1x Tsla Discussion Threads

[cfb2] 31/7/2024 13:47

Vauch: I've never heard that one! As far as I know Neuralink and Tesla only overlap in technology when the brain implant is able to pick up limb movement intentions from a missing limb and send them to an Optimus limb instead. I have heard Elon Musk say this as something way off into the future.

Current artificial limb technology detects intended movement by monitoring the nerve pulses in the remaining muscles at the point of amputation.

[cfb2] 31/7/2024 13:37

hosede: I won't disagree with you. The p/e ratio is far too high if you only consider their car manufacturing side of the business. It's even slightly high if you factor in their energy business.

As far as FSD and robotaxis go I'm just following the improvement trend of the software. Piper Sandler have just published a research note telling investors to buy due to robotaxis. They make the argument that nobody will want to buy a car in the future if it doesn't have the capability to self drive.

[vauch] 31/7/2024 13:01

I thought the plan for Tesla was to have the human neuralink so you can drive with the subconscious mind allowing you to get extra sleep before arriving at work.

[hosede] 31/7/2024 12:54

cfb
Well time will tell, but I think anyone buying at this price is paying an awful lot for a longshot

[cfb2] 30/7/2024 18:47

You're correct about there being more neural connections in the human brain but vast amounts of the human brain are goal orientated to things like getting your leg over, raising progeny and survival rather than driving a car.

Do you think Rainman's brain is wired better that way? If that were the case then Darwinism would have promoted autistic savantism over everyone else.

There is a chance at some point in our future generations the brain's goals may change but I expect the technology companies like Neuralink are developing will have redirected those goals.

[hosede] 30/7/2024 15:00

Cfb
I think the humam brain is thousands or even millions times smarter than any computer - and watching Rainman you realise that it's not wired up anything like as well as it could be.

[cfb2] 30/7/2024 13:00

If it can work for humans then it can work for an automated car. I've driven in English country lanes where you have to proceed cautiously and when you meet a car you have to reverse until you get to a wider part of the road. No reason an automated car couldn't do the same thing. More difficult are some Indian roads where you have 4 or 5 unmarked lanes and cars weaving between them!

I have to laugh at the WSJ video, with the mechanic discussing the complexity of the boards and the number of resistor components on there. I've worked on considerably more complex systems spread across multiple boards. The large number of resistors components are because they are limiting the current travelling around the board, which will be crucial for safety critical systems.

The hacking is just looking at the communication between the main integrated circuits. Data sent by the modem will mostly be in the clear and so they can see what communication is being sent between the car and Tesla. I don't think there are any trade secrets here but what is sent over this link will probably change between one version of the FSD software and the next. Given how easy it is to hack the boards to look at the data you can be sure that Tesla don't care if people look at it.

The secret sauce is what is going on inside the processors and neural networks inside the two big chips. If hackers managed to crack into this then that would be seriously impressive and I'm confident in saying they won't.

[hosede] 30/7/2024 10:36

cfb
Trouble is it HAS to be perfect - well FAR FAR better than any human. and it has to work EVERYWHERE! I can't see it working on English country lanes in Summer where you can't see round the corners for the grass
From WSJ

[cfb2] 30/7/2024 10:20

And I believe this because some ignoramus from the Telegraph says it?! Putting something in print doesn't make it fact, it just highlights the falling standards of education from those working in journalism.

[hosede] 30/7/2024 09:17

DRIVERLESS IS DEAD!! at least for the foreseeable future
It's the small EV tjhat is needed - but there will never be much money it - for anyone : just as with all other cars

[hosede] 29/7/2024 17:40

"Not even close to solving autonomy"

[blusteradjuster] 27/7/2024 15:50

Don’t let perfect be the enemy of good.

Would be good to get some holistic statistics (probably not from xyz youtuber) controlled for the kind of TSLA driver and the kind of driving they do.

If there’s a ‘never mind the millions, what about the thousands?!’ narrative then that narrative needs be firmly debunked.

[simon gordon] 27/7/2024 13:44

Yammie Noob - 19/7/24

Tesla's Autopilot Keeps Killing Motorcyclists. What do we do?

