The Next 6 Months for Tesla Batteries, Self-Driving AI and Factories

This article will start with a financial comparison of Amazon in 2017 and 2018 with Tesla today and a financial projection for the third quarter and the fourth quarter.

The article will also briefly review expectations and financial impact of battery day, the self-driving update due in 6-10 weeks and the gigafactories.

Amazon had small Tesla scale profitability or net losses up until mid-2015.

In second quarter Tesla financial report, Tesla indicated that their factories already had 690,000 car per year capacity. Full capacity production would be 172,000 cars per quarter. This would be 122,000 cars from Fremont and 50,000 cars from Shanghai. 105,000 cars from Fremont and 35,000 cars from Shanghai seems to be very achievable for Q3. The actual figure will be announced within 40 days, around October 1, 2020.

Tesla has had 2 quarters of auto gross margin at 25.5 percent. Going from 90,891 cars delivered to 140,000 would be 54 percent increase over the second quarter. This would be scaling the Tesla auto revenue from 5.18 billion to $7.98 billion. Maintaining 25.5 percent margin means an added $714 million in quarterly revenue.

Capacity was being expanded at Fremont by another 100,000 cars per year and would be done by the end of the year. The Shanghai factory is also undergoing various expansions. Hitting the full Q2 available capacity 172,000 cars and 18,000 more cars for the fourth quarter would be 190,000 cars for the fourth quarter. This would be another $2.8 billion in quarterly revenue.

Tesla is releasing an update to the full self driving autopilot within limited release in 6 to 10 weeks. During the fourth quarter, if this update is widely distributed then it could justify an accounting recognition move from 50 percent to 75 percent of the $8000. 25% of Tesla buyers have full self driving autopilot. This would be another $500 million in net income for the fourth quarter.

$3-4 billion in trailing 12 month net income would compare to where Amazon was at the end of 2017 or in the first quarter of 2018.


Battery Day

Tesla has an online invitation to the 2020 Shareholder meeting and battery day event on Sep 22, 2020.

The graphic appears to be nanowires.

A 100% silicon nanowire battery anode was developed by Amprius Technologies to provide up to 50% improvement in energy density compared to standard lithium-ion batteries.

The biggest financial impact would be for the battery technology or business deal that will enable Tesla to increase battery production.

The Maxwell Dry Battery Cell technology was reported to be able to made 16 times less factory space. The same size battery factory would in theory be able to produce 16 times the batteries.

Full Self Driving Autopilot

Elon Musk has reported already that the latest rewrite of the Full Self Driving Autopilot allows a commuter to drive from home to office and back with rare human interventions. Increasing the capability by from 50% to 75% has immediate financial revenue recognition benefit.


Billionaire Ron Baron has about $2 billion of Tesla stock. This assumes he did not sell any of his original position. He has access to face to face talks with Elon Musk. Ron Baron stated in a CNBC interview that Tesla’s car capacity factory construction cost had fallen from $15,000 per car per year for Fremont to $4,000 per car per year in Shanghai. This means a 500,000 car per year Fremont cost $7.5 billion while a 500,000 car per year Shanghai costs $2 billion.

The Berlin factory is being built at a faster pace than the Shanghai factory.

Tesla being able to build 500,000 car per year factories for $2 billion with completions in less than 1 year would mean the 8 billion in cash they have on hand could be converted into 2 million car per year capacity. It also means that 200,000 cars per quarter would generate the cash for another 500,000 car per year factory.

Not Long Term Predictions

These were not long term predictions. These simple forecasts mostly based on financial reports will be shown to be right or wrong within 6 months.

SOURCES- Tesla, Amprius
Written By Brian Wang, (author owns shares of Tesla)

37 thoughts on “The Next 6 Months for Tesla Batteries, Self-Driving AI and Factories”

  1. The nitrogen has gained +1 charge over then carbon. The emitted beta electron balances the charge equation out. See? Physics is pretty conssitent, as sciences go.

