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.
They are targeting low-wage in-person interactions that could be better served by a robot. Think:

- Delivery drivers (Hi Lyft)
- Takeout/Fast Food (Hi Uber)
- Outside work (Hi TaskRabbit)
- Personal Care (Hi Barber App)

No they're not. They're targeting the kinds of tasks that will be needed on Mars, BEFORE the first human colonists arrive. Think repair, maintenance, mining, construction.

The jobs you've listed are all the one's going on Golgafrincham Ark Ship B. :p

 
D1 (400w TDP, 645mm^2, 7nm)
BF16: 362 Tflops
FP32: 22.6 Tflops
On Chip Bandwidth 10TBps(or 1250 Gb/s)
Off Chip bandwidth: 4TBps(or 500GB/s, 25 D1 per Tile)
Off Tile bandwidth is 36TB/s reported (I think 9TB/s is more like it for tile to tile communication), 3000 D1 chips connected together

AMD Radeon MI100 (300w TDP, 750mm^2, 7nm)
BF16: 92.3 Tflops
FP32: 23.1 Tflops
On Chip bandwidth : 1228.8 GB/s
Peak Infinity Fabric™ Link Bandwidth 92 GB/s (offchip)

Nvidia A100 (400w TDP, 826mm^2, 7nm)
BF16: 312 Tflops
FP32: 19.5 Tflop
On chip bandwidth: 2,039 Gb/s
Off Chip bandwidth: 600GB/s (up to 12GPUs)
Off Chip bandwidth PCI4: 64Gbs

My opinion, the actual D1 chip is pretty good, but not mind blowing given the size/power usage. It's inline with Nvidia's best. But dat scalability holy S balls with them interconnect bandwidth off tile. Mind blowing...
 
Last edited:
Earlier today, I would not have believed the robot was possible in my lifetime. But hearing this from Tesla, it's now possible in my mind. For me it was convincing, but I've seen what humans can accomplish out of Tesla so that bar is quite high already. Most will see fantasy.

Fact is, I'd pay $100K for one of these even if all it did was follow me around, but knowing it will get better with every wireless update. It just needs to walk and talk at first... SOLD! 10 bagger down the road on this one I think. Buy and hold, no telling when this takes off. 🚀🚀🚀
And when it follows you into the next room you can ask "Optimus, why did I come in here?"
 
I was a bit late to the party. Just finished wathing the presentation. I am blown away - but I must also admit that I am blown away quite easy on the topics discussed. I did not understand half of what was said, not only due to the terminology but I had a challenge understanding the Indian guy. He had a very strong accent. He was genuinely enthousiast and I guess the thing he showed us must be very special. I would love to hear from people who are experts in the field of what they think about the presentation.

The robot is great. Can’t wait for one doing the ironing for us. Would the robot be the machine that builds the machine that Elon was talking about? These robots replacing more and more people in the factory?
 
Since everyone is offering theirs - my opinion on the event:

- FSD explainer/walkthroughs looked great, but I have no idea where competitors tech is at or what they use by comparison so tough to say how impressive it all was. Keen to hear from those versed in the area how good it was.
- D1 / Dojo was amazing, very impressive!
- the guy dressed in a robot suit dancing was very stupid
- Tesla Bot announcement was amazing! Those having trouble seeing why Tesla should make humanoid robots needs to think harder. Use cases are incredibly widespread, and the AI is the hard part which Tesla is already in the process of solving - the actual hardware is child’s play by comparison and Tesla might as well be the ones to dominate the likely multi-trillion industry as soon as it eventuates. Also I can’t imagine any AI researcher who wouldn’t want to work for Tesla now.
 
Robots for firefighting? nuclear cleanup? Mars colonization?

A day after Teslabots start rolling off the assembly their will be a startup getting a billion dollars in funding for sexbot conversions.

That is what humans do. The internet will be the greatest tool to bring education to every human on Earth? It delivered porn to Afghanistan, Madagascar and every corner of the globe.
 
Ok my opinions. I understood like 99% of what they said.

1. I am a bit disappointed with the reports from the ground. It was a recruitment event and everyone is talking about it like it was a product reveal. I want to know who was in the audience, were they impressed, did people line up to hand in their resumes etc. Not about the product revealed. Anyway I was impressed. Imo to understand this event I think you need to read ”Genius Makers” to learn how companies work to attract AI talent and how talent are attracted to companies. The question about open source was an important one. I think Tesla will attract engineers who want to do cutting edge stuff, not academics who want to contribute to academia. Not like what Facebook, Microsoft et al had to do to catch up with Google. Karpathy is likely a very strong magnet for talent.

2. Tesla Vision stack is seriously cutting edge. Rate of progress is epic. I liked what they showed about how they handle updates of their internal map and how they select which features at which times to update when stationary. Have some questions about loop closures when going around a block I wish someone from the audience would have asked.

3. Their path planning search was awesome. Better than I had imagined. The problem with end2end is that currently we don’t have a good way to implement search in neural networks. So they use these feature layers in vector space and then do search in that. Also once we have a way of doing search by NN, not only with NN things will get strange, wonderful and scary. We already have a version of it here greatly expanding what questions a neural network can answer and imo we are getting close to general AI with search by neural networks.

4. HW4 introduced with CT. Not sure if it’s so great to buy a HW3 car atm, maybe a lot of benefit to wait for CT and then it should not take too many months for the rest of the vehicles to switch.

