Machine
@M_IsForMachine
Followers
3K
Following
4K
Media
215
Statuses
3K
Joined March 2023
RT @M_IsForMachine: @freshjiva There's scant RCT's on protein effect on cancer and mortality. On kidney function, RCTs show no effect of….
0
1
0
Minor issue. Ready to scale for sure.
reddit.com
Explore this post and more from the SelfDrivingCars community
4
0
8
The misconstruing of the #Gaza war starts with fundamental misinterpretation of data. Death counts by popular news organizations like @nytimes @AlJazeera @BBC report report charts of unending death rises without acknowledging that - you know - people die in every country all of
0
0
2
In addition, it seems CAFE credits would be wiped out. Tesla earns about $1 billion from those per year. @ValueAnalyst1 @elonmusk what do you think?.
Estimating $TSLA Financials In Lieu of Tax Credit Removal:. Automotive:. Tesla automotive gross margins are currently ~ $5,000 - $6,000 worldwide. A third of sales are in the U.S. but not all of those will qualify for the $7500 Federal EV Tax Credit. J.D. Power, 87% of all U.S.
3
1
5
Estimating $TSLA Financials In Lieu of Tax Credit Removal:. Automotive:. Tesla automotive gross margins are currently ~ $5,000 - $6,000 worldwide. A third of sales are in the U.S. but not all of those will qualify for the $7500 Federal EV Tax Credit. J.D. Power, 87% of all U.S.
14
2
12
RT @M_IsForMachine: @elonmusk @SenMullin Just to be clear, you understand the difference between "deficit" and "debt" right?. If I reduce m….
0
5
0
How it started vs how it's going now:
$TSLA Sales to Plunge in Q4: Strong Sell.
3
1
27
maybe they can build local versions (like in Austin) that work well enough. We'll see. But it's clear we are hitting near the limit of what HW4.0 can do.
1
0
7
yes the trained model will always be bigger, but not necessarily the final tuned model that goes into product. it's the latter that is still clearly increasing with no signs of stopping. the next gen model - do you think it will be robust enough to support robotaxis everywhere?.
@M_IsForMachine Model size has to grow before you can distill smaller models that are better than the previous gen. By definition you cant have distillation without a bigger smarter model first lol.
2
1
9
Obviously keeping that 100x accuracy improvement is essentially a requirement for FSD, so right now I'm not seeing the ability to reduce model size by 90%. My "feeling" so far from what I've read through, is maybe you can reduce model size by 67%.
2
0
6
Model size keeps increasing faster than distillation improvements. If only someone could have predicted this.
@WholeMarsBlog @StuartZuckerma1 ~4.5X increase in params should be ready for wide release later this year. Super frugal use of memory bandwidth, caching exactly what is needed & squeezing microseconds out of everything are needed to maintain the frame rate. And the whole system needs to be retrained.
17
2
65
Congratulations to @elonmusk on successfully putting himself in this position wrt to people on the left and right
8
1
24
Really enjoying Lisbon, beautiful city, especially with me here.
5
0
6
Why do people extrapolate assumptions based on a small subset of data, when there is a much larger dataset available?. There are plenty of rideshare vehicles around already. You can calculate the sort of revenue you are going to earn at scale based off of what Uber pulls in.
EXCLUSIVE 🚨 Tesla’s RoboTaxi Math Just Broke the Internet
8
0
10
RT @M_IsForMachine: @AssalRad The entire propaganda around "genocide" started on October 7th, well before Israel even responded let alone r….
0
3
0
Yes there's so many ways government will twist the knife a bit. Say this fabled 10x in rideshares comes (and all robotaxis). The dead-heading may create many pain points of traffic in an urban area - there will be push back and restrictions in some cases.
@M_IsForMachine TAM is not what people are hoping for. Value is mostly in high cost parking areas. You also become a target for local government to raise revenue on. City of Chicago has large rideshare fees. @jpr007 and I have said this for years.
3
0
4
We get 40-60 billion in operating income per year. Value that with a PE ratio of 30 and you get 1.2 to 1.8 trillion in value. This is probably near best-case and *very* optimistic. That would value $TSLA share price at around. $400-$550. In 10 years. This is not a great.
8
0
2
Further, a more practical way of looking at it is assume a reasonable long term revenue and long term operating margins. Let's say I assume @Tesla grows business to 10x the revenue of Uber. 400 billion in revenue / year. Assume 10-15% operating margin (successful business).
2
0
1
In China we already know this to be true, and so price pressure will be downward and gross margins will not be very high. In the U.S. there is only Waymo right now, but in 10 years there will be competitors. And they will push prices down as well. There is no monopoly here.
1
0
1