2017 Investor Roundtable:General Discussion

[Drax7]

20 million all-electric cars replace only 1 million barrels per day of oil demand.

The world consumes nearly 100 million barrels of oil per day.

Y'all berate me for predicting Tesla will manufacture 10-15m all-electric cars in 2025.

Yet at the same time say "peak oil in 2022!"

Smh......

Predict whatever you please, though basing investment on the known unknowns
May have negative consequences. Just be prepared for that. The prediction
Business eventually humbles everybody.

 

[MitchJi]

If anyone can help with this, I would be forever grateful: Can Tesla simply add more Model 3 lines in 2018 if, say, July's Final Reveal event boosts reservations to more than 1m?

I believe that their first step will probably be to increase the production capacity (alien dreadnaught technology) of the first line.

It's also relevant to this topic that Elon said that they will be able to build the MY production capacity (including a step change in production) for half the cap-ex of the M3.

 

[ValueAnalyst]

Tesla Network

I've been doing further research on this, and here are some facts/expectations:

WARNING: THE FOLLOWING ESTIMATES AND FUTURE PROJECTIONS WILL ALMOST CERTAINLY PROVE TO BE WRONG. IT IS PRESENTED HERE FOR VERY PRELIMINARY DISCUSSION PURPOSES ONLY.

• New York City Taxi/Uber/Lyft/Others ("TULO") do 800k+ trips per day, which approximates 10% of NYC population.
• This percentage increased by a massive 60% just two years from 6% in mid-2015, which is when Uber and Lyft started to grow rapidly in NYC. This is likely due to (i) the improved customer experience, (ii) higher convenience of point A to point B transportation vs. train/bus/other mass transit, (iii) lower per-mile point A to point B transportation cost due to higher supply, and (iv) declining appetite for car ownership, which is a trend that is just starting to accelerate in urban areas.
• It is reasonable to assume that the number trips taken on TULO as a percentage of population will continue to further increase, quickly approaching 20% and likely higher, for the same reasons I listed in the previous paragraph as autonomous ridesharing launches and proliferates in the coming years, reducing cost of transportation by a factor of 5 or more.
• 4+ billion people currently live in urban areas across the world, and this number is increasing. Given that car ownership as a percentage of population is lower ex-US, I estimate that the number of trips taken on TULO across the world approximates ~500 million trips per day today, but read on.
• Elon noted during the recent TED talk that cost of transportation on ridesharing services will decline below that of bus ticket, which is likely why Tesla scrapped its plans to introduce a Tesla bus.
• The implications of this shift is monumental! According to the American Public Transportation Association, nearly 11 billion trips were taken in 2013 on public transportation in the United States alone. This approximates 30 million trips each day on mass transit in the United States.
• Given that (i) the current US population of ~325 million approximates only 4.33% of the global population, which is expected to exceed 8 billion in 2023, and that (ii) the reliance on mass transit is much higher ex-US, it is reasonable to assume that at least 750+ million daily trips would have been taken on mass transit across the world by 2025, if autonomous ridesharing never took hold.
• I expect autonomous ridesharing to completely replace all mass transit in urban areas due to the former's much higher convenience and future lower cost as well as the fact that government subsidies for mass transit rely on the fact that it is mass transit (i.e. any decline in participation in mass transit will substantially increase the per-person cost of government subsidies and/or the cost of a ticket, accelerating the shift to TULO).
• Combining the above points: (i) ~500 million trips per day currently on TULO, (ii) increasing percentage of TULO-to-population as TULO proliferates and autonomous ridesharing takes hold, (iii) lower cost of point A to point B transportation by a factor of 5 or more, (iv) 4+ billion, and an increasing number of, people living in urban areas with another 4 billion in suburban and rural areas by 2023, (v) 750+ million, and an increasing number of, daily trips on mass transit across the world that will likely shift to TULO in the coming decade, and (vi) the accelerating trend of lower appetite for car ownership, I expect demand for ridesharing services to easily exceed 1 billion trips per day by 2025 and continue to increase thereafter.
• Due to its upcoming fast, reliable, low-cost Supercharger v3 network that will be unmatched anywhere on Earth, its largest and fastest growing fleet that is already equipped with hardware needed for full self-driving capability, its exponentially growing battery manufacturing capacity, its unmatched speed of innovation, and its brand value, I expect Tesla to capture at least one-third of this market by 2025, or ~350 million trips per day.
• For reasons discussed in my earlier posts, I expect Tesla to take 25% rev share with an 80% gross profit margin, for an annual gross profit of $60+ billion in 2025, and growing.

The following summarizes my expectations, assumptions, and calculations:

Opinions welcome.