5. Hacker news at Ycombinator is serious negative. But this comment was gold: ”He replaced his marketing department with a R&D department”

6. Robot reveal seriously scared me. Not the robot itself but the reminder that the singularity is coming closer. Tesla might not work with general AI, but their work into Dojo, transformers, search etc will push the technology needed for general AI forward.

7. It was nice to see and listen to the other two heads of tesla FSD team. Seemed some of them had multiple roles. I assume they are all very hardworking, intelligent and motivated. Tesla are trying to find more people like them, but the questions from the audience didn’t seem like it came from those kind of people. Maybe those were quiet and had interesting give and take dialogues at the booth rather than just asking questions they want answered as a lot of the audience seemed to do.
 
Last edited:
Ok my opinions. I understood like 99% of what they said.

1. I am a bit disappointed with the reports from the ground. It was a recruitment event and everyone is talking about it like it was a product reveal. I want to know who was in the audience, were they impressed, did people line up to hand in their resumes etc. Not about the product revealed. Anyway I was impressed. Imo to understand this event I think you need to read ”Genius Makers” to learn how companies work to attract AI talent and how talent are attracted to companies. The question about open source was an important one. I think Tesla will attract engineers who want to do cutting edge stuff, not academics who want to contribute to academia. Not like what Facebook, Microsoft et al had to do to catch up with Google. Karpathy is likely a very strong magnet for talent.

2. Tesla Vision stack is seriously cutting edge. Rate of progress is epic. I liked what they showed about how they handle updates of their internal map and how they select which features at which times to update when stationary. Have some questions about loop closures when going around a block I wish someone from the audience would have asked.

3. Their path planning search was awesome. Better than I had imagined. The problem with end2end is that currently we don’t have a good way to implement search in neural networks. So they use these feature layers in vector space and then do search in that. Also once we have a way of doing search by NN, not only with NN things will get strange, wonderful and scary. We already have a version of it here greatly expanding what questions a neural network can answer and imo we are getting close to general AI with search by neural networks.

4. HW4 introduced with CT. Not sure if it’s so great to buy a HW3 car atm, maybe a lot of benefit to wait for CT and then it should not take too many months for the rest of the vehicles to switch.

5. Hacker news at Ycombinator is serious negative. But this comment was gold: ”He replaced his marketing department with a R&D department”

6. Robot reveal seriously scared me. Not the robot itself but the reminder that the singularity is coming closer. Tesla might not work with general AI, but their work into Dojo, transformers, search etc will push the technology needed for general AI forward.

7. It was nice to see and listen to the other two heads of tesla FSD team. Seemed some of them had multiple roles. I assume they are all very hardworking, intelligent and motivated. Tesla are trying to find more people like them, but the questions from the audience didn’t seem like it came from those kind of people. Maybe those were quiet and had interesting give and take dialogues at the booth rather than just asking questions they want answered as a lot of the audience seemed to do.
Rob Mauer has a video up talking about the people he saw at the event and the "job fair" style booths set up after the event.
 
Wow! I just love this company.

Of course in the short term, the humanoid is a recruiting tool to attract the best engineers and developers.

Actually I was just listening to Lex Fridman on Joe Rogan and he said his involvement/interest in Boston Dynamics is decreasing.
Since Boston Dynamics were purchased by Hyundai they're starting to become more marketing oriented, trying to sell the product, getting some ROI instead of focusing on engineering and development.
This might actually be a good argument for some Boston Dynamics engineers to go to Tesla to work on Bot.
I've been reading similar stories on unhappy Blue Origin employees who are fed up and want to go to SpaceX.

The computer vision/ ai (brain) part of a robot is a much bigger challenge than the physical engineering, so in that respect I think Tesla will have a much bigger chance of actually getting this to market than Boston Dynamics with all the investments and experience from FSD/Dojo.

As much as some media outlets will try to laugh or ridicule the idea in the short term, our future is going to have some humanoid robots in it if you like it or not. So I can only applaude the Tesla team for once again being pioneers in realizing the future!
(or at least try :p)
 
D1 (400w TDP, 645mm^2, 7nm)
BF16: 362 Tflops
FP32: 22.6 Tflops
On Chip Bandwidth 10TBps(or 1250 Gb/s)
Off Chip bandwidth: 4TBps(or 500GB/s, 25 D1 per Tile)
Off Tile bandwidth is 36TB/s reported (I think 9TB/s is more like it for tile to tile communication), 3000 D1 chips connected together

AMD Radeon MI100 (300w TDP, 750mm^2, 7nm)
BF16: 92.3 Tflops
FP32: 23.1 Tflops
On Chip bandwidth : 1228.8 GB/s
Peak Infinity Fabric™ Link Bandwidth 92 GB/s (offchip)

Nvidia A100 (400w TDP, 826mm^2, 7nm)
BF16: 312 Tflops
FP32: 19.5 Tflop
On chip bandwidth: 2,039 Gb/s
Off Chip bandwidth: 600GB/s (up to 12GPUs)
Off Chip bandwidth PCI4: 64Gbs

My opinion, the actual D1 chip is pretty good, but not mind blowing given the size/power usage. It's inline with Nvidia's best. But dat scalability holy S balls with them interconnect bandwidth off tile. Mind blowing...
I would like to correct that the TBps from Tesla's presentation is actually Terabyte and not terabits, which make the bandwidth from on chip 5x faster, and off chip 7x faster than Nvidia.