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

[willow_hiller]

Not discounting for not present value, the subscription price is 120 months of the purchase price of FSD.

You're being a bit pedantic here. Perhaps the purchase price will come down or perhaps this is the time when Tesla tries to convert everyone to a recurring payment (monthly) rather than a one time purchase.

The key difference is that a subscription price is subject to change. Subscribing now does not ensure that the price will continue to be $99 per month.

If you assume the subscription price of FSD will never increase, then you're correct that you need to keep your car for 120 months before you break even from a FSD purchase. But the subscription price will not remain $99 per month as FSD progresses.

 

[Big Time]

Wow, that's quite the tweet, I suspect a $99 FSD subscription will get a lot more folks and families on the bandwagon. $199 a month I could feel queasy...... $99 that's like a bag of groceries in these times..

On the P&D side where do we think we're heading into Q2? Has Berlin recovered entirely? Those pesky huthis still flubbing up the Red Sea? If anyone wants to lay it down with some nastrodamus style predictions im all ears.

Closed on my home 2 days ago!! What a sigh of relief. I was chatting with my mother about how when I sold the bundle of stock and got all my cash in line for this, a LOT of my stress has faded away. Prior to doing that, for the last 8 years I have opened my accounts every morning, first thing upon waking up. Looking back makes me feel borderline disgusting but on a personal note it's VERY strange I don't even consider reaching for the phone at this point.... Im up and looking for the coffee and perhaps animal/bird watching or just busy being alive I suppose.

On the other side of the coin I feel for all the people who are looking at the phone. More power to them, but I know the struggle! I think of posters here on the board who have always advised a smell the roses approach to investing. Both are probably fine and healthy ways to live, both with positives and negatives =). To each their own!

Great news with the FSD!

Sincerely
-Biff

 

[unk45]

I concur, and firmly believe that the only purpose for Income Tax is for use as a vacuum cleaner to suck vast quantities of inflated currency out of the system in order to offset the negative effect of the money creation side of this system.

The Federal Reserve and the Internal Revenue Service are private entities, both of which were created in 1913. They were designed to work in concert with one another to manage the money supply in the hope that they can keep the balls being juggled in the air. No fiat currency has ever stood the test of time, but some have lasted a few centuries.

Hopefully, the abundance on the horizon from drastically reduced energy production will bring a positive effect toward making a flawed monetary system last a little longer than it might otherwise have.

Tesla specifically, and Elon generally, are riding the wave of this convergence of technologies that with any luck will lead to change that will save our bacon.

I had to disagree n two points: first, the Federal Reserve System and Board are 100% US Federal Entities. They are NOT private. The regional Federal Reserve Banks are technically owned by the member banks in their District although they operate in policy set by the Federal Reserve Board. Here are the facts:

About the Fed

The Federal Reserve Board of Governors in Washington DC.

www.federalreserve.gov

The next one would by hysterically funny were it not to be a totally false statement.
The 16th Amendment to the US Constitution gave the authority. Reading that might help. Just a glance of the IRSntimeline does the rest:

IRS History Timeline | Internal Revenue Service

A brief, interactive timeline depicting a few highlights on the evolution of our American tax system. As you browse through this timeline, you will see the dedication of the Internal Revenue Service’s employees and our ongoing commitment to service and to all taxpayers.

www.irs.gov

To describe those false statements as concurring mine was sad. My post was a satirical comment on the evolution of papar currency. I may have actually misled you when I referred to leaving the gold and silver standards. Those events have exactly zero to do with actual tax policy. Finally the FRS and IRS were indeed legislated the same year. Checking their legislative history shows they were both intended to help prevent financial crises and provide stable funding base for thenUS Government.

All of this belongs in a graduate monetary policy course, nit a general investors forum.
However, without understanding thenUS financial system, including also the SEC and the DTC are less well equipped to understand the practical financial risks of holding TSLA.

Understanding all that allows a far less transactional view of TSLA.

 

[Artful Dodger]

Quick thought, this will further erode the value prop of the Foundation Series CT, going to see some unhappy people there that already didnt like they were "paying" more than the locked in $7k FDS price for early res holders.

Tesla will likely handle this easily. For example, tiered levels of FSD licenses could provide 'transferability' for grandfathered customers. That way, people who paid more in the past get more value in the future. Easy peasy.

 

[Stormy24]

https://twitter.com/x/status/1778911525865914441

 

[SOULPEDL]

Entire stock market at risk if the invasion leads to a wider war (more Iran proxies joining, Iran impacted, etc.)

I did a search on Iran, this is all I could find, so I assume it hasn’t been discussed recently.

Attack on Israel is likely this weekend. Gettin a bit concerned. Anyone have a read on how it impacts TSLA or markets?

I see potential for the Israeli strike and a measure counter back at Iran. If that’s it, will only be a blip next week. Hope for zero escalation after that point. I assume it puts oil prices on notice, but just guessing. Thoughts?

