. were they able to do a deal with Tesla to take care of the nitty gritty mundane tasks of standardization/ packaging for other carmakers, that would be a win win for all.

Most interesting is figuring out what VW's Diess is thinking. Use it as an excuse to fire most of his non competitive computer engineers re probably tough worker protection rights in Germany? Being the first to do a deal with Tesla for their FSD suite?

Finally, if other carmakers had any IQ and balls left, they could band together and force most western countries at least to enforce V2X (vehicle to everything) standards, and build up the signaling infrastructure. Imagine if each car knew what other cars destinations around it, their specs and short term route plans, and each car was aware of its infrastructure surroundings - they wouldn't need much of the advanced AI Tesla has developed. Perfect way to spend $ intelligently on infrastructure, re Keynesian boost to the economies. Not holding my breath at all there, this requires smart, ballsy car makers *and* legislators. Not going to happen... uh except maybe in China.
 
Last edited:
Don't think it is quite that simple. A prospective tenant of mine works in Fremont Factory as a tech on the team that maintains the production robots (e.g. Kuka and that other commonly deployed type). He said (hearsay from me!) they are constantly either breaking down or needing re-alignment, etc.

There will be teams of humans required in some sort of ratio: 1 human for every B Bots to maintain them, too! This will be especially true if said Bots are performing dangerous jobs, IMHO.
Are we sure we just don’t need someone to build a better robot that doesn’t break or need realignment all the time?? It seems to me if something constantly doesn’t work as it should, it simply needs to be redesigned/built to work more reliably. Oh, wait -
 
For anyone curious about wafer scale chips that Dojo is, here is an interesting article from New Yorker about Cerebras, a start up trying to build such chips. Talks about the same packaging and cooling challenges that Tesla touched on.

I would be interested in seeing how this compares to Dojo of course. (Cc: @Singuy)

Seems like on chip bandwidth is orders of magnitude more than Tesla, while off chip bandwidth is much less than Tesla. The two are definitely different in design. Teslas take more of a traditional approach than cerebral as mentioned already. So basically scaling cerebras is worst than Tesla but their single wafer design is much faster. So 1 tile depending on workload could be 10x slower than cerebras. But when you put 100 tiles together it's a different story. The two really need a good head to head using the same compiler to really determine their speed. But from some cherry picked benchmark, the cerebras system destroys nvidia dgx which means it'll destroy d1 tile as well.

However keep in mind that if cerebras can't do what Tesla wants, then it's as good as a dead chip. Tesla is not afraid of spending larges amount of money for their goal as we have seen the millions they have spent on nvidia. The compiler karpathy developed may not work on cerebras hence they didn't buy any. I can't imagine Tesla overlooked a system much better than Nvidia and has been on the market for a few years that only cost 2 million per rack for the cs1.
 
Last night, somehow I got into a thread from 2017 that had some of the posters projections for what this company’s stock will be worth in 10 years. Ultimately, we have already reached some of the highest projections.
That makes me wonder what, in 4 years by 2025, it will be worth and someone reading back on all this will think. 🤔
The next 4 years will be Interesting and exciting times for us, I know that much.
 
I was thinking about the AI presentation and the fact that The CEO of a company like this has such a significant interest and understanding of something that most would consider an arcane topic. That got me thinking about what watching this must be like for CEO's of a traditional car manufacturer.

[presentation starts discussing discussing D1 computational capacity and bandwidth/latency performance]

CEO: "What is this? He not talking about anything to do with cars. I don't understand any of it... go get me my wise men technical advisors!"

Wise men [watching presentation]: "Ummm... we have no idea what he's talking about either..."

[Collective head scratching commences... completely oblivious to how screwed they are...]
I’m not convinced any other OEM CEO even watched it. Maybe they assigned a personal assistant? Or they asked their kid about it?

I am reminded of a scene in Ford vs Ferrari where Shelby ambushes Ford and locks Beebe up so that he can change Ford’s mind about who knows best about making decisions about the race car and driver; the actual race car driver vs the PR suit who’s always in his ear.

This is a significant difference with Tesla and all the rest. The guy at the top actually knows and is involved in his company. He’s actually slept on the factory floor a time or two. Actually picked up a wrench and helped. Actually been involved in design and engineering elements, etc…. He has a distinct advantage over all the other CEO OEMs.
 
There's something to be said for communal knowledge. Adam Jonas admitted for years he didn't know how to value tesla's non-automotive components - although I think many here would argue he didn't even know how to do that. I know nothing of wallstreet but it seems like a lot of these guys are one man shows.

