General Discussion: 2018 Investor Roundtable

[ValueAnalyst]

Obviously, I'm not @jbcarioca but I'll try to answer this great question anyway. Let's start by breaking it down first:

• Drive Units (RWD/AWD/ P):
  At the moment Fremont is most likely QC'ing- and burning-in the front motor (AC induction) in one single production step. For RWD/P they not only have to do the same for the back motor (PPM) but also sort them by highest output. Overachievers will receive a second burn-in and will go into Ps, the rest will go into RWDs. Installing the front motor appears to be brain-dead simple: In layman's terms, the front drive unit only needs to be attached to the battery (2-4 cables) and the cooling circuit (2 tubes). Done. The only real question is, if they can pull of the installation of the front- and back DU in the same production step. Sourcing the front motors in adequate numbers is a non-problem in my eyes.
  
  
• Black & White interior (P):
  We already know that the B&W interior is supply constrained. No additional complexity should be expected, as this is the identical, premium interior beside the color trim.
  
  
• Wheels & brakes (RWD/AWD/P):
  Literally a matter of installing different sets of wheels and breaks.
  
  
• Carbon spoiler (P):
  Most likely installed by the guys who're installing or QC'ing panel gaps. Should be a matter of seconds. No additional production step required.

Relative to the problems they're still facing with LR+PUP, I'll rate those at

• Overall (adding multiple options & trims):
  • Difficulty: +1
  • Productivity: -0.25 to -1 (at least in the first few quarters)
• Performance:
  • Difficulty: +0.5 to +1
• RWD:
  • Difficulty: +0 to +0.25

Thank you for the detailed response; much appreciated. That's probably a bit less complexity added to the production line than I would have thought based on @generalenthu's comments this morning. I'm looking forward to any other responses. Thanks again.

 

[ValueAnalyst]

I agree w/your line of thinking except one thing I pondered yesterday. I believe that when Model 3 was designed, Tesla really believed they would be farther along with Full Self Driving. Many of the design considerations are obvious to me in this context. And FSD adds a lot of margin on top of a purchase. As it is, FSD is behind schedule and therefore, uptake on FSD can't be at the level that Tesla originally envisioned it being at. This also makes me wonder if this is why the other options aren't less expensive like Elon originally posited. For example, I was expecting AWD to be a bit less than $5K (like $3-4K) since Elon had said that Model 3 options would be less than their corresponding pricing on Model S. That hasn't been the case, as if they are trying to claw back more margin where people will pay for it.

Although I thought about the same, it's not clear if Elon incorporated FSD uptake in his 25% original margin estimate, which was stressed as late as November of 2017 as the target after reaching 5,000 weekly rate. Elon knew by then that FSD development was lagging.

I think that the primary reason for high option pricing is simply higher than expected demand for Model 3. Note that ASP is ~$55,000 1H18, or $13,000 higher than his original $42,000 estimate, and will be even higher in 2H18, whereas FSD at purchase costs $3,000. Big difference.

 

[mongo]

At the moment Fremont is most likely QC'ing- and burning-in the front motor (AC induction) in one single production step. For RWD/P they not only have to do the same for the back motor (PPM) but also sort them by highest output. Overachievers will receive a second burn-in and will go into Ps, the rest will go into RWDs.

Drive units come from GF1. They should be QCed, burned in, and binned (different part numbers) before arriving at Fremont.

 

[Anstandswauwau]

Drive units come from GF1. They should be QCed, burned in, and binned (different part numbers) before arriving at Fremont.

Yep, that's most likely the case.

Thanks, I've updated my post.

 

[Intl Professor]

I fear if all it contained were electrons, it would not contain the electrons very long...

And thank God they are easily seduced. They commit infidelity, like butterflies flitting from one flower to the next while participating in another species' reproduction. Nature is so economical—like poetry.