  2. Shame most of their claims have been fairly debunked. Quick youtube will give you more information. Their marketing has so many scam smells its unbelieveable. I use scam smells as they could still turn out to be true but it stinks at this time. (low data and huge claims that would sell themselves if true). If this product was as good as they make out on their website, when they published their research in a verifiable way, they would have to have an investment team to fend off investors trying to give them money.

  3. My problem with their device is that when I estimated the total energy (in beta radiation) that is emitted from 1 kg of the C14 material (assuming they get it to 100% pure), I got a couple of Watts. And still the engineers claim that they will be able to provide continuous power for phones, laptops and cars? I'm missing something, clearly.

    Also, the decay is into a neutral nitrogen atom + beta particla + anti neutrino. I don't see a positive charge anywhere..?

  4. Beta emission radio-isotope generators, also called betavoltaics if you are doing a google search, have been demonstrated quite successfully since the 1970s*.

    They have even been used in actual commercial products such as heart pacemakers.

    You are correct in that the production of electrons is no good if you don't have somewhere for the electrons to go. Fortunately, every time you emit electrons from a component, even through radioactive decay, you now have a matching positive charge left behind. So that becomes the other end of your circuit.

    *I have a vague recollection of seeing some stuff on betavoltaics from World War 2. I pretty sure that if they'd actually got them working that this was "in the lab" and not actually in use. I'm not going to try to find it again.

  5. …plus… Making diamonds with CVD process is not cheap. So supposed it will cost 1 million USD per kg, is it still worth it? I would say no, even if the battery would last 28 000 years…

  6. I've seen the video and read the article. It's not clear if they have even demonstrated the effect. Also, I'm actually not certain how it's supposed to work.. Presumably, you get a lot of surplus electrons.. But you have to let these electrons "go" somewhere to make an electrical circuit? So, is this to the ground with a cable?

  7. If they managed to get double the energy density (kWh/kg) of their current batteries, they could achieve the same range with a lot less batteries in their cars. There is an interesting feedback loop between energy density by mass and range. The higher the energy density, the less battery mass you need for the same amount of energy storage. That reduces vehicle mass quite dramatically because the batteries are so heavy (about 30% for the Model3). With reduced vehicle mass comes greater range essentially for free. So you get about 15% more range maybe, which would approach the range of the Long Range Performance Variant for free. Or you increase the range and the battery life by increasing capacity for the same weight. Or you further reduce the battery capacity to leave the range the same and reduce cost. Either way, it is a win-win.

  8. No way would a radioisotope battery ever, EVER be approved for use in consumer products in our society. Even if (or arguably BECAUSE) it was completely safe.

    I am still suspicious that such products might (may already be) offered as a super secret, black market, big dollar upgrade for the 0.01%. If you are someone whom could miss out on a million dollar deal if your phone went flat, or whose time is worth $5000/hour when you have to recharge your Tesla, it might be very tempting to have a little extra in your battery pack to ensure it’s always kept topped up.

  9. A number of battery chemistries project greater than kwh/kg specific energy if taken to their theoretical limits.

    Obviously taking them to their theoretical limits is not easy.

    But the current rate of about 10% improvement per year would get us there in a couple of decades (depending on your defninitions)

  10. Things to explain to someone from the 20th century: In 2020 it would be ridiculous to say that a car manufacturer could be as large a business as a bookshop.

  11. Algorithm for `primeness`?

    Is n prime?
    Prime = TRUE ## Establish original hypothesis;
    For i = 2 to n/2
    IF n/i = INTEGER(n/i): Prime = FALSE
    next i
    If Prime
    Print: N “is prime”;

