1/1/2015
guest I just want to point out that there has been some backtracking on the claims of autopilot learning in the thread in the Model S forum. I think it will happen, but the early reports appear to have been a little optimistic.�
1/1/2015
guest I can't find any such news? Could you point to the thread please?�
1/1/2015
guest Sure, the discussion is ongoing in this thread. Link points to a post that gives a description of what seems to be the current line of thought, from what I have seen from skimming the thread.
Autopilot is already improving. - Page 26�
1/1/2015
guest If Tesla isn't the cutting edge of the software, they are pretty close. Mobileeye is the hardware leader.�
1/1/2015
guest the algorithms that translate pictures into vision, artificial intelligence, are those of mobileye.
fantastically complex software, tesla adopts the vision to the car mechanisms and screen�
1/1/2015
guest True, there are different levels of software. For example with a desktop computer you have devices connected to the computer that may have their own processor running their own code, then you have device drivers that interface the operating system to the device, then you have the operating system itself (which in turn has many layers of its own), and finally you have user applications. In this analogy MobileEye has code running on the device, and they supply the device drivers. Tesla is among the leaders in the operating system and user application levels of software. Which is not the same feat as the basic algorithms MobileEye came up with, but it's still notable.�
1/1/2015
guest There are no entities in the world that have more (varied/quality) sensor data on automated driving than Tesla. All Autopilot enabled cars have the capability to gather sensor data in the driving conditions and environments that occur whereever such a car is driving. I don't know if Tesla is currently gathering sensor data from all the Autopilot enabled cars and shipping it back to headquarters, but it is an obvious idea that they will guaranteed leverage to get a huge advantage over all other autopilot/autonomous driving companies.
Google currently has the world's biggest database on automated driving and they also have a truly humongous amount of data on terrain, road conditions and maps from the their Google Maps/Street View initiatives. But Tesla is currently in the unique position of having a global fleet of sensors for gathering data on local details of driving, with the capability of automatically shipping this data home for processing. So to the extent that Google currently has an advantage in this area, it won't last. Google is frighteningly capable at artificial intelligence and software engineering, but they don't have the sensor network that Tesla has. I don't think that Google should be viewed as a dangerous competitor here.
I don't think the investor focus should be on automated learning as much as semi-automated tweaking or localization of their autopilot software. We know from the big data/machine learning fields that you can avoid a large amount of AI headache simply by having more data for training and tweaking your algorithms. I think people are discounting the advantage that Tesla has here. I wouldn't be surprised if Tesla will be the undisputed global leader in automated driving once it hits the mainstream, and that it will be very difficult for the second runner-up to catch up.
You simply can't have a better situation to gather detailed data on road conditions, weather conditions, traction, subtleties of road markings and signs etc. than by having your sensors constantly drive around. Long-term, I think Tesla will leverage this to e.g. predict road conditions in real time, but the far bigger advantage is simply knowing exactly where all the potholes are, where the maps are just slightly inaccurate, where drivers choose a different lane because there is always a big puddle when it's raining, having the ability to correct poor road markings by examining driver behavior, adding exceptions to software behavior in areas where dangerous situations frequently occur, etc. The advantage of this capability simply can't be underestimated. There is no way that any specialized AI can predict these subtleties better than actually having the data and going through it in a semi-automated fashion. Tesla will also have an unprecedented ability to reverse-engineer accidents and faults in their driving software, and they will also have the ability to build realistic simulation software from real-world data where their software can be automatically tested at 1000 times the driving volume than what happens in the real world. Have you heard about the simulations where SpaceX run their space rocket software against a simulated environment, having the spacecraft computer live in a simulated environment that to the computer is indistinguishable from the real world? Tesla is almost certainly building the same capability, with an unprecedented ability to detect and correct problems with their software.
I am super excited about this. This seems completely obvious to me as an average software engineer, so you can bet pounds to pennies that Tesla's soon-to-be world class team of machine learning and autonomous driving experts will be thinking the same thing.�
1/1/2015
guest Great post, Marvin. This type of information is undoubtedly valuable.�
1/1/2015
guest Not sure I agree with a lot of what you said. You're analysis of Google's approach to automated driving is probably accurate. It's a big-data approach where continuous communication with both central and local data sources is the foundation. I think you've underestimated the value of this data. By comparison Tesla is sensor and ai focused. And I don't see Tesla making a huge effort to change that. I believe it would be much harder than you think to collect data on a scale that Google has for autonomous driving regardless of how many sensors you have in the field.
I'm not saying Tesla will ignore the potential to collect road, traffic, and other driving data from their cars. I just don't think they will try to beat Google at it. And I don't think Tesla's fleet is a major advantage. It's more of a different approach.�
1/1/2015
guest I know this will sound like 'a friend of a friend told me'...but
A friend met a former Tesla engineer last week. He said to the engineer, 'I want to know when I can get in my S, push Navigate to a place 20 miles away, and have the car drive me there without my assistance.' The engineer said '6-8 months away'. Now I know that it's Tesla Time, but still that's much closer than I had thought. More cameras on the vehicle will be necessary.�
1/1/2015
guest It sounds exactly like that. [emoji3]�
1/1/2015
guest Yea, take it for what it's worth, which isn't much. Probably more likely than if I heard it from a Delivery Specialist (who have started more than a few incorrect rumors in the past), but I sure wouldn't take it as fact. The engineer's Model S did have more cameras on it than the ones coming off the line, though. Just sayin'�
1/1/2015
guest He may be a "former engineer " for a reason [emoji12]�
1/1/2015
guest Interesting. Aside from "Tesla time" I would not expect these capabilities to go public near that 6-8 month time frame simply due to regulations. As Musk has mentioned before, Tesla's autonomous technology will advance faster than federal regulations permit, especially in the US. He really thinks complete autonomous Teslas will be available in 4-5 years but won't be publicly used for another 3 years because of these regulations. Very hard to believe that anybody will be going 20 miles without any assistance within 8 months.
However, it will be extremely cool to see how this process develops and reaches mainstream attention. The last autopilot software update was a huge milestone when it comes to self-driving cars and each major update to follow will only shed more light on how fast this technology is coming and how far ahead Tesla is. Autonomous driving technology is one of Tesla's major advantages over other auto-makers and still has heavy skepticism, just like other aspects of the business.�
1/1/2015
guest There sure are a few factors holding TSLA down right now.
1) Oil approaching 9-year lows
2) Strength of the U.S. Dollar hurting gross margins
3) Model X delays
4) Significant cash burn.
Hopefully #'s 3 and 4 get solved in Q1 of 2016.
I can see TSLA hitting a market cap of around $40bil in 2016, however there will be significant resistance when it starts to approach the market caps of OEMs such as Ford and GM, both at around $56bil.�
1/1/2015
guest
Cash burn is not a long-term problem, it's called investment. It will be happening to some extent as long as EM is in charge, or until TSLA becomes very large.�
1/1/2015
guest Overall EV sales in the US are down for 2015 and worldwide growth has been slow, however the Model S is selling better this year than last year (though by small margins). When oil prices took a nosedive, sales for the Leaf fell sharply (3102 sold in the US in Dec 2014, but only 1100 in Jan 2015, fewer than Jan 2014), but Model S prices stayed strong. Tesla prices seem to be more immune to fluctuations in oil prices than other EVs. Most people who can afford a Model S could also afford a gas guzzling SUV if they wanted to. They are either buying for eco reasons outside of the short term price of oil, or they are buying for the tech or performance (or all of the above). While lower gas prices may make justification for those who use spreadsheets to justify every major purchase, it overall doesn't seem to hurt sales much.
The strength of the US dollar is hurting gross margins on exports, but it may help get some sales in the US against imports which now cost more.
Yes, the Model X delays has hurt them, but hopefully that will be turning around soon. I am concerned about the reliability of the falcon wing doors. I expect a lot of customer complaints about those. And the cash burn as MitchJi pointed out is investment and not fixed costs. GM went bankrupt because it had huge fixed costs it couldn't get rid of any other way. There was no fat to cut. If Tesla chose to, it could turn a profit in a very short time by slashing R&D, stopping supercharger development, and halting the Gigafactory. All those things are optional expenses, though anyone with any concern about the future of the company knows they all need to move forward.