I have been able to contribute thoughts here and there over the years, certainly not as much as some. But the only analyst that appears to have the cross disciplinary mindset about investing that forums like this posses is ARK. Cathy relies heavily on other members of ark. Everyone gets excited about Tasha going on TV but her expertise is the Autonomy side. Sam is the battery and energy storage guy. As a community there are individuals here who contribute various categories of knowledge and bring in sources of info from outside when appropriate.

Gordo claims he spoke to some NN researcher after AI day who said Tesla was 20 sigma behind. Who is this person? Some 19 year old grad student?

TL;DR Id say our advantage is each other and a dash of humility where Wall Street is all about showboating the individual and no humility.
With all the misdirection , spin and outright lies regarding tesla by the media why on earth would anyone here trust them regarding anything else?
 
.
There's something to be said for communal knowledge. Adam Jonas admitted for years he didn't know how to value tesla's non-automotive components - although I think many here would argue he didn't even know how to do that. I know nothing of wallstreet but it seems like a lot of these guys are one man shows.

I have been able to contribute thoughts here and there over the years, certainly not as much as some. But the only analyst that appears to have the cross disciplinary mindset about investing that forums like this posses is ARK. Cathy relies heavily on other members of ark. Everyone gets excited about Tasha going on TV but her expertise is the Autonomy side. Sam is the battery and energy storage guy. As a community there are individuals here who contribute various categories of knowledge and bring in sources of info from outside when appropriate.

Gordo claims he spoke to some NN researcher after AI day who said Tesla was 20 sigma behind. Who is this person? Some 19 year old grad student?

TL;DR Id say our advantage is each other and a dash of humility where Wall Street is all about showboating the individual and no humility.
Gordo's a liar, I very much doubt he spoke to anyone involved in NN's lol!
 
Here is what GJ said “I talked to a neural net researcher this morning… He said that if Tesla is using a simulation, which they said they are, they’re 20 sigma behind waymo”

It seems likely that the NN researcher was not familiar with what Tesla is doing and they answered an ill posed question from GJ.
I happened to hear a slice of that (SMR reaction video) and what Grodo actually said sounded more like Wham-oh -- not sure if that was intentional or just a Freudian slip from the whacked mole. 🙃
 
Actually, my guess is that he took an experimental shot from CureVac -- eating his own dogfood so to speak.
he would have to quarantine if he took an experimental shot. You can only avoid quarantine if you have one of the approved vaccines.

 
DUDE (remember it is OT/Week end), The MD's house on my street that paid a pro landscaper to come in and do a complete re-do of his entire yard and Paid another contractor to design whatever fashionable fake front porch/roof and then repaint the house in the most popular tropical color at the moment, and then tear out the concrete driveway. And now has a gardener come in every season and swap out the annuals, while the gardener's people come in a maintain the lawn and plants...
After a couple of years of my Elon(bot) (yeah I am naming my Optimus Subprime either "elon, or "elonBee" ..or I might start using the "B" word for ro"B"ot... whether calling a robot by its function, SolarBee, or its given name, "elonBee." it could become a universal part of nomenclature. Well back to elonBee, he/it/her(?) could make my home look like something perfect to the level of a Peewee Herman Playhouse dynamic...
And we have this one neighbor who him and his wife do awesome yard decorations for Xmas, Thanksgiving, July the 4th etc... let me see how elonBee can handle that (I have most of the decorations just get less enthusiastic as I get closer to the holiday).
My house could be this when I go to sleep
View attachment 699507

And THIS 24 hours later:
View attachment 699509
And what elonB needs he can either order online or go pick up in my cyber(nota)truck from Home Depot at 2 AM. Why 2 AM? because elonB "likes" going when the roads are empty. And he can just deal with other "Bees" instead of stupid dysfunctional humans in the day time.
And another thing, me and the guy down the street that also has his own Optimus Subprime can agree to share, and let our "Bees" coordinate jobs for more efficiency and tasks that require multiple Bees. "Hey Bill, Listen can you let your Bee come down and hold the ladder so ElonBee can paint the second story of my house for a few hours?"
And now I am thinking how my mango trees will be perfectly groomed for aesthetics and production ALL the time, and the fruit picked and on the countertop each morning... Ooops I see something that is going to not be easy for ElonB to do. Squirrel Elimination. I am sure "he" will come up with something, all I'll do is ask.
Mango trees in the backyard after I get my ElonB.
View attachment 699510
Well, use your imagination... replace the pears with mangos.