    Hope that Helps

  12. that makes no sense .. their is no algorithm for `primness` that equates to a snowy day and excited kids, you can`t question your black box to know if its right .. Health and Safety are very strict its not a game or maths its kids lives from their perspective. They like you want data and facts not hype. Brains have built in physics I said not driving but calculation of speed and trajectory is physics and brains do this and all other such calculations very very accurately, as you would expect from a tree climbing mammal, our social species brain gives us theory of mind and other drivers minds its not a tabla risa.
    1.1 You like to skip the hype but go to one tweet on pyramids, Musk is largely accurate when he speaks and here he was not joking. In effect you are saying you focus on facts.. what facts? on this issue you have provided none / zero just your opinion, at least I quoted a man building the bloody things and speaking to regulators.
    2 Culture is only vague if you have not looked into how Tesla operates compared to other companies if you have you understand, if you have not then you should look deeper at them.
    3 Climate change is becoming very very political and the petrol car is just one object in the firing line .. saying a ICE engine in any sentence ignores that political fact.
    3.1 .. Sorry again nonsense .. roads did cost trillions and they are there in place outside your front door and every bloody door .. no way is anyone building a electric tram system..

  13. 4-5) Again maybe, FSD taxi might dominate the market, but again it depends on several factors and it might by a shrinking market (due to mobility alternatives, telepresnece and what stated earlier). I wish Tesla best of luck, I simply think its success is not assured as Brian here seems to imply

  14. 1) While agree some physics come built in our brain I am pretty sure we have no innate sense of driving and while 16 years seem many is just 150k hrs (considering never sleeping and being fully aware of your surroundings 24/7) not billions of hours. NN can be taught I agree, but different architectures need very different amount of training and supervision. That can make the difference. Furthermore the amount of real driving required by regulators will vary depending on the quality of the solution to the problem: if you claim you have a black box that produces only prime numbers, but you don’t know exactly how, we might need to test a lot of them before accepting your claim, but if you found the formula/algorithm to “primeness” we might just need to verify that the formula is correct. This is true for automated driving.
    1.1) I did not say Elon always lie, I personally prefer to skip the hype (and the crazy stuff) and focus on the facts.
    2) Maybe, culture is quite vague.
    3) What ignores the environmental push? The laws of physics? Or the fact that electric cars does not necessarily mean batteries on the cars?
    3.1) Roads in every city are already trillions in infrastructure and they are already maintained spending billions every year: what I am saying is that if nation will push to go full electric the infrastructure can change and again batteries might not necessarily be the best solution. Even paved streets were a big investment and yet the transition happened fairly quickly

    1. We don`t in the sense you mean humans are not pure NN they come preloaded with a hell of a lot of Physics algorithms and Social knowledge .. honed by at least 16 years of training before they drive. Plus I gave that NNs could be taught

    1.1 . Oh well thats sorted Elon Musk tweeted on Pyramids, therefore he always lies, therefore Health and Safety and every other regulatory body .. won`t ask real world driving it will just accept that you are right that your vehicle is safe to drive in snow past a kids school in the dark .. so thats not a issue. Although I do wonder Musk was not hyping he was dampening saying even if Tesla does get FSD this year they have to jump through even bigger hoop’s to get it approved .. ah sorry I forgot can`t trust a bloody word he says.

    2.Its more than infrastructure its culture and Teslas is strong.

    3 Which ignores the environmental push, business does not sit in isolation from society, petrol is seen as bad and electric clean thats the perception

    3.1 .. nonsense the roads are built in every city to every business and home, you are talking Trillions in infrastructure

    4 That is Teslas business model software and production techniques combining high profit per vehicle but cheap you can look this up .. Ark Invests non FSD bull case covers this if you are interested

    5 Perhaps but whoever can do FSD taxis will have a major market

    still time will tell

  15. Tho’ phones were raised, rightly … also cars and planes: the nearly perfect fuel for either is hydrogen, which due to some evil joke of God, remains stubbornly a gas even when ridiculously refrigerated. And again, things start to scale mass-wise as compression is increased.  

    I keep saying that it is important to look at the WHOLE energy-storage-and-conversion-to-motive-power subsystem to see its meaningful limits, gotchas, and oh-dâhmns.  