 

[Hawkeye]

Is that you Tripp?

 

[ValueAnalyst]

Is that you Tripp?

I can neither confirm nor deny that I am Trip Chowdhry.

 

[MitchJi]

How Apple Is Moving in the Driverless Car Race -- The Motley Fool
<Snip>
market for self-driving vehicles will also likely be big enough to move the needle for a company as large as Apple, which has a $750 billion market cap and reported net sales of $215 billion in fiscal year 2016. The Boston Consulting Group estimates partially and fully autonomous vehicle sales could reach $42 billion by 2025 and $77 billion by 2035. And those numbers are just for physical autos -- a potential ride hailing mobility-as-a-service offering could be worth much more.
<Snip>
Apple was late to the self-driving-car game when compared with Alphabet(NASDAQ:GOOG) (NASDAQ:GOOGL) and Tesla (NASDAQ:TSLA). Its self-driving car project, named Project Titan, began in 2014, a full five years after Alphabet's program, and the same year in which Tesla introduced autonomous features in all its cars.

Still, with its vast cash hoard, Apple reportedly hired roughly a thousand engineers and automobile executives to get Project Titan off the ground. Despite that massive investments, Titan reportedly was soon beset with problems including strategy disagreements, supply chain issues, and infighting.

Changing strategy
In mid-2016, the company shifted strategy, replacing former Ford engineer Steve Zadesky with Bob Mansfield, a leader of the original iPad design team, to head Project Titan.

The company quickly shifted strategies away from building an actual vehicle, opting to focus on building the underlying autonomous technology. Bloombergreported in July 2016 that Apple hired Dan Dodge, creator of Blackberry Ltd.'s QNX car operating system, to head an Apple team developing autonomous programs, sensors, software, and simulators. According to Bloomberg, Apple has given the strategy until the end of this year to show meaningful progress.

The strategy is basically a shift away from trying to become Tesla -- as in, producing a full vehicle -- and more toward trying to become Google's Waymo, which currently leads in autonomous technology but (at least for now) partners with existing automakers.

Cook told Bloomberg Television in June that "... we’re focusing on autonomous systems and clearly one purpose of autonomous systems is self-driving cars. There are others. We sort of see it as the mother of all AI projects. ... Autonomy is something that's incredibly exciting for us, but we'll see where it takes us. We're not really saying from a product point of view what we will do, but we are being straightforward that it's a core technology that we view as very important."

The fact that Apple abandoned its hardware-oriented strengths shows just how difficult it is to develop and manage an automotive supply chain -- making Tesla's achievements to date all the more impressive.

Recent testing and developments
In April 2017, Apple attained a permit to test a self-driving system on California roads. The company outfitted three Lexus RX 450h SUVs with its technology, and one was recently spotted driving Highway 101 near Palo Alto. Apple also asked the state of California to further tighten its self-driving test rules, which would make more data publicly available, which could reveal more data to Apple, allowing it to catch up to rivals more quickly.

Don't count Apple out
Some may doubt Apple's ability to catch up to Waymo and Tesla in self-driving vehicles, given its late start and change in direction. However, Apple has historically benefited from what NYU professor Scott Galloway calls Apple's second mover advantage:

They weren't the first into object-oriented computing (the mouse), they weren't the first mp3 player, they weren't the first mobile phone. But they look at something, they improve upon it, they weigh it, and they come in and make it more user friendly.
-- Quote via Business Insider

Will self-driving vehicles continue Apple's second-mover success? It will surely be difficult to outcompete both Alphabet and Tesla's Elon Musk at the same time, not to mention other automakers and ride-hailing services all getting into the self-driving game.

Still, with Apple's brand strength, history of innovation, billions of dollars available for research and development, and second-mover modus operandi, I wouldn't count the iPhone maker out.

 

[callmesam]

There is a resistance and fear of losing personal autonomy. Fear of the unknown. I read it and see it, palpably, in the postings of some otherwise reasonable members.

Tesla Network: this product, which we are supposed to learn more about in July, will likely provide a huge amount of revenue to Tesla's bottom line. Shared Mobility is here. And it works even better when fuel costs 1/4 of gasoline and driver cost is 0.

As a heavy user of shared mobility (Lyft, Uber and long ago, Zipcar) I have a first-hand knowledge of this market. I talk to each and every driver (including in London, Paris last week) about Tesla vehicles and Full Self Driving vehicles. I've used these services for years in a variety of locations in the U.S., including every top 10 U.S. population center. Every driver gets the same questions: Have you heard of Tesla? Have you heard about Full Self Driving capability? How much do you spend on fuel every month?