The cash burn should get better with the Model X going into production and becoming an income generator instead of a cash suck, but there likely will be some cash burn until the Model 3 is in full production.�
1/1/2015
guest Worldwide plug-in sales through October of this year are up 50% through the same period last year. Not sure I'd call that "slow," : Monthly Plug-In Sales Scorecard
And I wouldn't be so quick to attribute the decline in Leaf sales to oil prices. Nissan definitely cannibalized their sales to some extent by announcing the long-range next-gen Leaf so early.�
1/1/2015
guest It looks like worldwide sales this year will only be about 35% higher than last year. If US sales in December are the same as last year, the US will only sell about 2000 more EVs than last year. A pretty small growth for a market niche that has been growing at large rates the last few years. Though you're probably right that Volt and Leaf sales may be off this year because new models are coming out this fall. Volt sales look stronger the last two months, but Leaf sales have remained flat.
We'll see if oil prices hurt the low end of the EV market or not. At the prices that cars like the Volt and Leaf sell, the price of gas may be a consideration for some buyers. Those cars are competing with the best gas mileage cars on the market in size and features and they generally have less cargo space, which may be a consideration when someone is weighing the cost of gas vs carrying capability.�
1/1/2015
guest Although it is apparent that demand for Model S clearly isn't tied to the price of oil, many investors aren't so quick to dismiss Model 3 demand concerns when oil is so low. Although I don't believe it will affect demand of Model 3, Mr. Market clearly doesn't care what I think, as TSLA has fallen in lock-step with oil since July.
The point of my previous post was to be introspective of my investment in TSLA. The conclusion that I came to was that those 4 reasons have been the primary drivers of the stock price in the last 5 months (early July the stock was at 280). 2 of the 4 reasons will most likely be taken care of in the next 3-5 months (hopefully).
And sure, TSLA is using significant cash to fund growth and invest in infrastructure, not operations...that is stating the obvious. But TSLA has to show that they can fund their growth plans, long-term, with revenue. Otherwise, all of us investors get diluted to the point of no returns with secondary offerings, etc. Don't forget that TSLA's mission statement isn't 100% correlated with stockholders' interests.
Best of luck to all!�
1/1/2015
guest OK, that makes sense. There is a lot of emotion in the stock market and that drives prices as much as the fundamentals, if not more over the short term. I can see a number of investors becoming bears on TSLA for the reasons you cited.�
1/1/2015
guest I own TSLA and the company mission statement is 100% correlated with my interests. But I freely admit I may be atypical.�
1/1/2015
guest Their mission statement (as well as the fact I really wanted to own a Model S) is the main reason I originally invested in TSLA.�
1/1/2015
guest Chevron's huge amount of oil field wastewater in California is sold to farmers to irrigate crops. This is just one more straw to add to the camel's back:
Crops raised with oil field waste water in CA
http://www.latimes.com/local/california/la-me-drought-oil-water-20150503-story.html#page=1
http://www.desmogblog.com/2015/04/08/urban-water-use-restricted-california-regulators-give-oil-industry-two-more-years-operate-injection-wells-groundwater�
1/1/2015
guest Not really a fundamental but I thought this was interesting.
Year end closing price and return from previous year.
2010 $26.63
2011 $28.56 7.25%
2012 $33.87 18.59%
2013 $150.43 344.14%
2014 $222.41 47.85%
2015 $240.01 7.91%
So far no year with less than a 7% return.�
1/1/2015
guest What was the process of obtaining the certificate? I do not yet own any TSLA, but plan on buying at least one share before the Model 3 is revealed. Can you get a stock certificate if you already own a stock? Specifically, I plan to purchase the share(s) via Robinhood.�
1/1/2015
guest Well, if you want to own one share on principle or to attend stock holder meeting or something just buy one share in a brokerage account. That stock cert was just to be cutesy; a gift to my daughter. It is also sort of fake, it has a corresponding account at computershare. I got it via oneshare.com. Easily the most expensive/worst way to buy a share. I probably paid $100 in service fees to do it. And no, I don't think you can get a stock certificate any other way.�
1/1/2015
guest Thanks for the reply. I found this on the Tesla website under Investor Relations:
Can I obtain a stock certificate for my Tesla shares?
Stock certificates are not practical or cost effective for the shareholder or for Tesla. For pricing please contact Tesla' s transfer agent at:
Computershare Trust Company, N.A.
250 Royall Street
Canton, MA 02021
(800) 662-7232.
�
1/1/2015
guest Anyone know how much this costs?�
1/1/2015
guest I was able to speak with a representative. She noted that the cost to obtain the physical certificate is as follows:
$25 if ordered via Investory Center on the website or written mail
$40 if ordered over the phone
The website is here: Computershare United States
�
1/1/2015
guest Plus $40 the frame
�
1/1/2015
guest This reminds me of when all the Boston Celtic fans wanted to own a share. This was during the Larry Bird era. Transaction costs were more than the share was worth.
I'd love to see what an actual Tesla stock certificate looks like. Traditionally a company will use lots of intricate art work and scrolly artwork to help prevent counterfeiting. At one time the certs for Disney were worth more as pieces of art than the actual stock value. Pics of Micky and Minnie Mouse.�
1/1/2015
guest I bought one piece of stock in certificate form, and hung it on my "Tesla Wall". Blue print, a seal, some signatures, the word "TESLA" on top in Tesla font. A couple stripes of blue pattern stuff down each side. It looks official, business-y, but I would not call it scrolly. If you really, really want to see it, I guess I could scan it for you, but it's hardly worth the effort. It is in a Target black frame.
Back then, I paid about twenty bucks for the stock, probably that much for the frame, too.�
1/1/2015
guest It's pictured right there in Nomad's post #3306, referencing austinEV's earlier post ...all of one page back:
Long-Term Fundamentals of Tesla Motors (TSLA) - Page 331�
1/1/2015
guest Yesterday�s Faraday Future debacle opened up another perspective on competition (at least for me).
Back in the day Tesla was considered dead on arrival for numerous (potentially valid) reasons. Car making is capital intensive. Nobody new survived in the past many decades. So on and so forth. There was even a �Tesla death watch� somewhere on internet.
Shorts proudly shared this list with the world:
List of defunct automobile manufacturers of the United States - Wikipedia, the free encyclopedia
Now Tesla somehow made it through, if not a mass market car company yet, at least as a high end luxury car company so far. With a credible long term plan for mass market.
Now shorts turn around and say that there is plenty of competition coming up. The thinking goes, because Tesla succeeded it should be easy for others to succeed as well.
Duh! wasn�t the original premise that almost all new car companies fail and thus Tesla will fail? So why would competitors not fail for the very same reasons?
The fact of the matter is, the situation is exactly opposite of what shorts imagined. Musk solved an extraordinarily difficult problem to solve. And because he is able to solve it, he created just as much of a lead/distance from potential competition.
This gives a lot of confidence in the very long term prospects of Tesla. Maybe we will end up with a single transportation company in the entire world in the very long run.
There is literally no competition in sight. And any new competition is prone to die.�
1/1/2015
guest Btw, this list has well over 1700 car companies.
List of defunct automobile manufacturers of the United States - Wikipedia, the free encyclopedia
And this list has mere 42 survivors.
List of automobile manufacturers of the United States - Wikipedia, the free encyclopedia
So purely statistically speaking, looks like we need over 40 potential competitors to have 1 genuinely competitive company.�
1/1/2015
guest Very interesting reading. I had no idea there were so many failed car companies. Perhaps this is obvious, but I'm struck by the number of companies that failed during the infancy of the automotive era - it's the vast, vast majority of the 1700 companies. As we are in the infancy of a new automotive era, I expect the proportion of failures (FF, looking at you) to be similarly high. Really puts things into perspective and strengthens my view of holding TSLA forever.�
1/1/2015
guest Elon Musk now has 4 start ups under his belt. 3 are still around as independent companies and his first is part of another company. He has had some luck, but most of his success is down to talent and skill. Tesla barely made it out of infancy, but it did. Tesla made it, so the thinking is other companies can now jump on the bandwagon and they will have an easy ride. Many of the pitfalls Tesla faced are still there for other start ups and if it isn't as difficult to break into the car market now, it's almost as tough.
Outside of China and a few other places in the world, cars are a mature market. Most people buy new cars to replace an existing car or they buy a used car. Very few people in the developed world get a new car as their first. Any new entrant in this market is going up against well established brands in a market that is only growing at the rate of the population, if that. Somebody new needs to bring a "killer ap" (as it's described in the tech world) to convert customers. Tesla did that bringing new tech into cars as well as a new driving experience. Anybody who follows in Tesla's footsteps now are considered like the IBM clones in the mid-1980s: inferior machines that might only be attractive based on price.