My point, the amount of work one Optimus SubPrime can do will equate to the real work of SEVERAL humans. And it will be extremely cheap, and do far better work than most humans. It will transform your personal physical world beyond today's wildest expectations.

I don't know if it is true but I have pretended through my life that it is... They say that Roman Ships were built better than any ships were for the next 1000 years because slave labor was used, and the demands placed on the slave weren't to get it done, but to get it done to perfection regardless of time. So Robots like OptimusB could be programmed on a sliding scale from "Meets current building codes" to Scientific Artisan" depending on various constraints the human wants for the job.

And in closing..if I just had a completely
DUST-FREE home I'd **sugar****************
Eventually they'll pick fruit...they could eventually pick you nose if you needed them too. My thinking for quantity needed in the OP was based on the comments from AI day around speed (5mph) and an expectation that the first few iterations will be like toddlers that can only handle relatively simple tasks.

But agree with your point that after a few iterations and these things put Olympians to shame then the numbers will come down in terms of quantity needed.
 
Here's an article on the boom in orders for EV/car manufacturing equipment. Some equipment suppliers are backed up quite a while which could delay plans of everyone trying to add new capacity. For example,

"We ran out of capacity for any additional work about a year and a half ago," said Mike LaRose, CEO of Kuka's (KU2G.DE) auto group in the Americas. "Everyone's so busy, there's no floor space."

The electric vehicle boom is pay-dirt for factory machinery makers
 
Forward Observing

Maintenance ~ way back in the beginning of time; okay computers becoming a part of the job (for me 1977).

I’ll bet, since I own an X and a Tacoma; that all our 3 & Y owners are boasting at the bar that they have not sat thru a 5K service since owning their car. The current X had one minor glitch in the beginning; while the Tacoma requires oil changes or some fictitious maintenance every 5K. And, all that can be said is, “our chip supply“ is short handed. We bought the truck in February and the X in June 2019. The maintenance schedule is a real pain in the bottom with the truck. Cost is a joke too.

Biggest issue is my wife (better half, smarter half, partner, significant other, better looking of us ~ etc) hardly ever let’s me drive the X! God I need that Cybertruck with all the bells and whistles of X. All I can hope is that the resale on the Tacoma is still high. What will Toyota’s excuse be then?

Keep you eyes (camera) on the ball:)
 
Couple thoughts from a recent 1240 mile road trip in our LR Model Y.

1) The supercharger network is pretty amazing. We had to stop and charge 3 times to get the full 620 miles with some to spare on the way down and back. The 150 kw chargers are not ideal so I found myself trying to plan to spend as little time there as possible. Meanwhile, at the 250 kw chargers we constantly found the car ready to go before we were. Overall, we did find ourselves seeing the same ICE cars on the road which means that we basically kept pace with them even though we had to stop and charge. Now, for those people who drive 8 hours straight at a time, obviously that isn't the case.

1.5) At one stop, we pulled in to a 250kw charger after someone in an ICE car had stopped to have a smoke break, gas, and bathroom break for the kids. He was intently watching to see how long we had to stay and charge before we left. We beat him out and he was surprised. I found myself consistently allocating a 100 mile range buffer from real world miles to get to the next stop.

2) We have FSD on the Model Y and also on our older Model 3 that has radar. We drove the I95 corridor and if you have driven it before its pretty straight and not full of much traffic south of VA. I had to turn off Navigate on Autopilot because it kept wanting to change to the passing lane even though there was no one in front of me. More work to do there. I have also found the phantom braking issue is better than before, but still its not quite perfect. A couple of times the car slowed down for no apparent reason -- my guess is the concrete on the road changed shades and it confused Autopilot. I didn't quite fully trust it in accordion traffic (0-60-0), but thats mostly my nerves about getting rear ended, not the car hitting someone.

3) I was able to pack enough stuff for our family of 5, including a bob stroller, cooler, and several boxes of toys, dry goods, booze, and a pack and play pretty easily in the Model LR Y 7 seater. It is quite impressive how much stuff that car holds even w/ the 3rd row with one kid in it. We had debated bringing our bikes and I am glad we didnt. Before the Cybertruck comes, Tesla needs to figure out how to upgrade the super chargers to allow towed items or bicycles. The network just doesnt make it easy to do that without screwing other people right now.