Headed to Santa Barbara service center today to help out with deliveries.
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
-
Tomorrow at 1PM PDT we will be discussing the latest Tesla Shareholder Meeting on the TMC Podcast. You can watch live on YouTube or X and participate in the chat.
- Status
humbaba
sleeping until $7000
You are then not comparing it to dealers, you are saying that not-rushed is better than rushed. That's fine.
But claiming it is relevant for comparison (Tesla service centers to legacy dealerships) is wrong. For one, as was already pointed out, that did not include used car sales which increase the "business" of a dealership. So it is not a metric for how busy a dealership is. It also does not capture the square footage you mention. If Tesla had three service centers that were the size of small towns the metric would be higher and by your take implying they were busier.
In short, it is only useful as a comparative metric if you are comparing similar things. For example, a Ford to GM dealership might work. But not for Tesla to others. They are just too different.
pyromatter
Member
https://electrek.co/2018/12/31/tesla-model-3-inventory-us/
"According to a source familiar with the matter, Tesla had over 3,300 Model 3 vehicles in inventory in the US on Sunday."
"According to a source familiar with the matter, Tesla had over 3,300 Model 3 vehicles in inventory in the US on Sunday."
EVNow
Well-Known Member
Less than I had thought.
But the SP takes a dive, all the same (concurrent with Nasdaq dive, so who knows why).
BTW, how much did Tesla have in inventory last quarter ?
8k M3 in transit at end of Q3, I guess it’s the same/includes inventory.
EVNow
Well-Known Member
So, 5k less than last quarter.
So, even if we conservatively estimate production at 55k, the deliveries would have touched 60k.
My guess is production was between 55k to 60k. So, deliveries will be between 60k and 65k.
ps : In the industry it is common to have 30 to 60 days of inventory. Tesla has just 3 days worth of inventory !
humbaba
sleeping until $7000
...okay, as I'm apparently not alone in being unable to decipher where to post and with all apologies to mods:
--> OT
I recently took a trip which lead to some observations about EAP and FSD. The short version is that this reinforces my belief that FSD is not just around the corner, but that some iterative developments are likely that will provide significant driver assistance. And, indirectly, this does relate to stock price, but I'm going into details that are not market action. This is long, so if not interested use the skip message button.
Please note: I love EAP. I consider it mandatory and it was my second reason to buy a Tesla (first was safety -- lowest risk of injury). Yes, I do realize that these features are beta and not complete. Yes, I do realize that this is without HW3. But there is some enthusiasm about FSD that seems completely unwarranted -- non-FSD features are still beta, and for good reason, and yet there's this belief that somehow FSD is going to be released in the first half of the year, or at any time next year. If it is available outside of a limited, controlled, beta then I expect it be disastrous for Tesla and $TSLA. Anyone who thinks the press was hard on Tesla in 2018 is in for a surprise if there's an FSD-related accident, much less death.
So, to be clear, I'm not talking FSD here. What follows are my observations about EAP beta features, most particularly auto steer, from a 6 hour road trip this last weekend. For reasons that will become clear, auto steer was engaged for less than half of the trip. However, I do want to note that insofar as it worked, it worked well and was a benefit.
Numerous Auto Steer Disengagements
There were many disengagements of auto steering. Not only were there more than I had expected, but I had to disengage auto steer myself in situations that I had not expected.
I believe that the cause of all of this drama was the marking of the center line. It was low frequency and the stripes were extraordinarily short. If there was insufficient marking for it to detect the center line that could certainly explain its inability to maintain auto steer. However, it should have disengaged. The hard left/right swerving is hard to explain as a rational consequence.
Computer Vision
If that were all that would be enough, but in terms of any dreams of autonomy the following short comings were also noted.
I've seen the opinion that a Tesla somehow "sees" better than a human driver due to camera coverage. Unfortunately, that isn't much of an advantage: humans can turn their heads to gain near 360 coverage. Naturally, they can't see all directions at once, but that is rarely a significant issue. What is a significant issue is ability to see stationary objects as well as moving ones. The ability to classify objects. The ability to understand missing objects, like a section of pavement cut out of the road (something that, while rare, I have encountered outside of marked construction).
If you watch a video showing object labeling and lane detection you can see that lines are drawn that actually go up the side of an underpass, apparently due to color blocking. The human eye has much better resolution and contrast than a camera, and the cameras in a Tesla do not appear to be pushing any quality boundaries. Quite the contrary, the video demonstrates (relatively) low grade electronics.
Going to higher processed resolution with HW3, the speed of HW3 to enable better classification, etc., will certainly help. But it is by no means certain that it will just by a HW3 swap to enable FSD. As I'm focused on Tesla here I'm omitting how lidar does not compensate for these shortcomings, but I don't see Tesla's focus on cameras as the problem. Its more general and computer vision does not appear to be within an order of magnitude of where it needs to be.
Maximum Speed
As an implementation flaw, while it isn't a deal breaker the traffic aware cruise control has a significant flaw: it works by attempting to maintain a maximum speed rather than an ideal speed. For example, it is a typical case to accelerate above target speed when passing. I do this whenever I'm passing and faster traffic comes up behind me: its basic consideration for other drivers.