I agree with your estimate and projections. But there is also an S curve at work with the Tesla Network. At first it will be a blip and then once the switch is flipped and money is promised, each location will reach critical mass, allowing for enough cars to serve each community.

I'm interested in how Tesla protects owners who are hesitant to participate. Tesla will be able to create a strong incentive by requiring riders to certify the previous occupant didn't harm the car. Pictures can be take through the app. Or video sent of damage. Cars can drive themselves to Tesla for service, cleaning, charging. Maybe owners will refer non-owners to the Network. Enough poor comments and fines and users are booted off the Network.

Tesla Network

I've been doing further research on this, and here are some facts/expectations:

WARNING: THE FOLLOWING ESTIMATES AND FUTURE PROJECTIONS WILL ALMOST CERTAINLY PROVE TO BE WRONG. IT IS PRESENTED HERE FOR VERY PRELIMINARY DISCUSSION PURPOSES ONLY.

• New York City Taxi/Uber/Lyft/Others ("TULO") do 800k+ trips per day, which approximates 10% of NYC population.
• This percentage increased by a massive 60% just two years from 6% in mid-2015, which is when Uber and Lyft started to grow rapidly in NYC. This is likely due to (i) the improved customer experience, (ii) higher convenience of point A to point B transportation vs. train/bus/other mass transit, (iii) lower per-mile point A to point B transportation cost due to higher supply, and (iv) declining appetite for car ownership, which is a trend that is just starting to accelerate in urban areas.
• It is reasonable to assume that the number trips taken on TULO as a percentage of population will continue to further increase, quickly approaching 20% and likely higher, for the same reasons I listed in the previous paragraph as autonomous ridesharing launches and proliferates in the coming years, reducing cost of transportation by a factor of 5 or more.
• 4+ billion people currently live in urban areas across the world, and this number is increasing. Given that car ownership as a percentage of population is lower ex-US, I estimate that the number of trips taken on TULO across the world approximates ~500 million trips per day today, but read on.
• Elon noted during the recent TED talk that cost of transportation on ridesharing services will decline below that of bus ticket, which is likely why Tesla scrapped its plans to introduce a Tesla bus.
• The implications of this shift is monumental! According to the American Public Transportation Association, nearly 11 billion trips were taken in 2013 on public transportation in the United States alone. This approximates 30 million trips each day on mass transit in the United States.
• Given that (i) the current US population of ~325 million approximates only 4.33% of the global population, which is expected to exceed 8 billion in 2023, and that (ii) the reliance on mass transit is much higher ex-US, it is reasonable to assume that at least 750+ million daily trips would have been taken on mass transit across the world by 2025, if autonomous ridesharing never took hold.
• I expect autonomous ridesharing to completely replace all mass transit in urban areas due to the former's much higher convenience and future lower cost as well as the fact that government subsidies for mass transit rely on the fact that it is mass transit (i.e. any decline in participation in mass transit will substantially increase the per-person cost of government subsidies and/or the cost of a ticket, accelerating the shift to TULO).
• Combining the above points: (i) ~500 million trips per day currently on TULO, (ii) increasing percentage of TULO-to-population as TULO proliferates and autonomous ridesharing takes hold, (iii) lower cost of point A to point B transportation by a factor of 5 or more, (iv) 4+ billion, and an increasing number of, people living in urban areas with another 4 billion in suburban and rural areas by 2023, (v) 750+ million, and an increasing number of, daily trips on mass transit across the world that will likely shift to TULO in the coming decade, and (vi) the accelerating trend of lower appetite for car ownership, I expect demand for ridesharing services to easily exceed 1 billion trips per day by 2025 and continue to increase thereafter.
• Due to its upcoming fast, reliable, low-cost Supercharger v3 network that will be unmatched anywhere on Earth, its largest and fastest growing fleet that is already equipped with hardware needed for full self-driving capability, its exponentially growing battery manufacturing capacity, its unmatched speed of innovation, and its brand value, I expect Tesla to capture at least one-third of this market by 2025, or ~350 million trips per day.
• For reasons discussed in my earlier posts, I expect Tesla to take 25% rev share with an 80% gross profit margin, for an annual gross profit of $60+ billion in 2025, and growing.

The following summarizes my expectations, assumptions, and calculations:

View attachment 233680

Opinions welcome.

 

[ValueAnalyst]

Can anyone explain why nearly every company testing for autonomous driving is using an SUV?

 

[Starno]

Can anyone explain why nearly every company testing for autonomous driving is using an SUV?