In China, there is room for a successful start up to gain traction. Owning a car is a big status symbol there and the market is growing fast. With few used cars, most people buying their first car are buying a new one. China is where the US was when Chrysler started in the mid-1920s. With a growth market, start ups have more room to make mistakes and survive. There is enough demand, even a product that is decent, but with some flaws can make it.
Apple has become the company they are by finding markets people haven't exploited yet and coming out with a product everyone wants. They did it with the iPod, the iPad, and the iPhone. There were products of some kind in those markets before Apple came in, but they were not strong sellers. Apple blew those markets wide open. Apple's products had some problems at the start, but they still sold well because it was the hot product everyone wanted.
But a lot more people can scrape together $500 for a new iPhone than people who can scrape together $100K for a Model S. Cars are, by their nature, one of the biggest investments most people make. Owning a house is the only one bigger for most people. Without customers lining up around the block with cash in hand, the room for mistakes is much narrower.
A couple of weeks ago I posted my essay on why I thought Tesla was not going to fail, running through the 20 reasons start ups fail. If you apply those 20 to the "Tesla killers' the press are talking about, most have quite a few potential failure points.
Of the potential competition in BEVs, Apple might make it because they have a lot of money to make a lot of mistakes. It might cost them 10X what they were expecting though. Some traditional car companies will likely make the pivot and survive. Even if the public makes the flip and starts demanding long range BEVs, supply of batteries is going to be a bottleneck for more than a decade at least. It takes time to build 200 Gigafactory equivalents (could me many smaller factories) to supply all those batteries needed. Tesla will be in a position to supply more BEVs than anyone for a while, but because of the potential size of the market and the physical impossibility for Tesla to spool up that fast, traditional car makers will have an opening to fill demand, but there will be a couple of years of short supply of BEVs while the big guys retool.
Some ICE companies will go under. I would put Chrysler at the top of this list. Many others are strong enough to survive like Toyota and Nissan has more experience in BEVs than any other traditional car maker at this point. I don't know about GM or Ford. Ford is under massive debt, but they are seriously looking at PHEVs which could be a stepping stone and GM is only a step or so behind Nissan in the BEV market, though it's a tiny part of their overall business. BMW too is innovating into the battery vehicle sphere. At least those companies will have some in house expertise when the change in demand comes. Companies that are only building compliance cars because they have to will be in a lot of trouble when the time comes.�
1/1/2015
guest All four are still around as independent companies. PayPal spun off from eBay last year.�
1/1/2015
guest Zip2?�
1/1/2015
guest Yes, Zip2 is the one that doesn't exist as a separate entity anymore. It was bought and absorbed by Compaq in 1999. I was counting Paypal as a separate entity. Even when owned by Ebay, Paypal's identity remained separate.
In the end, Elon has made money from all 4 of his start ups, which is an amazing track record.�
1/1/2015
guest http://thinkprogress.org/climate/2016/01/08/3736911/electric-vehicles-washington-post/
Tesla has 10% of the global market
�
1/1/2015
guest Charles Lane is indeed clueless and remarkably resistant to the facts in evidence, but I would note that Apple has not "revealed it is working on a PEV" or BEV or any sort of EV. At this point it is just speculation that Apple plans to build a vehicle. It seems possible, but far from certain.�
1/1/2015
guest Capital Market Laboratories — The Secret is Out: The Apple Car is Coming
�
1/1/2015
guest The linked article presents as evidence that Apple has acquired 2 domain names with suggestive names. Though I agree with the conclusion, I find the cited evidence inadequate to draw the headline's conclusion.�
1/1/2015
guest I had an interesting conversation during last weekend with an old friend from university. About 15 years ago during university I was lucky to being able to work on interesting advanced computing research in the area of stochastic computing, fuzzy logic algorithms, neural computing, pattern recognition and others. It has been an incredible interesting couple of years as HW and SW had made advances that made it possible to solve problems with an accuracy much more precise by an order of magnitude compared to only a couple of years (less than ten years) earlier. It was a great pleasure to travel around the globe to share the new research results from our university with people working on the same research at the well known universities on this planet. During the following years progress seemed to slow down. But the most recent research results make me smile again like 15 years ago. They show that the slow down is no longer true for the most recent years (less than four years), e.g. I am 100% speechless about the progress of deep neural network learning algorithms. The progress is incredible and far bigger compared to the revolutionary progress I experienced about 15 years ago. To put this progress during the last couple of years into perspective we tried to estimate the percentage of the population in developed countries that might understand what is currently going in neural computing. While my estimate was about 1% my professor said nah, more like less than 0.1% of the population. I said 'But this is crazy' and he answered 'Yes it is, cheers!'. Btw his reasoning is as follows: you need to have studied computer science, you need to have had the chance to work on the topic of neural computing and you need to have worked on the topic during the last four years. Only then you will be able to know what neural computing is really able to accomplish in year 2016 and more important you might be able to estimate the progress for the next couple of years. Think about this for a second, only 0.1% of the population in developed countries might understand this technology and might be able to judge if this new technology will be a success or not.
Why am I sharing my weekend conversation with my former professor?
There are two reasons:
1. I have been wrong sceptic about fully autonomous driving. I have even shared my sceptic opinion here on TMC. But occasions like the talk with my former professor or people working on autonomous driving projects at German car manufacturers (friends from university working at these companies) make me change my mind. The progress of autonomous driving is directly linked to the revolutionary progress of deep neural network learning during the last years.
2. I have been wrong sceptic about the Tesla Motors OTA update mechanism. I know that German car manufacturers want to copy that, but they can't. The hurdles are too high.
The well known iPhone-hacker who demonstrated autonomous driving with an off the shelf production ICE and cheap of the shelf sensors makes this revolution visible to a bigger audience that might might not have insight into newest neural computing research.
And please don't fall for the statment that his autonomous driving tech is not perfect yet.
The main issue is the established car manufacturers dumped millions of dollars during the recent years for autonomous driving tech and now there is one lonely hacker out there showing the world a much simplier and cheaper solution. This is what I call a revolution!
Next point:
Companies like Uber, Google, Apple might do some good research on neural computing, but they are missing a competitive vehicle to deploy their autonomous driving tech. This is the reason why I would not be afraid from competition of these companies.
Next point:
Because of this technology change to DNNs for autonomous driving the first mover advantage of the traditional car manufacturers working on autonomous driving tech is gone. This is the reason why I would not be afraid from competition of these companies.
Next point:
Tesla Motors currently is far ahead of the pack with their autonomous driving tech and OTA updates are helping the company to further increase the distance to the slower moving players as the innovation rate is high enough to deploy improved and additional features every quarter to the fleet of all autopilot capable vehicles.
My thesis:
I now belive the way Tesla Motors is implementing developmemt of autonomous driving tech, this feature will generate more sales than any other feature of a Tesla Motors vehicle.
Just compare a Tesla to vehicles from any other car manufacturer, only the Tesla will stay up to date during the next couple of years. Any other car of any other brand without comparable prerequisites will get old soon and the owner bored. The next time this owner will join the Tesla brand.�
1/1/2015
guest well maybe not 100%... :biggrin:
It it is good to hear that, though. There are so many challenges with fully autonomous non-highway driving (human drivers being unpredictable/incredibly stupid, unusual combinations of road conditions/markings/construction, dealing with flat tires and other mechanical breakdowns - the more you think about it the longer the list grows) that it's reassuring to hear that the latest HW/SW might truly be up to the task...�
1/1/2015
guest Would you please explain that statement?�
1/1/2015
guest Yes, this is easy. DNNs are able to solve more complex tasks compared of traditional sensor fusion or outdated neural networks. This makes a lot of algorithms developed during the last years by the traditional car manufacturers useless. The DNN is now capable of this functionality.
Just watch the video about the self driving tech of the iPhone hacker.
Afterwards you will know what I am talking about.
What I would like to point out is not the money lost during the last years for developing useless code is important, the big issue is the advantage that they started earlier with autonomous driving tech development is gone for the traditional car manufacturers. They have to start 'from scratch' like Uber, Apple, Tesla, Google and others!�
1/1/2015
guest One thing that has always puzzled me is exactly *why* tesla seems to have a technical lead on some things. OTA is pretty trivial. Autopilot is largely a product of mobileye, or *could* be even if TM isn't using it that way. I am starting to think that it is just courage. Existing big automakers are slow to move because they expect regulatory hurdles and the expectation of years of testing and qualification for every little thing. TM is just doing it and pushing it on the market. That might eventually backfire (eg the first autonomous fatal accident) but for now it looks like a tech lead.