The maximum speed should be set like the TTACC speed offset -- that is, configuring +5mph as a maximum above the desired speed. The speed should then be adjusted for road and traffic conditions, permitting upward speed as well as downward. Another example would be accelerating when in front of a semi after cresting a hill and it is (temporarily) going faster due to the slope. Maintaining speed will result in the semi going to pass, then failing as it can't keep speed up, and consequently increasing congestion.
Even more, both of those speed adjustments should be set as percentages in order to be as broadly applicable as they are used. You can set TTACC at 30mph (and I do it for a 35mph speed zone on my morning commute) but using a +10mph setting would be too fast for a 35mph zone despite being quite acceptable for interstate travel. I kind of get why they do flat add/subtract, but even outside of my extreme 35mph to 70mph comparison it is still better to use proportional increases.
There is also no "perception" of traffic flow outside of the "vehicle in front is slower, suggest to pass." For example, when approaching a slow vehicle (but not close enough to trigger a pass) while being approached from behind there is no suggestion to pull out early to avoid being trapped in the right lane.
Conclusion
The implication from Elon's tweets about Auto Steer was that it was "good to go other than validation" in its current release. Auto Steer needs to be substantially improved before it can be validated by a beta release. We're not even talking autonomy issues, but basic "can it track an interstate lane" and not risk accident by serving back and forth like a drunk.
And if auto steer is this far from being out of beta it is hard to conceive that FSD is anywhere on the horizon. I look forward to incremental improvements, but also hope that they address the fundamental approach of how EAP is intended to work (e.g., maximum speed vs ideal speed).
--> OT
I recently took a trip which lead to some observations about EAP and FSD. The short version is that this reinforces my belief that FSD is not just around the corner, but that some iterative developments are likely that will provide significant driver assistance. And, indirectly, this does relate to stock price, but I'm going into details that are not market action. This is long, so if not interested use the skip message button.
Please note: I love EAP. I consider it mandatory and it was my second reason to buy a Tesla (first was safety -- lowest risk of injury). Yes, I do realize that these features are beta and not complete. Yes, I do realize that this is without HW3. But there is some enthusiasm about FSD that seems completely unwarranted -- non-FSD features are still beta, and for good reason, and yet there's this belief that somehow FSD is going to be released in the first half of the year, or at any time next year. If it is available outside of a limited, controlled, beta then I expect it be disastrous for Tesla and $TSLA. Anyone who thinks the press was hard on Tesla in 2018 is in for a surprise if there's an FSD-related accident, much less death.
So, to be clear, I'm not talking FSD here. What follows are my observations about EAP beta features, most particularly auto steer, from a 6 hour road trip this last weekend. For reasons that will become clear, auto steer was engaged for less than half of the trip. However, I do want to note that insofar as it worked, it worked well and was a benefit.
Numerous Auto Steer Disengagements
There were many disengagements of auto steering. Not only were there more than I had expected, but I had to disengage auto steer myself in situations that I had not expected.
- Missing entrance/exit lane markings. This was the most frequent cause of disengagement and occurred because many entrance/exit lanes did not used dashed lines so EAP was unable to determine lanes. While it can tolerate brief gaps every time there was no marking it would bail close to the end of the ramp. For example, with an entrance ramp the car would drift right in an attempt to maintain center lane positioning and then jerk the wheel left and abort when it gave up.
- Missing exited vehicles. When passing vehicles that took an exit it would remain in the left lane instead of suggesting merging right. This required manual disengagement to correct.
- Losing vehicles. When passing a semi around a curve on a hill the EAP lost track of the semi that was still quite visible and obviously there. This also required manual disengagement to correct.
- Failure to pass. I have it set to "mad max" so it is supposed to suggest lane changes on any speed drop. Usually this worked as expected, but multiple times it would drop speed and follow without suggesting a lane change despite there not being any other traffic. This required manual disengagement to correct.
- Going over heavily patched road. This happened to be a bridge, but I'm confident the reason auto steer disengaged was due to the multi-colored road surface from a mix of different patches. Regardless of cause it was a needless disengagement (just a short piece of roadway in a long straight away).
- Missed lane change notification. The notifications are put in the lower part of the screen which is partially obscured by the right hand being on the wheel. Not responding in a timely fashion apparently results in disengagement. Less likely to happen when accustomed to the system, but it is awkwardly placed.
- "Random" disengagement. Being my first experience I'm not sure as to the cause of the first disengagement. It might have been the sharpness of the curve. I may have inadvertently put too much force on the wheel. It might have been something else.
I believe that the cause of all of this drama was the marking of the center line. It was low frequency and the stripes were extraordinarily short. If there was insufficient marking for it to detect the center line that could certainly explain its inability to maintain auto steer. However, it should have disengaged. The hard left/right swerving is hard to explain as a rational consequence.
Computer Vision
If that were all that would be enough, but in terms of any dreams of autonomy the following short comings were also noted.
- Road debris. This is invisible because it is stationary and would be difficult for computer vision to detect in any case due to small features. But hitting a rock about eight inches across at highway speed would be a bad driving day. Nor is driving over a blown out tire a good idea.