I wonder as well, but I might have a theory :

Once full autonomous driving will be there, most cars won't be owned by individuals, but by companies that will have a fleet of them and use them as taxis. And so they want to be able to transport multiple people at the same time with a lot of room. Thus the SUV.

 

[3Victoria]

More room for people and equipment, both for processing the sensors, data, and driving AI, and for monitoring of same.

 

[22522]

Can anyone explain why nearly every company testing for autonomous driving is using an SUV?

There are several reasons:

1. LIDAR Elevation - particularly at intersections. 42 inches is standard height for cabinets on corners that provide communication services. You have to be able to see over that with some sensor when stopped. It also helps to have the sensor as far forward as possible. A lot of google car accidents were rear endings by other vehicles as they tried to look around corners with their LIDAR by rolling forward a bit. Tesla put cameras in the tops of the B pillars.
2. Safety for the driver.
3. Gross Vehicle Weight Rating. Prototype stuff is sometimes heavy. It helps if the car can carry the weight.

 

[Dwdnjck]

We might also have rolling recessions, for instance healthcare that takes up about 20% of GDP
May suffer a recession if and when market forces are introduced.

Well, at least my holdings in Stonemore will go up..

 

[22522]

View attachment 233680

Opinions welcome.

Thank you.

This is about error bars and driving gradients and maybe the need for a graph (like a stock price graph) that shows events that trigger rapid increases in behavior changes. For example, when Model 3 new car goes away after 3 years, the number of real participants goes up, as someone mentioned.

Also, error bars are traditionally on the Y axis, but for this stuff ambiguity in time (x axis) is important. Some say, take any schedule and multiply by Pi if you want to know when it will happen.

For this stuff, does it make sense to add 3 years and then say +/- 3 years on the x - axis?

 

[Snapdragon III]

Can anyone explain why nearly every company testing for autonomous driving is using an SUV?

My theory is that they have 100-1000's of pounds of computer gear, and batteries jammed in there, and a normal sedan doesn't have room, and can't take the weight. It is one more area that Tesla could be way ahead of the competition that is not obvious to the casual observer.

 

[Alketi]

Tesla Network
Opinions welcome.

Without getting into all your numbers, I see that you're assuming full self-driving (L5) by this time next year...

But it's not just full self driving that must be achieved, it's permission from authorities to go driverless -- which brings with it regulatory approval. The NHTSA has a "hardware redundancy" clause in their guidelines. Tesla has no such redundancy in their vehicles (backup computers, etc). So even if by some miracle Tesla achieved FSD in 12 months, there's the additional requirement of having backup hardware systems. This likely means that no Tesla in existence would qualify for approval.

IMO, it's not remotely realistic given all the observed datapoints that the Tesla network will be up and running a year from now. Musk himself recently admitted that self-driving has been a "tough slog" and just fired the head of his software department. The car still has trouble staying in its lane on a sharp highway turn. Tesla just turned on full data collection in May.

To imagine FSD suddenly appearing means believing that Tesla has a second secret team working on FSD and a primary team having trouble achieving silky-smooth L2 with the same hardware. I believe your estimates are at least a year too early (technology-wise)-- with the added risk of a slow regulatory approval and that only NEW vehicles may someday qualify.

 

[ValueAnalyst]

There's nothing left to reveal as far as I can tell. I think they'll just announce a start date for production, whether it's already started or will start at some time in July.

Without getting into all your numbers, I see that you're assuming full self-driving (L5) by this time next year...

But it's not just full self driving that must be achieved, it's permission from authorities to go driverless -- which brings with it regulatory approval. The NHTSA has a "hardware redundancy" clause in their guidelines. Tesla has no such redundancy in their vehicles (backup computers, etc). So even if by some miracle Tesla achieved FSD in 12 months, there's the additional requirement of having backup hardware systems. This likely means that no Tesla in existence would qualify for approval.

IMO, the Tesla network will not be up and running a year from now. It's not remotely realistic given all the observed datapoints. Musk himself recently admitted that self-driving has been a "tough slog" and just fired the head of his software department. The car still has trouble staying in its lane on a sharp highway turn. Tesla just turned on full data collection in May.

To imagine FSD suddenly appearing means believing that Tesla has a second secret team working on FSD and a primary team having trouble achieving silky-smooth L2 with the same hardware. I believe your estimates are at least a year too early (technology-wise)-- with the added risk of a slow regulatory approval and that only NEW vehicles may someday qualify.