Regardless of the method, an edge is an edge. But I think we make a mistake when we say TM has amazing tech that no one else can match. It is probably more accurate to say TM has tech that no one else is matching out of caution.�
1/1/2015
guest I wouldn't say OTA is trivial. OTA updates in itself is of course trivial, most modern cars do have the tech to do OTA updates. They have the modems and at least for the entertainment they have software driven interfaces. The big difference is the way Tesla has been incorporating OTA updates in their entire architecture from the design process. The impression I'm getting is that a Tesla is more safe from software hacks than any other car on the market (excluding older car without any software of course) because they accepted that someone will try as they designed their processors and software. That's why no other manufacturer gives you more features, like adding creep, through OTA updates.
Cobos�
1/1/2015
guest Honestly I see computer science principles again and again. Think about layers, modularity, quick incremental updates (both vehicle HW tech as well as computer HW tech As well as SW), communication strategies for the fleet, building networks, plug-in architecture, capability for updates. This all does not happen by accident, this is the Silicon Valley DNA at its best.
Some time ago I thought a lot about if Tesla Motors is really needing OTA updates, today I think Tesla Motors would not be Tesla Motors without all these computer science principles.
- - - Updated - - -
Thank's for chimin in.
I totally agree.
It is useless for an ICE engineer to get remote access to an internal combustion engine. No way to diagnose a broken part of the engine. For a computer science geek you always want remote access capability. Only the computer science geek wants remote access and remote updates!
- - - Updated - - -
BTW mobileeye is adding an other data point. They have been able to become the market leader and such a big player in no time because there is currently a revolution happening. Nobody stunned that it is not Bosch or Continental or others that are delivering leading edge autonomous driving tech?! They want, they even demonstrated something but they are not able to compete with mobileeye, they just can't!�
1/1/2015
guest I concede the point about OTA. Probably others can't do it because it wasn't designed in from the ground up to support it. You are essentially self-flashing the cars firmware, which is a fairly delicate and mission critical problem. Probably GM can nail 98% of OTA flashing easily but the last 2% requires a major architectural rethink (yes, I know conti or bosch).
To my point, TM doesn't have inventions, they have conviction to use available tech.�
1/1/2015
guest I know Audi needs to bring some vehicles to their service centers for a day just for a SW update, think about it, this is crazy!
Their SW is so badly structured and badly packaged that they need up to 2 DVDs for a firmware upgrade. These 2 DVDs contain lots of useless SW functions only needed for other vehicles!
I do not want to disagree with you
Tesla Motors does not posess inventions like modularity, upgradeability, layering structures and so on.
The important thing is they implement their tech by following these paradims.
The result is amazing things like an old Model S getting more efficient over night, or getting spotify over night or self parking, or ... I could go on and on�
1/1/2015
guest This is relevant to the long term thread: These OTA feature advancements never move the stock... but they might move me. I have a strong suspicion that the good model S sales is due to people upgrading old model S's that don't have autopilot sensors. My lowly 2013 P85 is starting to feel obsolete.... I might have to upgrade my dang car to be one of the cool kids.�
1/1/2015
guest Next thesis:
First time I heard Adam Jonas talking to Elon Musk about fully autonomous driving and fleet services and pressing Elon Musk for an answer during an earnings call I tought this is a Little bit far fetched and crazy.
With the information from Elon Musk from yesterday I say Adam Jonas knows more than we do!
The stars are aligning and the bigger picture is getting more and more obvious.
This is the reason they are all working like crazy at Tesla Motors.
They are surprised that they got a second open sliding window for becoming world leader for autonomous driving tech and fleet services like they got for their electric propulsion drive train based on Li-Ion tech.
If they get this right during the next three years, they know Tesla Motors is here to stay for ever!�
1/1/2015
guest It is interesting. If you read about amazing tech advances on the internet you pretty quickly get a cynical thick skin. Sensational headlines that signify less than the say are the norm and you just tune it out. 99% of people who hear about Tesla autopilot, self parking etc probably already think "yeah right", sensational click-bait story. I admit I sometimes think "yeah right" when Elon wants fully autonomous driving in a short time period. But... here is that rare case of the headlines meaning exactly what they say. So, a higher than average credibility factor needs to be assigned to Elon's plans. We have an information advantage here, how to make money off of it is the harder questions.�
1/1/2015
guest A single OTA feature advancement does not move the stock, no reason to do this.
But the general capability of OTA feature advancements for a vehicle during the next years is driving Model S sales!
In order to get new autonomous driving tech for one of the vehicles of the traditional manufacturers you'll have to buy a new one, this is just a result of the SW developing process of these companies. With a Tesla you will not feel any pressure to buy a new vehicle because of better autonomous driving tech is abailable, you just go with the over night OTA!
- - - Updated - - -
Normally I am not able to add anything substantial to TMC.
But I do know this business
Basically Elon Musk gave Adam Jonas the answer to his Uber-question, only a couple of months later than Jonas hoped.
Musk publicly stated that Tesla Motors will be there in about two to three years!
And this is going unrecognized by the general public?
This is ridiculous!!!
- - - Updated - - -
Honestly, I recommend to everyone with some spare time of weeks to months, get an account like ieee and dig into current DNN research it'll be worth the time!�
1/1/2015
guest I heard JB Straubel in an Youtube talk where he was talking about working with other automakers and how conservative they were about software and firmware. He said most of them were scared to hand over major systems to complete software control without some kind of mechanical back up. One example he gave was Mercedes who was concerned about a motor going too far and the Tesla solution was, we just make sure it doesn't in software, but the Mercedes people wanted a steel bar to prevent the motor from swinging too far and that's what they put in the car.
Tesla evolved out of Silicon Valley thinking and their core expertise going in was software. The engineers there fundamentally trust software and trust they can write code that prevents bad things from happening. Other car companies, coming from the world of mechanical systems are wary of software and don't want it controlling critical systems unless they have to and even then they want mechanical backups to prevent the worst from happening. It's a mindset that comes out of the history of those companies. The second youngest car company currently building more than a handful of cars is Chrysler who has been bending metal and building mechanical systems for 90 years. Their roots predate the transistor and electronic control systems in general. And that's the youngest of them. Tesla is a product of the 21st century and probably everybody working there was born after the invention of the transistor. There are probably few working there who have ever used a slide rule. The bulk of the junior employees have probably never known a time without a personal computer available to them in most places.
This software reliance can go too far. I suspect that's happened at Faraday Future. I watched an interview with Nick Sampson earlier today and he sounded like they were so enamored with the software they aren't really thinking enough about the hardware.
Tesla may have excellent tech and the guts to innovate to the edges, but they also build good mechanical systems. The Model S has had some reliability problems, but compare it to products from other start up car comanies. It's been unusually reliable for a start up company's car.�
1/1/2015
guest Thank you ev-enthusiast for the clue-in
If that is the case I can see where this 2-year estimate confidence comes from. If the "eye" already is pretty close to correctly interpreting what it sees, it really is just a matter of putting effort into developing decision-making algos. It's not rocket surgery and quite predictable, incremental job.�
1/1/2015
guest Nonsense. Look at the Summon function announced last year compared to the limited functionality delivered in 7.1. Musk is advertising to fully staffing full autonomy development last month, and this month predicting full autonomy in two years.
What I see a guy who needs to go to the capital market to continue building the gigafactory, and is worried at what that will cost.�
1/1/2015
guest ..and your argument is? We're having a discussion about the technical details of what the solution architecture might be here. It's not not even specific to Tesla but might have substantial influence on some investment decisions.�
1/1/2015
guest As much as I want to believe Elon Musk, two years is the usual over optimistic Musk speak. What we are likely to see is hardware included in 2 years but the actual autonomy maybe by 2020 updated OTA for everyone who has sensors 2.0. Which puts Tesla far ahead of the pack simply from having a large existing capable fleet but the whole 2 years needs a large dose of salt. But, for everyone else the improvements will be incremental and unavailable to the existing fleet. So even if everyone reaches autonomous cars by 2020, Tesla will already have 250-350,000 cars capable on the road, while they will just be starting their rollouts.�
1/1/2015
guest I don't remember a timetable given for fully functional summon. I'm quite sure they have much more capability ready to go but they are releasing it in small steps to avoid any potential issues, not to mention regulatory backlash.�
1/1/2015
guest Exactly!