    ⋅-=≡ GoatGuy ✓ ≡=-⋅

  16. 4)     I agree with you that Tesla is exactly like Apple. Steve Jobs managed to move the perception of computers and phones from something work related to something cool to have. But I would like to point out that companies can be measured also in terms of sales rather than collective market opinion. In that case Apple is behind Samsung and Apple marked evaluation is mainly due to the fact that they can net 400$ of profit on every phone they sell making the Iphone a relatively cheap luxury product. In general this kind of things do not scale up well in the car industry, with a wealthy minority spending way more (buying Ferrari, Lamborgini and so on) and the majority being more price conscious.
    5)     As I stated it is not clear how pandemics will affect mobility, but it might be very well likely that in the cities people might start to avoid taxis and robotaxis. If batteries become more efficient very lightweight scooter might do the trick in case of reduced requirements for mobility. Outside the cities where the population is not dense health risks might be lower but the economy of robotaxis/fully electric car might not be as good

  17. 3)     Batteries have unavoidable limits due to physics and chemistry, while I agree with DrPat that advantages might outweigh the issue, I think in general the advantages are advantages of the electric motors rather than the batteries (lower mechanical wear, no classic transmission, regenerative breaking). A constant RPM ICE+generator feeding electricity to an electric motor moving the car (with a small battery for regenerative breaking) is probably way better than fully electric.
    3.1) Furthermore even considering the assumption that cities and nations will push to go full electric the cheapest and most efficient solution is no batteries at all with tramway/trolleys power lines and pantographs: if the general assumption is that self driving cars will function as a parallel public transport system nobody will really care if they have batteries or not as nobody really care if the taxi they ride is electric or not or with an arm connecting it to a powerline. Cars with no batteries will be cheaper, with less maintenance and more energy efficient (as they will not need to speed up and slow down the weight of the battery itself) and the technology does not need any specific next gen development.

  18. Hi, I will split my reply in few comments as it is too long for one:
    1)     We know that autonomous drive can be solved by NN without billions of hours of training as every single human being gets their license with much less. NN research demonstrated in recent years that in some cases big data sets are better (GPT3 for example) but in other cases it is more relevant to increase the level of network abstraction (autoencoders that compress and distil the essence of the data). My position on the matter is that is mainly a methodological/architecture approach rather than a dataset approach.
    1.1)  Musk is on the record also for claiming that aliens built the pyramids… so not everything he says is good.
    2)     Amazon (and Goolge) developed big infrastructures that are very big entry barriers for competitors. Then again Amazon and Google provided unique (or almost unique services) that allowed them to conquer a new market space.

  19. The real problem is that a tank-full of high thermodynamic energy fluids, requires a rather substantial mechanical combination of engine, tankage, exhaust flue, and rotational translation devices (transmission) to convert that thermodynamic potential to motive power (‘energy’ if you prefer)

    It is only when a full-power-plant mass-to-power and mass-to-energy comparison is done between competing technologies, that ‘the visual’ can be seen.  

    FOR INSTANCE, start by setting design-goal limits of comparable nature. 

     × 300 miles of range
     × 250,000 km of useful range between major overhauls
    • reasonably peppy performance (0 to 100 km/h in 5 sec);

    Thing though is that the “all in” mass must then be extracted.  

    And we have to also remember (especially so for aircraft) that an additional Δm power plant mass EITHER requires losing payload, or increasing the whole structural mass to handle BOTH the structural mass and the powerplant weight.  

    Ultimately ‘multiplier’ factors of safety, durability, longevity and reliability become the post-comparison normalizers that take a bunch of parts, numbers and physics to real-world vehicles having comparable characteristics, but dissimilar technologies.

    And it doesn’t help one iota to include self-driving with electric power-train article text.

    ⋅-⋅-⋅ Just saying, ⋅-⋅-⋅
    ⋅-=≡ GoatGuy ✓ ≡=-⋅

  20. Fair comment, I was meaning more as in your comment but I was too short and snappy as you write in in more depth, your reply is more nuanced and I agree with you ..
    Yes pound for pound petrol is more energy dense than current batteries and X (we don`t know) amount of time will be needed for parity, if that is even possible.
    The main point we agree on, the current advantages outweigh any disadvantages .. so in this area EVs (Not Tesla alone) have the advantage which will continue to grow.