 

[Intl Professor]

Bearing in mind that there is no meaning without relationships. Without motion there is no energy. Without energy there is nothing. Our minds are quantum magicians. We can create relationships from nothing.

 

[avoigt]

I always did know that the Model S is a submarine car.

10 days under water and it still works. Thats is just hilarious ....

Tesla rebuilder discovers Model S still functional after being underwater

 

[DragonWatch]

FUDar Tesla's Model 3 isn't really for the mass market. It's a luxury car

If average Joe or Joe Jeep (old military jargon), as I would call him/her, will buy the basic model because they can, and the ride will exceed any other make or model out there. Dollars and cents the mid-level Model 3 would have provided everything that we needed or wanted in a sedan. Then we could have waited for the pickup without skipping a beat, in other words, we would have kept our Tacoma until the Tesla truck was available. With the Model X we are down to one car, and I can barrow our old Tacoma from my son-in-law, a win-win deal

By the time Joe Jeep buys his Model 3 all the issues will be resolved and he/she will be happier than any click bait author Oh, and we will be burning fast and furious towards our second 450K Model 3s

 

[Waiting4M3]

I always did know that the Model S is a submarine car.

10 days under water and it still works. Thats is just hilarious ....

Tesla rebuilder discovers Model S still functional after being underwater

Not that big of a surprise actually.Comparing with an ICE car, it's much easier to seal a BEV air tight since it doesn't require any air.

 

[dc_h]

Thanks for the summary. It’s an interesting conclusion. Because of what has happened to YY and other drivers, with autopilot struggling in certain areas to the point of becoming dangerous when it veers out of an existing lane if there is an exit ramp or if the lane appears to widen, I would advise friends and family to generally just use smart cruise. I think many of them would be too trusting and not on guard for the limitations. I love Autopilot and use it daily. I too have been occasionally surprised when the car veered unexpectedly, causing me to take over. Each time I have wondered what would have happened if my attention had not been fully on the road. If for a split second you are looking at something toward the side of the car or in the car, you could be caught off guard and not prepared to correct the car instantly. It takes bad luck to have it happen just at that moment when you are glancing away but it obviously can and does happen. Even with smart cruise, people have to be alert for the corner case of a stopped vehicle on the roadway, especially when the car you are following changes lanes. I believe we are going to see hundreds more of these crashes over the next couple of years until Tesla solves the limitations. I still want the option to use it now even with the limitations. I’m speculating, but I suspect even YY would continue to use it in its current form despite having a serious crash while using it.

I think Elon gets this as well. Overall safety is improved, but few will report the saves few will report the saves and all fails will be shared vigorously. If autopilot is curt twice as safe numbers can be pared. Even at 10 times the sophistry will only need to be a bit more sophisticated to explain how even one accident per billion miles is inexcusable.
I think they could do it h cross country trip now, but it will make the temptation to check out worse and cause more accidents. The transition window for any major tech change is always the hardest part. Two years and autopilot will have won. Could be supercruise or waymo, but automated driving will be measurably safer than human driving.

 

[Waiting4M3]

<snip>
autopilot struggling in certain areas to the point of becoming dangerous when it veers out of an existing lane if there is an exit ramp or if the lane appears to widen
<snip>
I too have been occasionally surprised when the car veered unexpectedly, causing me to take over.
<snip>

I think a lot of mistrust of AP comes from when AP acts differently from people's expectation. I wonder if some GUI showing what AP is planning to do would help alleviate this mistrust. Maybe an outline of the road and arrow showing which path the car is planning to take. If there is any ambiguity such as at a fork, maybe show 2 arrows with different level of transparency showing which side AP is leaning towards. This would give user some advance warning that the car may go where they don't want the car to go.

Come to think of it, a heads-up display projected onto the windshield overlaying the road could be a cool way to achieve this. Sorry to beat another dead horse.

 

[vigleik]

First, I want to apologise to ValueAnalyst for the tone in some of my previous posts. I will try to be more polite in the future.