Listening carefully to Elon Musk's statements about predictability of the outcomes one realizes that he makes a reference to the driving errors a human driver is doing on average ('...better than a human driver...'). This is the way to go. Nobody will be able to get to a 100% failsave system. And nobody needs to, only avoiding most of the vehicle accidents resulting from a distracted driver would be a great benefit for all traffic participants.
And yes, low level decision making can be done by DNN. They can provide contextual information. One of the reasons why this tech makes it easier to build autonomous driving systems by an order of magnitude!
Would be great if we had someone from Tesla autopilot project here who could chime in and comment on the way Tesla is implementing this in general!? Without going in company relevant details
Again, I strongly recommend digging into the latest research results on deep neural networks, most stuff is public research articles and available on e.g. ieee. And please skip every article older than 5 years! Not worth reading.�
1/1/2015
guest Hey ev-enthusiast, this is a bit off topic but are you a computer scientist? Are you actively doing research in deep learning? If so, what's your area of focus?
For those that don't know, nvidia recently announced (at CES I think) a gpu specifically designed for cars, I think this unlocks a lot of potential. As transistor density increases, and the physical footprint of computational devices get smaller, we move towards a world where a lot more computation can be done on the device without having to communicate back up to the cloud infrastructure (think cars, cell phones, wearables etc.). More compute power in the device means more complex algorithms, better analysis tools, and a generally more robust system. I would suspect any self-respecting autonomous driving company to start thinking about ways to integrate the nvidia gpus in their cars.�
1/1/2015
guest Actual autonomous driving will require resolving the environment precisely enough to determine the intent of pedestrians. That is not going to be done with radar and sonar, regardless of what nvidia processor is used.
Tesla has given no evidence that they are working on such a project, other than musk predicting delivery in two years.
Straubel said in 2015 that autonomous driving would take nine years. That's prediction is equivalent to saying that commercial fusion will work in the next decade.
So what changed at Tesla?
If Tesla implements autonomous driving this decade it will most likely be by integrating a system from google or another company. That's a perfectly acceptable choice for a car manufacturer, but not particularly useful in boosting the stock price or decreasing the cost of the additional capital they need.�
1/1/2015
guest I think they mean autonomous driving with a driver at the wheel who is responsible for intervening, aka "autopilot everywhere". That is the only way the time frame makes sense IMO.�
1/1/2015
guest Yes, I am a computer scientist. I did some research on neural computing and other stuff at university. Was an amazing group of people and an amazing time as the technology made some progress back then and eberybody got better and better results by the improved algorithms on the same old input data. But as I mentioned earlier, the current progress by DNNs is a much bigger leap forward compared to what happened 15 years ago. We have been working on a lot of areas, e.g. security, medical, pattern recognition, industry applications.
Right now I am working in the automotive industry in Germany, some friends working on autonomous driving projects.
Nice to point out that nvidia is doing an amazing stuff on DNNs right now, thank's.
They are absolutely aware about what is currently going on and are marching full speed ahead!
Any cooperation between nvidia and Tesla would be amazing!
Do you have some experience on DNNs or autonomous driving tech?
- - - Updated - - -
Electracity, I do not want to sound arrogant.
I do not know much about the Tesla autopilot system, but this is what they are actually doing.
They have a pretty damn amazing model for the environment, I would love to learn more about what they are doing there!
BTW what the hell are you talking about Google? Just recently mobileeye admitted that they are surprised about the performance Tesla is able to squeeze out of the current mobileeye sensor suite with their own algorithms. What reason do you see for Tesla to integrate a third party solution and throw away their working algorithms that are already deployed to most of the vehicle fleet and working under real driving conditions?�
1/1/2015
guest Yep, I currently am doing research on deep nets and un-supervised learning. Have worked on (and continue to work on) a bunch of vision/nlp related projects on the side. I might also end up working with some autonomous driving projects fairly soon.
We can discuss the details of DNN offline if you want.�
1/1/2015
guest Humans can't determine the intent of pedestrians, why do you expect computers to be able to do so? Even if humans can (sometimes they do), most of the time it's not a good idea to make driving decisions based on that. Just follow the rules of the road and you'd be better off.
Now understanding a flagman at the intersection, that's an important ability. I'm pretty sure it's been worked on or solved already, and it's definitely within the envelope of what the existing technology can do.�
1/1/2015
guest Autonomous driving algorithms today can probably handle 99% of situations, it's the 1% of odd scenarios that throw them for a loop. Humans make mistakes in those situations too, but we forgive humans more than we forgive machines.
Humans also have some advantages over computers. Silicon can compute things many times faster than we can, but humans are far better at quickly winnowing down the options that aren't worth considering and only working on the options that are most likely. With things like fuzzy logic computers are getting better at thinning the possibilities without calculating all scenarios, but humans are still better than that.
Frequently we draw comparisons between human minds and computers because it's the best comparison we can make, but we really are very different on some very fundamental levels.�
1/1/2015
guest The thing is, unlike humans computers are not moody, always pay attention, and once tested to work good for a particular situation are good to go. Humans, you have to train each one individually. This is I think the crux of what Elon is talking about. As time goes by that 1% where computer stumbles becomes 0.5% then 0.1% and at some point you're done. Every accident that actually happens can be analyzed, replicated and fixed in software so it won't happen again. Not the case with humans, we don't have instant over the air update to our driving algo. Being 10 times better than humans is not that great actually since we kinda suck at this.�
1/1/2015
guest I agree the fringe cases where the AI doesn't know what to do will reduce over time. They are probably under 1% now.
Whether we like it or not autonomous driving is coming. The older generations like boomers tend to not want it (I know a lot of people over 50 who hate the idea of autonomous driving), but millennials really want it. In the long run, cars and driving will be safer overall than it is now.�
1/1/2015
guest I'd bet many of those over 80 or so would really appreciate the potential return of mobility a truly autonomous car would provide.�
1/1/2015
guest Anyone with compromised eyesight/mobility/coordination.�
1/1/2015
guest Here is what autonomous car development actually looks like:
http://www.wired.com/2016/01/google-autonomous-vehicles-human-intervention/�
1/1/2015
guest Then there is my father who is 96 next month and bought a new car in 2013. He doesn't do road trips anymore, but he does like driving himself.
But to the bigger point, people with handicaps that make driving difficult would probably like self driving features. For example people of any age with seizure disorders aren't allowed to get drivers licenses. They either need to rely on other people to give them rides or on public transport.�
1/1/2015
guest Yep. Humans are god-awful terrible at driving, and our actions cause needless traffic. Your numbers observation is correct; if we can get to 10x better than humans that is an enormous public good. The thing to think about is, yes autonomous cars will be in crashes and occasionally even kill people. Nasty business. I don't look forward to sensational headlines. But, it is a numbers game. If the autonomous cars are 10x safer we just really need to accept that, and not hold them to a false standard of anything more than zero is unacceptable. Besides, if ALL cars were autonomous there probably would be zero fatalities (barring people jumping into roads and natural disasters and stuff).
I remain an optimist. People seem to fret that the public will resist this change, or than policy makers will judge issues harshly. But I think people get that we are bad drivers. Policy makers REALLY know this. Insurance companies know this to the 4th decimal place. I think the power structures will skew pro-autonomy and the country crowd pro-manual. Might even become a right-left or urban/rural split, god forbid.�
1/1/2015
guest Very cool to see that Google went from a potential crash every 6K miles to 95K miles in such a short time (about a year). That looks to be a trend that will get them to a potential crash every 1M miles this time next year!�
1/1/2015
guest That's NOT what EM meant, because he was talking about summoning a Tesla from California to NY and he also said that plan B hardware will be required due to not having a driver to take over if necessary.�
1/1/2015
guest Correct MitchJi.
One thing that doesn't add up: Musk said on the one hand that you could summon your car from anywhere in about two years (meaning full autonomy), but on the other hand Musk said in the press conference last sunday that the redundant hardware is still a long ways off. Something along the lines of "not in production for quite some time". He said they are still designing that hardware, so how this can be integrated in vehicles in two years time is beyond me.�
1/1/2015
guest Good point. My guess is that it has something to do with the dance Elon has to do being CEO of Tesla Motors between simultaneously attracting investment while not turning off current and near term customer base. In other words, he does believe it is coming soon and we should invest accordingly, but does not want people to sit on sidelines not buying Model S and X while waiting for it.�
1/1/2015
guest I know what he said. I am applying an Elon filter.