 

[Drumheller]

Regarding Sawstop: the tech to detect the current change when skin makes contact with the blade and immediately trigger a hard stop before the injury is more than superficial costs a few hundred dollars. Some other manufacturers have value offerings that are less than $500 in total. If they implement the safety tech, their value line will cease to be offered for under $500. If they choose to implement the safety tech in their higher end models but leave the low end model without it, then they will seen as not caring about poor people.

So the other manufacturers decided to forego the tech altogether and compete against SawStop by offering lower prices. However, some people do comparison shop, value safety, and have the funds available to spend more, so SawStop has done well.

How this applies to FSD, Tesla, and competitors, I'm not certain, but I am curious to read other's thoughts regarding that scenario as applied to Tesla and FSD.

 

[Artful Dodger]

But until there are true cababilities suitable for business use,

I see what you did there

 

[Joe F]

Good question. I engaged FSD right before stopping at the circle entrance.

Okay, but your first post stated "on my next approach into the circle" implying you were already under control of FSD and moving. Perhaps I read it wrong...

It's not clear to me that this is a valid use case for FSD, initiating routing to a location that is what, 100 feet away? Perhaps a valid test would be routing to the BP from further back, say from the Pancake House or Davis Bakery? You know, give it a chance to make a left turn into the first entrance to the BP?

Not sure why you would label this as "bad, really bad."

Junk yard dog here, btw, but my last post on this issue.

 

[unk45]

Yes, and does anyone remember the price of gold when we nixed the standard? Answer: $35 per ounce!!!

Today gold touched over $2400 per ounce! What’s changed over the last 50 years? An ounce of gold is still an ounce of gold, and technically it has no more value today than it did back then.

The value of the dollar is what’s changed! The $35 dollars that’s been in my Dad’s safe since the early 1970s is still worth $35, and the ounce of gold next to it is still an ounce of gold, but it’s worth a lot more than the $35 he paid for it.

THIS is inflation, and despite what our politicians tell us, THEY control it! Unfortunately, it hurts those living paycheck to paycheck the most. People who are able to buy gold or real estate or S&P 500 Index funds will do okay, because over time those assets grow in value relative to the dollar.

FWIW, keeping those standards also delivers economic fluctuations with any control. Those are the consequences if making something mined and used industrially as a standard for exchange value. The history if things used to exchange value show that none ever are perfectly suitable. Perhaps I’ve studies this subject far too much…

Amazon.com

 

[Thekiwi]

Model Y LR RWD (600 km / 373 mile WLTP range) now available to order in Europe. Not yet available in the UK, due to driving on the wrong side of the road.

https://twitter.com/x/status/1778353650668105923

This is a great demand lever to pull. The LR RWD Y is 10% cheaper than the LR AWD, and gets 12.5% more range.

I know many here are all about the G’s when it comes to their preferred Tesla variants, but for me this is the perfect option, give me more range over performance any day. The fact that it is cheaper is icing on the cake.

It is of course cheaper for Tesla to make this model than the AWD, given the lack of the 2nd motor. So even though margin is a bit smaller, it’s still substantially better margins than the entry level SR Y.

I hope Tesla rolls this variant out globally.

 

[unk45]

Has anybody other than me had recent updates (I e since 12x) being made via Tesla cellular?
My P85D had those but the ones since then have only works via WiFi. This makes me very, very happy. Especially, I was in Brazil when I just made another update on my Miami resident Model Y.

 

[DarkandStormy]

https://twitter.com/x/status/1479595757388845059

All the mind blowing products coming soon get their price cut in half, right?

 

[mrmage]

It's not "fast compared to training". It's a totally different application process. Training requires petabytes of information to be processed to build the algorithm. Inference compute can fit on a smart watch for edge computing.

Currently Tesla is in the process of removing bottlenecks on the inference side by having the algorithm using more of the npu. The npu was done with the answer a long time ago and it was the cpu having all the control C code that took forever to utilize the npu answer and would offen output the wrong thing.

Compute has plenty of headroom because prior to v12, v11 was bogged down by a cpu core capable of only 16k flops per core. Assuming 4 cores were used, we are talking 64k flops as the rate limiting step vs the npu which is 72 trillion flops. The question is all about memory size.

Training creates the model parameters, which are then used for inference to predict the best behavior. The amount compute required for inference is closely related to model size. A small LLM of 10 to 20 GB parameters generates only a few tokens with a fast CPU, and perhaps dozens of tokens on a high end GPU. None of these models nor FSD will run on a smart watch in real time, nor in non-real time due to memory constraints.

This is one of the most important posts on this forum given the high importance FSD robotaxis has suddenly become in supporting any future apprecation of share price, so it's surprising there's not much discussion.