Second, I fear that some might have mistaken me for a bear. Nothing could be further from the truth, so let me try to explain. With what seems to me like reasonable assumptions a simple model gives me a market cap of not one but two trillion dollars in ten years. When something seems too good to be true it usually is, so in order to avoid disappointment I force myself to assume that lots of things will continue to go wrong.

Third, I think 45k model 3’s in Q3 is not a bad result. I’m hoping for 60k, but even with 45k Tesla should be close to breakeven. Maybe a small GAAP loss and a small non-GAAP profit, plus substantial positive FCF from operations. And that’s good enough.

 

[MP3Mike]

Not that big of a surprise actually.Comparing with an ICE car, it's much easier to seal a BEV air tight since it doesn't require any air.

But he had just got done pressure washing the interior including the MCU and CID.

 

[jbcarioca]

Obviously, I'm not @jbcarioca but I'll try to answer this great question anyway. Let's start by breaking it down first:

• Drive Units (RWD/AWD/ P):
  At the moment GF1 (thanks @mongo!) is most likely QC'ing- and burning-in the front motor (AC induction) in one single production step. For RWD/P they not only have to do the same for the back motor (PPM) but also sort them by highest output. Overachievers will receive a second burn-in and will go into Ps, the rest will go into RWDs. Installing the front motor appears to be brain-dead simple: In layman's terms, the front drive unit only needs to be attached to the battery (2-4 cables), he cooling circuit (2 tubes) and obviously to the chassis (which should be automated). Done. The only real question is, if they can pull off the installation of the front- and back DU in the same production step. Producing the front motors in adequate numbers is a non-problem in my eyes.
  
  
• Black & White interior (P):
  We already know that the B&W interior is supply constrained. No additional complexity should be expected, as this is the identical, premium interior beside the color trim.
  
  
• Wheels & brakes (RWD/AWD/P):
  Literally a matter of installing different sets of wheels and breaks.
  
  
• Carbon spoiler (P):
  Most likely installed by the guys who are installing mirrors or QC'ing panel gaps. Should be a matter of seconds. No additional production step required.

Relative to the problems they're still facing with LR+PUP, I'll rate those at:

• Overall (adding multiple options & trims):
  • Difficulty: +1
  • Productivity: -0.25 to -1 (at least in the first few quarters)

• Performance:
  • Difficulty: +0.5 to +1

• RWD:
  • Difficulty: +0 to +0.25

I agree with everything you said.
The only complexity is the simple added steps in assembly. So none of the variants are material, Difficulty of 1 sounds reasonable.

The new front motor does need design and production testing so I would give it a 2 for design, a 1 for installation.

Because there are content options with AWD, P and interior I might rate the totality as a 2.

The whole point of waiting was to make certain the processes were all stable prior to adding any complexity at all.

 

[sundaymorning]

$634 million Australian project:

Tesla's Massive Virtual Power Plant in South Australia Roars Back to Life

 

[bambam4171]

Drive units come from GF1. They should be QCed, burned in, and binned (different part numbers) before arriving at Fremont.

I think that is an assumption but not proven. We know rear motors come from GF1, but front? Couldn't that just be the same old Model S/X front motor?

 

[bambam4171]

Thanks for the summary. It’s an interesting conclusion. Because of what has happened to YY and other drivers, with autopilot struggling in certain areas to the point of becoming dangerous when it veers out of an existing lane if there is an exit ramp or if the lane appears to widen, I would advise friends and family to generally just use smart cruise. I think many of them would be too trusting and not on guard for the limitations. I love Autopilot and use it daily. I too have been occasionally surprised when the car veered unexpectedly, causing me to take over. Each time I have wondered what would have happened if my attention had not been fully on the road. If for a split second you are looking at something toward the side of the car or in the car, you could be caught off guard and not prepared to correct the car instantly. It takes bad luck to have it happen just at that moment when you are glancing away but it obviously can and does happen. Even with smart cruise, people have to be alert for the corner case of a stopped vehicle on the roadway, especially when the car you are following changes lanes. I believe we are going to see hundreds more of these crashes over the next couple of years until Tesla solves the limitations. I still want the option to use it now even with the limitations. I’m speculating, but I suspect even YY would continue to use it in its current form despite having a serious crash while using it.