He has this curious habit of creating corporate goals in the public sphere. This is him telling his team to do full autonomy in 2 years, I think what we will get is autopilot everywhere if for no other reason it won't be legal by then.�
1/1/2015
guest You know that next gen sensor suite already available at mobileeye?
Again, I strongly recommend to dig into current search results on DNNs at e.g. ieee!�
1/1/2015
guest Airliners have pretty much been self flyable for years, but regulations still require two pilots. There was a joke around Boeing when we were developing the 777 that the modern flight desk has just a pilot and a dog. The pilot's job is to feed the dog and the dog's job is to bite the pilot if he touches anything. Most airline accidents in the last 20 years have been due to human error or human malfeasance, but the regulations still require humans at the controls.
There are a lot of advantages to self driving cars and once the software is worked out, they will be safer, but it will take a while before people trust them.
Another thing that will be possible when cars drive themselves and can communicate with each other is they can drive much closer together at much more uniform speed. If some kind of slow down forms, cars miles back will know about it instantly and can start slowing along with all the cars in front of them. This will allow them to travel much closer together than a human could safely manage and most cars in the pack would get energy savings from drafting the car or truck in front of them. It would turn rush hour back into rush hour on freeways and save shedloads of energy.
I expect in 20 years or so the only roads where humans can drive themselves will be back roads and the major highways will be autonomous vehicles only. The technology to do it will be there a long time before regulators feel OK with it. Tesla is going a long ways towards helping this to happen by proving the autopilot can operate safely on existing roads with mostly human drivers all around.
- - - Updated - - -
Reading between the lines, I interpret this as meaning the technological capability will be there in two years, but from what he said elsewhere, the technology will not be available for sale for some period longer than that. Tesla may do some sort of publicity stunt driving a test car across country with the driver never touching the wheel or something like that.
They showed off the auto plug in arm last year. It will still be a year or two before that is installed anywhere. The technology is there (meaning a prototype exists), but it is not available yet.�
1/1/2015
guest There will probably be crashes still that happen due to unexpected road conditions, hardware failures and things like that. But due to the nature of the system, preventable crashes that happen due to the system making sub-optimal decisions will soon disappear.
The way I see it is there will be a point where these things are "safe enough". Then we'll see a lot of continued investment in investigating crashes or near crashes and modifying the system to cope better. At some point there will be very, very few preventable crashes happening.
I also expect that there will be public investments into modifying road infrastructure to better suit auto-driving cars. Pretty sure there will be a bunch of optimizations available that aren't possible with humans.�
1/1/2015
guest Yep. That is the most exciting thing of all. A 4 lane highway could become 5 lanes just by packing tighter. Cars could stage before the onramp while the rightmost lane has "gaps" developed for them. Got a slow old gas burner? The gap is bigger and further back so you can hit it because the system knows your specs. Debris in the road? one car reports it, traffic routed around it parting-of-the-red-sea-style and traffic continues.
Stoplights go away. First they are turned off but the system mimics it with each direction taking distinct turns, then eventually the traffic becomes interleaved where east/west is bunched up into groups with gaps before and after and north/south is bunched up with gaps before and after and they just all flow throw with no one ever stopping. There are loads of flow optimizations you can do once the driver is removed.�
1/1/2015
guest Apologies I posted this earlier in another thread but should have posted this here since it deals with Tesla as an investment.
I don't think Tesla is a bad company. I applaud the lofty goals of changing the world. I love solar, and I think there will be a role for partial- or fully-electric vehicles in the future, since EV's are better for the environment than gasoline.
I also like the model S and applaud the audacity of starting a new US-based automobile company. Model S has a great exterior look and great performance. Although inside, I hate giant touch screens and prefer simple buttons and knobs. I also think the interior is somewhat unimpressive for the price.
IMHO Model X is ugly ... looking like an inflated Prius. Falcon Wing Doors are a ridiculously complicated solution to a problem that did not exist. From an investment perspective, the current market valuation seems "ludicrous". My valuation analysis suggests the stock worth about $25 per share based on a discounted cash flow analysis, which assumes the following:
1. Tesla actually sells 300,000 cars in 2020 (still a huge increase from 50,000 in 2015). On any planet, 6x growth is a huge leap and just plain unlikely.
2. Tesla and somehow actually makes 10% gross margins on the Model 3, even though many believe Tesla will be lucky to break even on Model 3.
3. Tesla achieves continued 25% gross margins on Model S, as well as 25% margins on the Model X despite its massive complexity.
4. Tesla overall company operating margins of about 4% (even though F and GM have massive economies of scale and still manage only 4% margins on $100 billion in sales). This despite Tesla's massively negative operating margins today.
I think all these assumptions are optimistic for Tesla, and still the analysis suggests TSLA is only worth $25 per share. I put my money where my mouth is and am short via owning long-dated puts and shorting the shares directly.
The gigafactory seems like a terrible idea, since Tesla has zero experience manufacturing batteries, and others (LG Chem, Panasonic) have massive economies of scale and minimal margins. Compare to Apple, which realizes it core competency is designing beautiful, simple products and outsourcing the low-margin component supply.
Nothing against Tesla as a company. But basic impartial math suggests the $200 stock price is ~8x overpriced.�
1/1/2015
guest Actual users seem to disagree.
Your personal preferences in looks are irrelevant, and you seem to have forgotten that the Prius sold quite well in spite of it's looks.
They are an eye catching advertisement every time the are used. Any star wanting to make an entrance better be stepping out of a Model X.
Tesla has been making batteries for years. I assume you mean cell manufacturing, and while Tesla has it's own cell development research team it is of course partnering with Panasonic.
Are you completely unaware of the Panasonic partnership?�
1/1/2015
guest You do realize the Gigafactory is a joint project between Tesla and Panasonic. Panasonic will be making their batteries there for Telsa. Elon Musk also has a track record of figuring out how to do things he had zero experience doing. He built a successful rocket company from zero experience and he has lasted longer as a car start up than anyone, starting with zero experience building cars.
I think the gross margins on the Model S are closer to 20%, they make around $20K per car and I think the average price is around $100K. They also have Tesla Energy, which is a much needed technology for utilities. Even if the Model 3 is behind schedule, the Gigafactory can build batteries for Tesla Energy in the meantime.
Tesla is an audacious venture, but then so was Apple in 1977. Maybe Tesla is only worth $25 a share. I'm not an expert on stock analysis though very dynamic people like Elon Musk can be a wild card that makes stock analysis go wonky (in both directions).�
1/1/2015
guest Lets say that enough people agree with you about inside of S, and cool factor of X doors. What do you think Tesla stock is worth? I say $25 per share. Tell me your sales projections. Margin assumptions. Agreed Tesla is a cool brand.�
1/1/2015
guest Thank you for this. Really. It's hard to come here and tell the short side, and put money on it. Cudos to you.
The reason Tesla is more vertically integrated than Apple is, in my opinion, scale. Apple has the buying and bullying power to get suppliers to do anything, anything, needed to deliver on time and on budget. Tesla doesn't have that ability, so if it absolutely has to be there overnight, they make it themselves. Thus things like building their own 2nd row seats, and the gigafactory. No one was going to take Tesla's 500,000 cars' worth of cells seriously enough to build the factory needed to make the cells in that much quantity in time for the Model 3. Elon hoped others would see the light, but they didn't, so he lit his own light and put Tesla's money where it's mouth was.
I know nothing about evaluating companies, but I do know innovation when I see it. Tesla has it, Apple has it, Ford, GM...not so much. Many have said Tesla is a technology company that happens to build cars. Perhaps that's true, which is one of the reasons the bullish analysts value Tesla much more than just an auto maker. I believe (but remember, I know nothing about evaluating companies) technology companies have a higher PE ratio than auto makers. Perhaps that is partly where the extra pricing comes from. Also, I believe you have left out the valuation of the Tesla Energy business, that some have said could have more value than Tesla's auto sales.�
1/1/2015
guest �
1/1/2015
guest $25. LOL.