  21. Whr/$, and sometimes Whr/kg are useful metrics. Whr/L isn’t useful. Hump batteries were byproducts of not designing cars around the battery in the first place.

  22. Seriously was just about to rant about this. Wh/L is only a useful metric for Apple who is squeezing batteries in to headphones.

  23. 3) Batteries have unavoidable physical limits compared to gasoline as energy storage medium, we can pretend this is not the case, but it is.

    3 OK so you say, many experts disagree

    I don’t think they do. I would be surprised if you can come up some experts who disagree with Batteries have unavoidable physical limits compared to gasoline as energy storage medium.

    But this is missing the point. Yes, a tank of liquid hydrocarbons is cheaper and smaller and lighter than the same energy in a battery, even a battery a decade or two more advanced than our own, because of fundamental chemical properties. But the current batteries are good enough that their ADVANTAGES can outweigh the disadvantages for a growing set of consumers and applications.
    It’s not that the claim about physics is wrong, it’s that it isn’t important enough to be a show stopper any more. And the physical problems are minor enough to be overcome with engineering work arounds.

    At least for cars. Aeroplanes are still a work in progress.

  24. Quote
    This is all good and well, but:
    1) Autonomous drive is the kind of AI problem that once is solved for one is solved for everybody capable of that level of investment: the inherent black box nature of Neural Networks and the fact that you can start with a different training set make it very difficult to defend it in court against similar approaches from competitors.
    Musk is on record in person as saying his contact with regulators makes him believe he will need about 6 Billion miles (10 Billion Kilometres) of REAL world date to show the system is safe not in theory but in practice
    My question is
    “How do you see anyone who copies the system obtaining the regulatory real world data, to actually licence it on the worlds roads?”

    2 Its not something just Tesla can do but then not just Amazon can do online shopping, its very easy to see what companies do its very hard to replicate.

    3 OK so you say, many experts disagree

    4 But Apple (whom you don`t mention) did not, computers and phones were very established businesses and yet Apple is worth how much?

    5 and that means? .. less cars? yes but FSD is expected itself to reduce car ownership, less demand will kill the inefficient first as in nature.

  25. This is all good and well, but:
    1) Autonomous drive is the kind of AI problem that once is solved for one is solved for everybody capable of that level of investment: the inherent black box nature of Neural Networks and the fact that you can start with a different training set make it very difficult to defend it in court against similar approaches from competitors.
    2) The most innovative part of Tesla is their approach to the factory which in itself is designed and redeveloped as a product to be continuously optimized: this is a relevant change in perspective compared to the big investment/sporadic retooling approach, but again is not something only Tesla could do.
    3) Batteries have unavoidable physical limits compared to gasoline as energy storage medium, we can pretend this is not the case, but it is.
    4) Electric car sales exist in a preexisting market (the one of car sales) while Amazon (and Google) conquered (almost) completely new market niches that were ready for the taking (targeted internet advertising, and the shipping/logistics required to directly connect the manufacturers and final customer).
    5) Amazon (and similar services) and the risk of pandemics (present and future ones) reduce the need and willingness of physical mobility .


  26. Also, while energy versus weight is most important for electric aircraft, once again smaller size enables optimizing design, think center of gravity, design of the airfoils, etc.

  27. Batteries in early EVs gobbled space…hence limited trunks or rear seating.

    Newer batteries reduce the issue, but you still see designs like GM jamming the batts into a centerline “hump” to provide better legroom (think Chevy Volt).

    Smaller is almost always better. Provides design flexibility. Reduces ancillary packaging size, weight and cost. etc.

  28. For phones, Wh/L is much more important. But for vehicles, I agree that per kg is more important. Though even there, the volume matters to some degree.

  29. Do we care about Wh/L? Now Wh/kg is a big deal, but are there really any applications where the volume of the battery is at all a limiting factor?

Comments are closed.