Before anyone else takes off on this (the bolded sentence above), it makes essentially no difference to my intrinsic value estimate if FSD/regulation comes about next year or in 2020/21. I believe most here agree with me that Level 4/5 autonomy (Level 4 would be enough to launch Tesla Network) will be achieved by 2020/21; three years after Elon's estimate of late-2017, which he made just a few weeks ago. And also, before anyone says "but that's only on one route from LA to NY," no, he also said during the TED talk that the software will be able to handle dynamic routing and said "no controls will be touched at any point during the trip."

SO LET'S BE CLEAR: Elon Musk predicts FSD capability (hardware/software) by the end of this year.

I don't think it's a stretch to assume regulation will follow in the following three years at least in some major states in the US. I expect California to be first, followed closely by New York. These two states alone would be enough for my projection until 2020/21.

Having said that, I want to emphasize that in my DCF projection, the FSD-capable fleet and the participation rate assumptions are so low for the earlier years that estimated Tesla Network revenues and cash flows feeding into my financial model are very minimal until 2020/21.

 

[22522]

What bothers me even more is the personal attacks.

If you understand the attacks better, they bother you less.

One person is concerned about the credibility of the group, and outlandish statements destroying the credibility of the group. Self policing has benefits since external correction is a blunt instrument. It is almost always better to self police.

The other person comes across as more of a promoter than analyst. They are showing more of their work, and the more clearly they show how they got their answers, and present 3 different possible scenarios for the reader to choose from, the better things will get. For example: Slow (probable), Medium (maybe), and Fast (motivated/driven).

 

[Alketi]

Before anyone else takes off on this (the bolded sentence), it makes essentially no difference to my intrinsic value estimate if FSD/regulation comes about next year or in 2020.

Then, if it makes no difference to your numbers, start your chart at 2020, and assume only new cars from that point on will achieve approval. It would more than likely reflect reality, if that's the goal of the projection.

As to your numbers, are you assuming $1/mile?

Your 25% revenue split for Tesla is probably ballpark correct.

I'm confused how you get 91 days at 12 hours per day. Is this assuming weekends and something else?

 

[ValueAnalyst]

Then, if it makes no difference to your numbers, start your chart at 2020, and assume only new cars from that point on will achieve approval. It would more than likely reflect reality, if that's the goal of the projection.

I personally think Tesla Network will be launched in 2Q18 so that's what I'm presenting, but I would encourage everyone to run their own DCF analysis.

As to your numbers, are you assuming $1/mile?

I actually haven't converted my model to per-mile basis, but given that current avg Uber fare is $10 and trip length is 7 miles (and increasing), I'm starting my projection with slightly more than $1/mile in 2Q18, but that number is declining precipitously to much less than that in outer years, which is what matters to the intrinsic value calculation for Tesla. Especially the 4Q25 assumptions. I think that the length of an average trip will substantially increase in the coming years and ridesharing will increase so I think $2.50 per trip is a very conservative estimate.

Your 25% revenue split for Tesla is probably ballpark correct.

Great; I'm glad we agree. Note that this is lower than Uber/Lyft's effective cut of more than 30-35% today, and arguably, this percentage should increase as Tesla will effectively be the driver. So 25% I think is a very conservative estimate for a very impactful input.

I'm confused how you get 91 days at 12 hours per day. Is this assuming weekends and something else?

There are ~91 days in a quarter. Weekends and holidays are included.

 

[Starno]

Before anyone else takes off on this (the bolded sentence above), it makes essentially no difference to my intrinsic value estimate if FSD/regulation comes about next year or in 2020/21. I believe most here agree with me that Level 4/5 autonomy (Level 4 would be enough to launch Tesla Network) will be achieved by 2020/21; three years after Elon's estimate of late-2017, which he made just a few weeks ago. And also, before anyone says "but that's only on one route from LA to NY," no, he also said during the TED talk that the software will be able to handle dynamic routing and said "no controls will be touched at any point during the trip."

SO LET'S BE CLEAR: Elon Musk predicts FSD capability by the end of this year.

I don't think it's too much of a stretch to assume regulation will follow in the following three years at least in some major states in the US. I expect California to be first, followed closely by New York. These two states alone would be enough for my projection until 2020/21.

Having said that, I want to emphasize that in my DCF projection, the FSD-capable fleet and the participation rate assumptions are so low for the earlier years that estimated Tesla Network revenues and cash flows feeding into my financial model are very minimal until 2020/21.

I agree however it should be noted that a East Coast - West Coast trip means it will be 95% only highways. FSD on highways is fairly easy as it's basically just straight lines, and slight curves. There is no or very few stop signs, red lights, pedestrians, obligations signs, ... to deal with.

What will be very interesting and a major stepping stone, in my opinion, will be for example the first full day in a major city fully self driving. Like in NYC or Los Angeles. Going almost to every part of town.