The market has disagreed with you for a long time now. Even if you are right (PS - you aren't), just remember the market can stay irrational longer than you stay solvent. You are going to get smoked.�
1/1/2015
guest So your name is valuationmatters but in this case I don't think it does. I think your views are pretty clear and I could try and convince you otherwise and maybe I will another time. But if I assume in the early 20s Tesla is making millions of vehicles at high margins thanks to their R&D lead and Tesla Energy is selling as many Powerpacks as they can roll off the assembly line then I arrive at a number far, far higher than $25. But if I use your assumptions I get $25. So, really, valuation does not matter when it comes to TSLA. What matters is who predicts Tesla's fate. Who is right? Will Tesla ultimately do little more than prod the major automakers into making electric cars and retreat into a niche market? Or will Tesla become an incredibly large company making premium cars, offering premium automated transportation services, and selling large battery projects to utilities and businesses? These 2 outcomes have drastically different valuations.�
1/1/2015
guest Well, I think your numbers and assumptions are wrong on so many levels. I remember in the past people predicting the stock price in the future years. A quick search found this table (author unknown). This is pretty old, and the 2015 number and later are a little off, I think total share numbers are pre-capital raise last year, and there is no Tesla Energy/powerwall contribution which would increase the stock price. But I think it is still fun. TSLA Valuation - Google Sheets�
1/1/2015
guest What is your short open position? At what time and price would you cover? Only fair given your question.�
1/1/2015
guest And as I suspected, the arguments for Tesla being worth higher than the math of $25 boil down to:
1. The battery business will be huge. (even thought today it is zero, Tesla has no history in this business, and there are established competitors).
2. The stock must be worth way more than $25 because valuation does not matter (would be the first time in stock market history, but hey maybe Tesla is the first).
3. The stock must be worth way more than $25 because it is trading at $200 today.
I invested about $80,000 in Tesla puts. If Tesla goes to $25 I make $1.5 million. If it goes to zero I make $2.5 million. If it stays above $150 I lose all $80,000. I'm fine with that.
I expected better arguments than 1-3. No math.
LOL�
1/1/2015
guest What expiry? I expect to see you in here on opex day.�
1/1/2015
guest What do you think the value of Amazon should be then? Tech stocks don't do valuation like you might think.�
1/1/2015
guest @valuationmatters - have you seen this thread: Blind Faith Price Targets�
1/1/2015
guest I believe Tesla is just another bubble stock that will pop. In the long-run stocks ultimately come back towards their true value. It's not Enron. I do believe the company will survive in some shape/form but ultimately be much smaller. When the stock falls, the company will be forced to cut costs quickly to remain solvent. Tesla reminds me of the following companies at their peaks:
Blackberry
Nokia
HTC
3D Systems
GoPro
Fitbit
all nanotechnology stocks
all 3D printing stocks
SunEdison
If you look at the message boards, all these companies had absolutely rabid followers who defended the mindlessly high stock prices. All came back to reality.
I am actually far more optimistic on Amazon which has sustainable competitive advantages by being the largest, having best customer service, generally lowest cost of operations due to scale, and strategically placed warehouses. And Amazon trades at reasonable valuation based on revenue multiple.
Interesting discussion but I am disappointed at the lack of quantitative arguments. Seems like a lot of emotional arguments, religion, and annoyance that anyone does not believe their religion. LOL it's like Mac vs. PC, Android vs. iPhone.�
1/1/2015
guest The lack of quantitative arguments is because it is not our job to explain to you where TSLA should be valued. I am not your financial adviser, why should I waste my time? Plenty here have done their own research and made their investment decisions based on it. I presume you have done yours before going short. In due time the market will determine who is right. There is no need to copy and paste the exact same post to multiple threads to try to instigate a response. If you are right, you will eventually be rewarded. If you are wrong, there is nothing you can say here that will change that fate.
BTW I don't think there is anything wrong with shorts. Someone has to be on the other side of the trade. If they get proven right then all the credit to them. However, if they(continue to) get proven wrong, they are the ones who will provide liquidity for Tesla to fund their expansion for years to come. So good luck to you.�
1/1/2015
guest TRUE. There are a lot of smart shorts publishing articles in SA and I do see their logic is much better than "dummy" bulls. Paulo Santos is one of them (although most people here will disagree), I know he made quite some money by shorting TSLA right after model X reveal while a lot of bulls lost big since that.
�
1/1/2015
guest Thanks Jesse and maoing. You guys are somewhat more balanced. It is a fascinating story playing out before our eyes. On the one hand I really think Tesla S is cool, and I'm coming around on the X. But I just can't fathom the valuation using any mathematical logic. And historically, when the math does not make sense, the stock price ultimately returns to its true value. FYI I do like solar, and I like the idea of electric vehicles. I also like the idea of electricity production getting cleaner gradually (less coal, more natural gas, hopefully more nuclear, as well as solar/hydro/wind if they make sense). Ideally these non-coal sources of electricity provide so much cheap electricity that we can then have plentiful power to do things like build tons of water desalinization plants so that we have infinite water as well. Can you tell I live in California?�
1/1/2015
guest There is no doubt TSLA will be hammered hard in next market crash based on its valuation, but you don't know when it'll come. With 7 years bull market, investors should be more and more cautious for TSLA trading. 2016 should relatively safe because there are many positive catalysts coming. But I'll worry about the 2017 and going forward until Tesla becomes a truely profitable and self-sustainable business.�
1/1/2015
guest Suggest you read some of the investment threads and quote some sections that you'd like to discuss. Your current position is well known at this point. Your arguments for Amazon I agree with in general and I'd draw very similar characteristics with Tesla.
Maybe Tesla delivers 300K in 2020 with an average ASP of $70K and dominates all markets they compete in. What would your valuation be at that point? Hypothetically of course.�
1/1/2015
guest That is exactly what is in my DCF model that produces $25 per share: 300,000 in 2020, much higher ASP though. I assume Tesla can maintain 25% gross margins on S and X with average ASP's of $94K and $102K respectively in 2020. But I assume only 10% gross profit for Model 3 (with average ASP of $47K) since I think Tesla will have a tough time even breaking even on Model 3. I also assume that Tesla ultimately gets to just a 4% overall operating profit margin which is not dissimilar to Toyota, F and GM. Finally I assume just an 8% discount rate even though many experts would say to use a higher rate. I think all these assumptions are optimistic for Tesla, but even so they only spit out a $25 valuation. Of course, I give no value to Tesla Energy which I am very skeptical about. If that business actually is profitable, it could make the company much more valuable. But as of today, I am unaware of any revenue for Tesla Energy.
By the way if the model spit out a $500 valuation I would be long the stock. But it says $25 so I am heavily short. Which is a shame because it's a really neat company. It just burns a lot of money and I think it will be tough to even get to overall profitability.
I should point out that my valuation of $25 is as of 12/31/2015. So it you are asking my DCF valuation as of 12/31/2020, just increase it 8%/year (discount rate) for five year. $25*1.08^5=$37�
1/1/2015
guest The phone companies on this list were hurt by a revolution in technology from smart phones that destroyed their market. Unless someone invents some radically revolutionary battery technology and one of the major car companies patents it and Tesla can't get it, it isn't applicable.
3D printing is one of those emerging technologies and its a smaller parallel to the early internet. A lot of companies become darlings of the market only to collapse when the money runs out. There will eventually be companies that are leaders in that market, but there will be a lot of casualties along the way.
SunEdison's problems are similar to what did in Solaryndra and other solar start ups. Prices have fallen dramatically in the solar market which is good for consumers, but it's hard on early players who went with more expensive technology and ended up too deep in debt to survive. Whenever technology changes dramatically in a short time, especially when the costs drop rapidly, companies that weren't prepared often get into trouble.
Tesla isn't facing those problems right now. Tesla is more in the position of the first smart phone rather than being Blackberry in a smart phone world.
And history also has examples of companies that were thought to be bubble stocks that turned out not to be. They are rare, but they do exist. In another post I pointed out Amazon and Apple were once written off as being overpriced too.�
1/1/2015
guest Tesla Energy has generated over $1 billion worth of reservations since its launch in the second quarter of last year. Elon Musk believes demand for stationary energy storage products will be approximately double that of cars going forward.
If we project Tesla Energy product sales very conservatively at ~77.6k Powerpack systems and ~60k Powerwall systems globally per year by 2020 and assume a $3.25k price tag on the Powerwall and $25k for Powerpacks it suggests nearly $1.9 B in annual revenue by 2020.
