He was part of the team that created internet payment systems out of thin air. They had an idea of how the world should be and successfully brought it to life; and it's still a successful company today.

The jury is still out on SpaceX profitability but history is littered with rich guys trying and failing to make rocket/aerospace companies. Elon has done it successfully and has pioneered new technologies in the process. Again, maybe landing/reuse isn't economically viable....but his company had a vision of the future and made it happen. Rockets landing on barges might not quite be science fiction but it's a step down the path to fully reusuable which is basically sci-fi.

You could say the same about EV car companies......For all it's faults Tesla has at least 1.) gotten further than any other stand alone EV car company....way further 2.) Made a car that, at least for a while, seemed like it was sort of from the future.

The Model S was a brilliant car imo. He made real a vision of a futuristic luxury EV.

And that's the theme imo.

Again....the culture he fosters seems wholly incompatible with a mass market auto manufacturer and I think he's more likely than not to be ousted from the company, TSLA goes bankrupt, whatever.

But GOOG hasn't shown us really anything since search. They turned straw into gold with a better search engine and a healthy dose of luck. Since then they've mostly just watched money pour out of the fire hose that is their Ad monopoly.

It would be an interesting study to take all of GOOG's retained earnings that they've tried to branch out into non-advertising businesses and see what the ROI is. I don't know what the answer is, but I do know that despite practically unlimited resources they are still (essentially) a one trick pony. I don't view them as a tech leader/pioneer. And the Gmail refresh a few months ago only reinforced that view in my mind.

Put another way....Elon's mostly been successful at raising money is a valid stance at this point and I don't even necessarily disagree with it. But the jury is still out imo. There's a chance that he does (mostly or somewhat or somewhere in between) live up to the hype.

We know who Google is; and it ain't a world changing tech company imo. At least not anymore.

I like the “I’m not posting anymore, it’s not because I’m wrong and losing a lot of money, it’s because you guys use words like cuck and it’s not nice”

I mean talk about a mic drop exit


See ya heltok, go back to Paying for that ValueAnalyst patreon subscription

Google can be the clear winner in AI (not just in driving too), I think cruise is a ways off. To me, cruise just seems like a ****ty waymo.

I agree that Tesla is probably toast if they don't hit their moonshot. I agree that Tesla's system as it currently stands (ie. what people are driving with today) is not very impressive [either compared to waymo or just in general]. Tesla however has much better vertical integration and the only ones with a very large set of live data (not simulated data which lacks many real world corner cases.) The goal is to get a system capable of regulatory approval, not to have the best demo. Against this goal I think Tesla is in good shape.

"Omg it folded a royal so its a bad ai" are weak args imo, you have to go by winrate not individual hands. Unless you want to say that Tesla is cheating on stats (this is a workable argument actually that I am willing to accept)

I don't agree with this. This type of technology isn't created by a big visionary going for a moonshot kill, but rather brick by brick with incremental improvements until a cohesive system is built, refined and tested to death that brings all improvements together.

Paradoxically, I think this is also why Google will not succeed because they have the Silicon Valley mindset and track record of taking great leaps forward and not caring about the refinements or maintenance aspect of improving technology (i.e., move fast and break things). I think full autonomy is one of those areas in which you actually don't want visionaries or alpha level programmers, but rather beta level programmers who understand systems and are willing to grind out the dull and dreary refinements and maintenance to minor details to build it up brick by brick, and test by test.

My guess is the eventual winner will be GM/Ford or the Japanese tech/automotive companies like Toyota/Denso/Hitachi et al. that have embraced the LEAN/six-sigma approach to building this system through individual ADAS functions first, rather then just going for full autonomy like Google is from the get go.

This is extremely wrong and if you are going to wager six figs on this you might want to study the state of the industry a bit more and realize how laughably off you are.

Google is a perfect example of how hard catching lightning in a bottle actually is. Its also a testament to the skill/luck(probably luck) that Jobs possessed to be the driver behind a long string of breakthrough products.

That's an interesting persective.....I'm trying to think of some analogues to human drivers -------> SDC to test your theory against but every thing that springs to mind doesn't seem very analogous. Something out of Xerox PARC or BellLabs comes to mind to test your theory but I don't know enough about either to hazard a guess; and I know even less about the culture of their employees.

Even accepting your premise don't you think it's far fetched that GM/Ford, Toyota/Hitachi, companies for whom software is, in the absolute best case, a tertiary competence, would advance the state of the art in what amounts to applied computer science?

The best analogy I've heard for this is... it's a mile rice and tesla has a 30 second head start... but the competition has bicycles.

PayPal was already a product when Musk's X.com, an online bank, merged with Confinity, the developer of PayPal. Musk was the CEO of the combined company for just 6 months before being replaced. Musk had little to do with the initial development of PayPal or its eventual success.