But do we know it was Autopilot? Or could it be a blown tire? Or the rattle in the front suspension he heard before? Or a rock?

 

[Esme Es Mejor]

Anecdotal warning:
The two accidents where I was hit were a driver changing lanes while I was decelerating to enter a parking lot (looking too far ahead) and while stopped at a red light (driver was looking ahead but at the car next to me)

The one time I tagged a truck, I was looking forward, but tired and misjudged the distance to their hitch reciever.

I've also witchesed a bumper tap in stop and go where the driver had to be looking up due to knowing to start moving.

So looking up does not mean looking at the correct spot nor paying attention. It may remove some blatant non looking cases, but if that is the goal I'd say non-AP cars should be forced to have that feature first.

I hear you, and I agree it won’t stop every case of distracted or terrible driving. But it would stop the most egregious cases of looking at one’s phone while driving. That would be a worthwhile benefit.

I also agree with you that Tesla shouldn’t be forced to do this. If a driver insists on driving distracted, they’re no less safe in a Tesla than in any other car. In fact, they’re probably safer. But it’s still a terrible decision, and if Tesla can change their behavior with eye-tracking software, they should. They shouldn’t be forced to do it, but they should do it.

 

[bambam4171]

I also agree with you that Tesla shouldn’t be forced to do this. If a driver insists on driving distracted, they’re no less safe in a Tesla than in any other car. In fact, they’re probably safer. But it’s still a terrible decision, and if Tesla can change their behavior with eye-tracking software, they should. They shouldn’t be forced to do it, but they should do it.

Thing is that they get a false feeling of security doing distracted driving more and more. Just look at XueXue, he said Autopilot has saved his life many times, so what harm can it do to just check your phone? And bang..

 

[mongo]

If a driver insists on driving distracted, they’re no less safe in a Tesla than in any other car. In fact, they’re probably safer. But it’s still a terrible decision, and if Tesla can change their behavior with eye-tracking software, they should. They shouldn’t be forced to do it, but they should do it.

If Teslas are less likely to do something bad while on AP with a distracted driver, wouldn't it follow that they should be one the least in need of distracted driving detection?

I can see the case for requiring industry wide distraction monitoring, (industry wide breathalizers too?), but I don't agree with Tesla specifically being held to a higher standard.

Car safety test: Time how long it takes the car to do something bad after the driver puts car on highest functioning driving mode (cruise, lane assist, TACC, whichever) and then takes hands off the wheel.

My truck has no autosteer, it will go in the ditch in <10 seconds. Wife's car is new with lane assist and adaptive cruise. It might ping pong 3 times before going in the ditch, and the cruise will not stop the car from rear ending the car it front if it goes below 20 MPH or so.

For distraction monitoring, I'd rather see interactive prompts from the cluster/ center display. "Press steering wheel switch twice", this cuts out zombie / zoned out driving. Alternatively, it can provide voice prompts: "Are you awake?, say yes or no". Zero cost (other than SW which is a one time thing). Although, that also has the possible issue of increased driver workload depending on situation. Could link to level of other driver interaction (wheel torque, lane changes).

If they have monitoring, the safest option is what Tesla does: eventually flip on hazards and slow to a stop. The Tesla result is better than other car makers. (No way to abort on most cars, including a Tesla with EAP turned off)

Once FSD is operable, I expect Tesla will make all cars as safe as FSD allows, regardless of options purchased. I.e. it won't drive for you, but it also won't plow into a stopped car. So, even if distracted, the car will be safe.