[johnwise] 25/7/2024 08:56

Tesla's Latest Earnings Report Leaves Investors Reeling


Tesla's recent earnings miss has sparked significant investor concerns. The company reported its fourth consecutive earnings miss, with adjusted earnings per share at $0.52, below the expected $0.61. Despite a 2% increase in annual revenue to $25.5 billion, automotive revenue dropped by 7%, and net income fell by 45% to $1.48 billion. Factors contributing to the miss include aggressive price cuts, increased spending on AI projects, and rising competition, particularly in China. The earnings miss led to an 8% drop in Tesla's stock. Tesla is heavily investing in AI infrastructure to advance autonomous driving and robotics, but these initiatives come with high costs. The future outlook is mixed, with significant uncertainty and cautious investor sentiment.

[johnwise] 24/7/2024 20:26

Expert sets next TSLA stock price target after Tesla ‘catastrophic earnings report’

Electric vehicle (EV) manufacturer Tesla (NASDAQ: TSLA) recorded disappointing results in the second quarter, missing key targets. Reflecting these results, the company’s stock has suffered significant losses, recording a double-digit decline

[hosede] 24/7/2024 17:09

In Britain they would need an extra one to make the tea :-0

[cfb2] 24/7/2024 16:47

Apologies all. That's some embarressing mental arithmetic! Thanks for picking it up zho!

Also the $28/hour was the rate prior to the renegotiation. CBS website doesn't say what GM renegotiated.

Yahoo quote the Ford UAW assembly line workers as getting $42.6 per hour in 2024.

The shift is 8 hours for Ford, which is $88,608. So a $40K Optimus doing the same work would be 45% the cost. If Optimus was doing two shifts then it's 22.5% the cost.

[zho] 24/7/2024 15:49

>>UAW workers earn around $28 per hour. Lets say they do 7 hours a day, 5 days a week. They are earning in excess of $200k/year.>>

Doesn't that work out at $50,960/year?

[cfb2] 24/7/2024 14:05

Industrial robots on the production line operate from a fixed location. This will often give them accessibility problems. No human is allowed within the radius of the robotic arm (they often operate within a cage).

A robot that is able to move and operate like a human is a major advantage and potentially can operate closer to the industrial robots. Reducing the length of a production line has multiple benefits: time, floor space, energy.

UAW workers earn around $28 per hour. Lets say they do 7 hours a day, 5 days a week. They are earning in excess of $200k/year.

If Optimus costs $40k and replaces a UAW worker then that costs 20% as much.

Optimus could potentially run 24 hours a day, 7 days a week. UAW workers may think they've made the deal of a lifetime but in the next couple of years they will get a big shock.

[hosede] 24/7/2024 13:35

cfb
But we already have vast numbers of robots on production lines. They are programmed in general to perform one very specific task.
Do you remember " Modern times" when Charlie Chaplin's sole job was to tighten one nut by about half a turn with a very large spanner, and when he came home he couldn't stop going thru the same motion :-o

[cfb2] 24/7/2024 13:01

Being able to fold towels would be considered a breakthrough in robotics. If it can tie its own shoelaces that'd be better than most politicians. Musk has said by the end of 2024 it will be able to thread a needle.

What we consider useful to us is a much higher bar than would be useful on the production line.

[hosede] 24/7/2024 12:10

cfb
When a robot can tie its own shoelaces, It could become useful.

[cfb2] 24/7/2024 11:50

No analyst is even thinking about Optimus for revenue predictions yet. Manufacturing of Robotaxis is at last forcing them to be considered though. My suggestion for "Optimus" is "Nihilus".

[cfb2] 24/7/2024 11:44

I agree zho, the quarterly call was a good performance by all the contributors but too "pie in the sky" for some investors and certainly analysts.

Version 12.5 of FSD software was released yesterday (was it a coincidence it happened on the day of the conference call?) but it only had one of the new features that Elon Musk promised (increased parameter count to the NN and HW4 support only). Things that weren't in the release:
Smart summon and banish
Support for Cybertruck
Combined highway and local driving models
Ability to reverse

Lack of required features being demonstrable is not a good sign for the upcoming robotaxi reveal. There will be a capex for the robotaxis and that needs to align with the software becoming plausible.