This is completely absurd. Google is one of the most successful and transformative companies in the history of mankind. Google also has been able create multiple dominant products, each one far more dominant than anything Musk has ever managed. Outside of search, the following are all category leaders:

Gmail
Google Maps
Android
YouTube
Chrome
Google Ad Manager

This is before you get to either non-products or not yet dominant but still impressive products, such as:

G-Suite/Google Drive
Google Cloud Platform
TensorFlow
TPU
Google Analytics
reCAPTCHA
Google Calendar
Google Photos
Nest
ChromeOS
Google Assistant

Compared to Google, this is also true of Tesla.

Google has lots of different successful products - most of them happen to generate ad revenue. GMail has ads. Google Maps has ads. Youtube has ads. Android/Chrome/ChromeOS integrate with Google accounts to help Google deliver other ad-supported services and also improve quality of ad targeting. Google Ad Manager is an enterprise product for publishers and advertisers to buy/sell advertising outside of Google services.

It's because they are making money faster than they can spend.

Amazon is far less profitable than Google, so they don't have this particular problem. Amazon is incredibly frugal, literally about everything, and that's how they manage to be barely profitable. It's not because Amazon decides to invest more - the exact opposite is true. All else equal, high margin is better.

This is why I am leaving. I like to discuss arguments, some people here like to discuss posters and they have scared away lots of posters who like to discuss arguments.

Thanks to candybar for pushing back on newsavman's mistaken notion that Google isn't really a tech company and that their advantage over Tesla isn't that great. Apart from the dozen-plus Alphabet Inc. endeavors (YouTube, GoogleMaps, TensorFlow etc.) that candybar pointed out, they also put forth one of the most amazing deep-learning AI demonstrations in human history with AlphaZero conquering the entirety of human and computer knowledge of chess (and go and shogi) after four hours playing against itself. So I'd say they have pretty legit tech chops, and aren't just some lucky advertising company on par with Tesla as far as hardware and software developers go.

Alex, I didn't understand the time horizon of your Tesla bet. Are you thinking Tesla could weasel approval in a year or two and be first to market, by virtue of Musk's government manipulation skill; then you'll bail since you think they'll be overtaken by Waymo et al.? Or is your bet even longer term because you think their deep-learning, camera-based approach will actually win out and they'll be world leaders in self-driving for many years to come?

First I will re-evaluate over time, I may bail at any time including just a few months etc.

I believe a camera-based approach can be successful. I mean that's what people do. Also Tesla doesn't just have one front facing camera, they have a bunch of cameras and other sensor data.

I tend to think about advantages, and Tesla has natural advantages of having some ramp up of production, and the data from the fleet. This is not a light point, I think it will end up being that corner case data from real use cases (filtered over billions of miles) is at a very high premium and necessary to add 9s to safety (99% vs 99.9% vs...) and there is no way out (besides some breakthrough like totally unsupervised methods.) However, if a breakthrough happens, tesla will also be poised to take advantage as it seems the advantage of lidar is more about helping label and not necessarily as a first class input modality.

All the other advantages of the other guys can basically be bought. For example every player with deep pockets is basically already going after the best AI engineers and offering them 7 or 8 fig etc. to work there. I looked up the top AI people in Tesla after the newest shakeup and everyone seems good and capable to deliver and I believe them. Academic types basically don't do this, they are basically honest about technical aspects.

Elon can lie or exaggerate all he wants, he's a CEO that's almost par for the course. Seems actually like a necessary trait of a CEO, if you think about it. I think a lot of inside people see "he's saying something off" and just automatically jump to the conclusion "lying -> it's ****". But actually another motivation to push is to keep the price from falling through the floor. Though it is rare, technically speaking you can still be right.

What I don't understand is why you are buying in now?
You agree it's very low probability. Why not wait until it's clear TSLA will be able to develop a commercially available AI first? Sure, you may lose out on some multiples, but you avoid the high probability that they dilute multiple times or potentially eventually file for BK.

If they actually get fully automated driving, the stock price will be in the many thousands. You'll do just fine if you wait for more clarity, and may get a SP 75% lower than what it is now.

Is the data even that much of an advantage? If some of the majors decided to change approach and copy Tesla, it wouldn't take long for them to surpass them in driven miles?

The collected data thing is pure Musk bull****. It is useless.

Camera recognition is so hard that even robotics can't do it for defined paths. This is the main reason you still need humans in car production lines - a fact which Musk was so incredibly, incredibly stupid as to not understand this basic concept - hence tried to robotize his whole M3 line and nearly lost his company as a result when he had to scramble to backtrack, ending up with a 1980s hand line in a tent. You think this utter ****** can solve the orders-of-magnitude more difficult problem of FSD via vision machine learning? What are you smoking, I want to invest in that pot company.

I'm not sure what this has to do with anything. Their sensors and software are so inadequate they fail to detect trucks alongside on a highway, and stationary objects in front. Yet they're going to solve fully self driving in 8 months?

Musk is the God of plausible sounding bull****, and he's done a number on you here. Fleet data is worthless. For one, it's imprecise and requires human tagging. For two, it doesn't given decision feedback. Training a model on "**** happened on this feed, let's somehow magicaly not do that ****" is absolutely worthless for a machine learning decision engine. Google in comparison have billions of miles and actual precise simulation training/testing going through vast decision trees with all possible consequences, and millions of miles of real world usage. They have Tesla crushed on incoming data.