Musk says margins on Tesla Energy are fairly low right now, but he expects them to increase to 20% once the gigafactory is online. If we assume half of this 20% gross margin passes through to Tesla�s bottom line then that gives Tesla an extra $190M of earnings in 2020. If we apply a 20x multiple to these earnings (and then discount to present dollars) that makes Tesla Energy worth about $3.8B, or $25.88 per share.
Then there is Tesla's automotive business.�
1/1/2015
guest Would your numbers change if your comparison stock was BMW, not Toyota, F and GM? It seems that would be a more apt comparison based on model types and ranges. Just curious.�
1/1/2015
guest Using DCF on growth companies like this is not wise. Better to stay out if you are value investing and use DCF. The market are not valuing them based on DCF, but other factors that differs per company... with Facebook it was user growth, with Amazon it is free cash flow, partly because they told everyone those metrics were the most important.
With TSLA it seems to be revenue growth, gross margin and stock dilution. If that is right or wrong is up to debate.
By 2020 I think they are going to make atleast 100k S and 100k X, asp 90k. I also think they can make 200k Model 3, asp 50k. That would be 28B in revenue. Add to that lets say 3.6 billion in energy storage sales (calculated by assuming an output of 300 000 * 60 kwh = 18 GWh for Gigafactory for storage and sold for 200 USD/kwh per pack). 31.6B in 2020 is a bit under the 50% growth per year Elon has mentioned, it is more like 35%.
I still think Tesla at this point would not turn a profit but be like Amazon with everything going back into the business. The gross margins according to guidance would be close to 30% on S and X, and I assume about 10% on 3 and the same for the storage business. I think at this point the market would value Tesla about 3x revenue for a market cap of 90B, or put another way 10% to the bottom line for a imaginary profit of 3B and then because of the growth rate a P/E of 30 for a market cap of 90B.
This is my model. I don't really see much that can stop this from happening. They will get there with a lower growth rate than the current one. That said, I think the general market can take the stock down quite a bit if things get bad. I also think it will be hard to break the 50B market cap barrier as Tesla then have a higher cap than GM, Ford, VW etc.
As for dilution and capital needs I think that will be much less than what many are predicting. This is mainly because the S and X with scale increase and vertical integration will generate more per year and also because they can start sell stationary storage to fund a good chunk of Gigafactoy. Stationary Storage should be able to scale reasonable quick as they can get the cells needed for Panasonic.�
1/1/2015
guest Actually Toyota and BMW have overall operating margins of 11% and 9%, respectively, which are much better than F and GM which are mid-single digits. So you could attack my assumption that Tesla only achieves sub-4% operating margins in the long run. You could argue that Tesla should achieve 10% or more and get twice the valuation I get.
Adjusting my operating margin assumptions for Tesla to 5% in the long run, and rolling forward the valuation date to 12/31/2015 (which gets us past some massive cash outflows), my model now says Tesla is worth around $65 per share. This includes a slight increase in vehicle deliveries assumed in 2020 (from 290K to actually the 300K that I previously rounded to). I still place no value on Tesla Energy. My model says total sales of $19.3 billion in 2020. Base on all this, again now my model says Tesla is worth $65 -- as of 12/31/2015.
But I think it is a hugely optimistic assumption to say that Tesla will go from 50K vehicles in 2015 to 300K vehicles in 2020. Also a huge assumption to assume Tesla ever makes a penny on Model 3 given they only make 25% on Model S which is 2.5x the price of Model 3. And also a huge assumption to say that Tesla makes 25% gross margins on S and X perpetually. IMHO my model now include bullish assumptions. And I think the odds are overwhelmingly against this level of success. Ultimately someday could Tesla get to 300K vehicles? Sure. I just think it will take a lot longer than 5 years from now. And EV competition is going to massively heat up (thanks in part to Tesla).
$65 per share seems more reasonable. And still 4x the IPO price of $17 back in 2010. So I am not irrationally bearish. Just trying to be impartial even though I probably have my own biases. Elon is probably right about autonomous driving. But more like 20-30 years from now, not 10. Perhaps a man way ahead of his time.
- - - Updated - - -
Our posts passed each other. Thank you for detailing your assumptions. Yours are just higher than mine. You just assume modestly higher sales (my 300K vs. your 400K in 2020) plus higher your higher gross margins on S and X (my 25% vs. your 30%). And you further give weight to Tesla Energy whereas I am very skeptical of any revenue until I see actual sales. But a good discussion.
- - - Updated - - -
Actually Lango -- the one suggestion I have for your model is that you should tax-affect your $3 billion pretax profit by say 35%. So that means ~$2 billion net income x 30 P/E using your numbers, making the stock worth $60 billion equity value in 2020. I just think it will be more difficult and take longer than you think to get "there." Further factor in the likelihood of a recession sometime during the next 5 years.�
1/1/2015
guest I should also add that the numbers I have for 2020 is at the end of 2020, not the start. So five years of further growth.
I think they can get to 30% margins on S and X. There is no real competition in the premium segments in the foreseeable future. Mercedes and BMW will both launch new platforms in 2016 and both are focused on ICE and hybrids, not BEV. Same can be said about the asian and the more expensive american cars. No other has Superchargers either. With Model 3 I think competition will be tougher but Tesla should still come out on top.
So where is the price pressure going to come for S and X? With no price pressure and increased production the margins will improve, not go down.
Five years from now though the competition I think will be real. But then it is very possible Tesla can keep its lead for another five years with further 30%+ growth.
Yeah, but I don't think they will make a profit in 2020. Just that it could be thought as 3B out of 30B in revenue. You can also just assume a 13% margin instead of 10%�
1/1/2015
guest Sure enough, when I plugged in your assumptions, I got a valuation per share more like $288 than my $65. So it is all about the sales and profitability assumptions. I just think it is more likely it takes much longer, with more hiccups. Those differences produce a very different valuation.
Good discussion.�
1/1/2015
guest It is possible it will take several more years but I don't think it will. The pessimistic projections I think comes from them always being like a half year late.
After 2016 both S and X will be in full production and stationary pack assembly has started at the GF. What remains after that is Model 3 production at Fremont. Actual ramp of GF cell production is not as important for reaching my 2020 revenue number if they can sell more expensive model 3 2017-2018. It only assumes 200k Model 3's.
To me it feels they have already accomplished more between 2013-2016 than they need to do from 2017-2020.�
1/1/2015
guest I do think there is a risk of poor guidance in Q1 2016. I have nothing to base this on, other than a hunch. I still have a tough time believing the massive ramp of sales that occurred in 4Q 2015. Something smells fishy. We shall see when the company releases its Q4 financials. Then we can see whether there was mass discounting (implying sales to resellers) in Q4. I just don't understand how Q4 sales increased so much more than Q3, Q2, or Q1. Based on nothing other than my gut. But I wonder if they stuffed the channel. I fully admit I could be suffering from confirmation bias.�
1/1/2015
guest Tesla's prediction for max production of the S and X are 50K per year each. I think that's about right unless they lower their prices. They currently dominate the sedan market in that price range and it's not going to grow much.
They predicted production of 400,000 Model 3 by 2020 which is probably a bit optimistic. They may reach that by 2022, but probably not by 2020. They do have some hurdles to get over scaling up from a 50,000 car a year company to a 500,000 car a year company.
- - - Updated - - -
Tesla does not stock many inventory cars, but they always have some cars at their stores and sell them off after using them for test drives for a few weeks. They got aggressive about selling off these cars at the end of last year. They also have a second factory in the Netherlands now that builds cars from parts shipped from California. They had a big spike in sales in Denmark at the end of the year because a major tax incentive was due to expire and the tax on Teslas was going up 180% January 1. I suspect they ran the Netherlands factory full out and burned through all their parts inventory to meet the deadline in Denmark. The Netherlands factory may be out of parts right now and the Fremont factory may be short on some parts too which is contributing to the shutdown in January, though the Fremont factory shutdown is supposedly for some changes to the production lines and maintenance.
I do expect deliveries in Q1 2016 will probably be off, though not catastrophically so.�
1/1/2015
guest Yeah, I am aware of the 50k+50k+400k numbers they have mentioned for 2020. But I think a more likely outcome is 100k+100k+200k. We are already at 50k yearly for Model S, with most of it being in a few countries (US, Norway etc). The guidance for 2016 will help to clarify the current rate. I just don't see the growth stopping for S and X. I think Elon just recently mentioned 150k as a steady state demand for them? ICE and Hybrids are improving but not as fast comparably as Tesla with OTA updates, autopilot and SC expansion�
Không có nhận xét nào:
Đăng nhận xét