You have given some good evidence for why it's quite likely the models are going to grow much bigger to support lower error rates / lower disengagements. There will probably be advances in ability to distill these big models into something smaller that works mostly as well but still... this a serious issue with inference compute. If Tesla eventually trains a model with high enough fidelity for robotaxi, it may be orders of magnitude too big to fit on HW3, HW4, or who knows, HW5.

This is a very serious risk.

There are other risks wrt to training: How hard will it really be to improve performance 100x?

There are papers and studies around "neural scaling laws". They look at how much increase in compute, data, and model size affects the abiliy to lower error rates.

Here's one on vision transformers (on images) which could relate reasonably well with FSD.

https://openaccess.thecvf.com/content/CVPR2022/papers/Zhai_Scaling_Vision_Transformers_CVPR_2022_paper.pdf

Here's the thing, even on a log scale, as you increase the compute, data, and model size, you get sublinear returns in error rate improvement.

I.e. it becomes harder and harder to reduce the error rate further. The model size, data size, and compute needed to get crit disengagements (CD) to 3000 miles / CD may be a 10x increase, but then to get from 3000/CD to 30000/CD may take more than another 10x increase, it might be another 30x increase in data / compute / model size.

Tesla has the ability to scale up data and compute some, but not these orders of magnitudes in a "short" time frame of a few years (let alone by August).

Point being, there's a lot of empirical data hinting that training FSD to a robotaxi level could take a lot longer than people think.

View attachment 1038036


View attachment 1038037


View attachment 1038040

The first graph looks pretty discouraging after eyeballing the error rate and replotting along a log scale for the X and Y axis. If the newly released FSD 12 is represents the leftmost data point, extrapolating to 10^4 fewer interventions would require going from -0.3 to -4.3 error rate. From the slope of the line, 10 training periods are required for 10x improvement in error rate, implying 10^40 training units are needed.

The sun will extinguish in 10^10 years, so Tesla will need to do much better.

Presumably Tesla will optimize the training with state of the art techniques, but I posit that the behavior for FSD actions is more important than image classification. Nonetheless, adding having one extra 0 increases training from 1 year to 10 years. There are 40 zeros, so a *lot* of zeros need to be eliminated.

 

[Thekiwi]

https://twitter.com/x/status/1479595757388845059

All the mind blowing products coming soon get their price cut in half, right?

I’m sure it will be nice to have a big FSD subscriber number for the upcoming earnings call, to offset the tepid earnings.

However arguably it is all about the data for training at the moment, and Tesla should perhaps consider offering for free for a limited time (say 6 months) but blowback from paid upfront FSD users may be a little high if they did that. Maybe worth offering for free in areas with low amount of users and where training data is lacking.

 

[Mengy]

https://twitter.com/x/status/1479595757388845059

All the mind blowing products coming soon get their price cut in half, right?

Do you understand the concept of Supply and Demand economics? If not you might want to read up on it. Lower prices greatly expands market adoption, and usually enables making even more total income than could previously be possible at the higher pricing. This is particularly powerful for software with high margins, which FSD is. Plus the larger FSD customer base brings in a LOT more driving data, and now that Tesla is no longer compute constrained this will accelerate FSD training.

Lowering the price of FSD is a very good thing for Tesla.

 

[jschwefel]

FWIW, keeping those standards also delivers economic fluctuations with any control. Those are the consequences if making something mined and used industrially as a standard for exchange value. The history if things used to exchange value show that none ever are perfectly suitable. Perhaps I’ve studies this subject far too much…

Amazon.com

Agreed. Clearly the gold standard wasn’t without its own problems, the Great Depression is evidence of that. I was simply trying to explain a reality of today’s monetary policies. A world full of fiat currencies is really a race to the bottom and it’s fascinating to watch. The dollar’s status of the reserve currency is also something to consider.

 

[Dutchie]

https://twitter.com/x/status/1778911525865914441

There goes the golden goose…

 

[SOULPEDL]

I did a search on Iran, this is all I could find, so I assume it hasn’t been discussed recently.

Attack on Israel is likely this weekend. Gettin a bit concerned. Anyone have a read on how it impacts TSLA or markets?

I see potential for the Israeli strike and a measure counter back at Iran. If that’s it, will only be a blip next week. Hope for zero escalation after that point. I assume it puts oil prices on notice, but just guessing. Thoughts?

Starting to wonder if this is OT… maybe that’s a sign… I guess it could already be baked into the market.

Anyway, from what I read, Israeli LR defenses are not necessarily iron clad… meaning we don't know how effective Iran can pierce through, and the more missiles results in higher risk.

The iron dome is only good for short range, however Israel has about 3 or 4 defense systems for long range (after watching a recent Bloomberg summary).

I would assume the target would be political, but who knows. US has a lot of industry such as a Chip FAB over there (not related to Tesla). They wouldn't strike industrial would they, or Jab at the US?

Hope for the best folks. Maybe they don’t strike at all.

 

[geometro]

So odd… neither my 2018 M3 or wife’s 22 MX have gotten the FSD trial.