One thing that LIDAR allows you to do is to simulate, since the incoming data is a precise 3D map and you can simulate the exact same incoming - enabling you to train billions of miles in days if needed. You can't do this with vision as it's not a precise 3D point and motions map - this tiny details of color and shading matter - and it's something that requires precise contextual analysis and correction with real data, something orders of magnitude harder to machine learn and possibly requiring close to a human level AI, which we're perhaps a decade off. The lack of simulation ability alone means Tesla can't get there.
The advantage of LIDAR is reliable 3D mapping and precise high resolution no-failure positioning of all objects and their relative motion. Vision cameras have to infer that via very complex (AI level) processing instead of having hard 100% reliable data coming in in the stream. This is the core of the problem. Until you have Skynet getting these camera feeds, you have nothing, especially for the thousands of edge cases you can't machine learn where a bad recognition means disaster.

Bought with what? Tesla are losing vast sums of money and have no net cash even after the raise. You think they can outbid Google? Tesla were down to hiring some dickhead Stanford academic with little real world experience to head their project, and Snapchat's head of monetization (no joke - look it up!) as the VP of engineering!!! You think this team is going to delivery autonomous driving over Google, or what Cruise has achieved in a short time?
I suggest you listen to the 3 hour autonomous driving investor day video. That should turn you completely cold on Tesla. They are lost and clueless and praying that machine learning will solve problems, which it won't for a decade. The commentary on snow and solving complex driving decisions (we'll just machine learn it!) is particularly hilarious.

Starts at 1hr 9 minutes



This is a bigger bull**** story than Solar Roofs.
Liar and fraud isn't liar and fraud because he's trying to keep the stock price up?

the fleet is not an asset.

if they were serious about their autonomy plans (they are not), they would spin off this giant debt- and law-suit-ridden, under-warrantied, money burning pile that is the car business (and all the other stuff besides self-driving) to the dumbest retail shareholders in history and start off new with their ai-team and a few billion $ from investors as a software company.

just yesterday they paid $11m to settle a 5 year old work accident case. there's a lot more where that came from.
the car business is a liability.

There are a couple points the "bulls" keep making that are just so wrong it hurts.

1. "Data from the fleet" - this is just absurd. People think each tesla is streaming back real-time status and updates about driving conditions at all times back to Tesla HQ. This would take Terabytes of data per car and a guy hacked his Tesla to see what data is being sent back said it is a tiny amount of data, nothing that could "recreate a scene". People think "oh they can query the fleet data to find a situation where a car is surrounded by trucks and a dog crosses the road in a construction zone" - Uhhh, no. Just no.

2. "Tesla is ahead" - They have 0 autonomous miles driven in California. ZERO. What they have is a fancy adaptive cruise control + lane control that many other cars also have (but they aren't stupid enough to call it Full Self Driving)

3. "Feature complete this year" - They mean actual level 5 autonomy, which is absurd. Level 5 is ALL situations and ALL road conditions. Someone on the "Autonomy day" even let it slip that they haven't even STARTED on snow conditions. Snow is a really tough problem for a camera-based system, but ok they will nail it in months. Righto.

There are more but those jump out at me...

It blows the mind. Alex Wice is a decent thinker who often gets the right answer in uncertain situations and the conman Musk has managed to get him to put >$100K in his stock on a pure fraud that has zero chance of happening after a long history of lies, fraud and failing to deliver, and an autonomous investor day that basically admits they are nowhere with no path to level 3 in two years let alone 4 or 5.

China sales not good

Let's just fast forward to the money shot:

Tesla is dying a slow death and the industry has noticed. As one of my contacts in the auto supply community noted to me, "the hamster is dead but the wheel is still spinning." Tesla shares have limitless downside here. Be very careful.

We have been through your posts and we have the fact that the thread is now six years old with the name "TSLA showing cracks?". The bears have been terribly incorrect on this stock and have only been right this year - the easiest year to be right in. Even bulls like myself have predicted 2019 to be negative for TSLA.

I love this nonsense. GM and Ford will be nowhere. They are already nowhere and have a terrible track record of innovation.

Toyota/Denso/Hitatchi are just Japanese companies that you think do something other than mass produce stuff, they are not exactly the big innovators either. What are you basing your nonsense on?

The LEAN/Six-Sigma approach doesn't help you in innovation.

It is no laughably off. The LIDAR based systems today crash very frequently. Only Google is even in the position to show some track record. Cruise is just a producer of car crashes basically.

The big advantage that TSLA has if the approach they take ends up working, they will be on the street having more than a million cars. If they fail, they are basically as far as everyone else and have to buy technology from Google. The payoffs are vastly in their favor - it's just a long shot at the moment.

"Study the state of the industry" - what does that even mean? Go to university if you want to study things and see how wrong you are (or just run bull**** regressions that prove your point, but can't be replicated). If TSLA was so far off, they wouldn't be able to attract the talent they are. It's a myth that no one capable works at TSLA, it's actually the complete opposite.