- Stopped vehicles. These are also invisible due to being stationary, but the law requires using the left lane.
- Merging traffic. I avoided a dangerous situation because I knew better than to initiate a pass of a semi when I saw that another semi was heading down the entrance ramp at speed and the one on the roadway was going to want to pull left to allow him on.
I've seen the opinion that a Tesla somehow "sees" better than a human driver due to camera coverage. Unfortunately, that isn't much of an advantage: humans can turn their heads to gain near 360 coverage. Naturally, they can't see all directions at once, but that is rarely a significant issue. What is a significant issue is ability to see stationary objects as well as moving ones. The ability to classify objects. The ability to understand missing objects, like a section of pavement cut out of the road (something that, while rare, I have encountered outside of marked construction).
If you watch a video showing object labeling and lane detection you can see that lines are drawn that actually go up the side of an underpass, apparently due to color blocking. The human eye has much better resolution and contrast than a camera, and the cameras in a Tesla do not appear to be pushing any quality boundaries. Quite the contrary, the video demonstrates (relatively) low grade electronics.
Going to higher processed resolution with HW3, the speed of HW3 to enable better classification, etc., will certainly help. But it is by no means certain that it will just by a HW3 swap to enable FSD. As I'm focused on Tesla here I'm omitting how lidar does not compensate for these shortcomings, but I don't see Tesla's focus on cameras as the problem. Its more general and computer vision does not appear to be within an order of magnitude of where it needs to be.
Maximum Speed
As an implementation flaw, while it isn't a deal breaker the traffic aware cruise control has a significant flaw: it works by attempting to maintain a maximum speed rather than an ideal speed. For example, it is a typical case to accelerate above target speed when passing. I do this whenever I'm passing and faster traffic comes up behind me: its basic consideration for other drivers.
The maximum speed should be set like the TTACC speed offset -- that is, configuring +5mph as a maximum above the desired speed. The speed should then be adjusted for road and traffic conditions, permitting upward speed as well as downward. Another example would be accelerating when in front of a semi after cresting a hill and it is (temporarily) going faster due to the slope. Maintaining speed will result in the semi going to pass, then failing as it can't keep speed up, and consequently increasing congestion.
Even more, both of those speed adjustments should be set as percentages in order to be as broadly applicable as they are used. You can set TTACC at 30mph (and I do it for a 35mph speed zone on my morning commute) but using a +10mph setting would be too fast for a 35mph zone despite being quite acceptable for interstate travel. I kind of get why they do flat add/subtract, but even outside of my extreme 35mph to 70mph comparison it is still better to use proportional increases.
There is also no "perception" of traffic flow outside of the "vehicle in front is slower, suggest to pass." For example, when approaching a slow vehicle (but not close enough to trigger a pass) while being approached from behind there is no suggestion to pull out early to avoid being trapped in the right lane.
Conclusion
The implication from Elon's tweets about Auto Steer was that it was "good to go other than validation" in its current release. Auto Steer needs to be substantially improved before it can be validated by a beta release. We're not even talking autonomy issues, but basic "can it track an interstate lane" and not risk accident by serving back and forth like a drunk.
And if auto steer is this far from being out of beta it is hard to conceive that FSD is anywhere on the horizon. I look forward to incremental improvements, but also hope that they address the fundamental approach of how EAP is intended to work (e.g., maximum speed vs ideal speed).
sundaymorning
Active Member
Whether Troy’s estimates are accurate or not, I’m not a bit worried. Just sitting in my chair waiting for official numbers. This feels much better than any other quarter I’ve been through. Results will be the same regardless, we’re going to be profitable.
As I pointed out, given that there are also used car dealerships, the market multiplier average including used vehicles is less than 3.5. So, even if you include used car sales, then based on reports from people who've taken delivery both during the week and at weekends, the delivery rate for Tesla is much higher than the average per dealership.
Given that Tesla center lot sizes have not been noted as being especially large, and given that it has used former dealerships, I think it's reasonable to track the number of sales per location as a source of profitability and/or competitivity.
Short term, I think the SP will be determined by the Q1 deliveries or projection. Anyone one has a clue what Q1 number could be?
I can do that really quickly--your problem is that the car has an engine.
Anecdotally, I know three people who took delivery in Q4. One was home delivery, two were at centers.
EVNow
Well-Known Member
Nasdaq is trending lower - and market is jittery about Q4 P&D numbers. There have been contradictory signals about the numbers.
O I see so people are still stupid and think there is no demand...or demand has "peaked"
Well I smell a buying opportunity.
Zhelko Dimic
Careful bull
What do you do when 100% of your portfolio are your core shares? And any new money you can bring is less than mouse nuts? And swapping to leaps is not possible without market time gap, as your broker enforces shares/options settlement date difference, i.e. will not let you sell shares, buy options on the same day?
It's a rhetorical question, just so you consider that your advice doesn't work for some people that are fully invested
Hmmm. A surfer and a salmon. I'm going to see your biography at the cinema next week.
The shorts need a story to drive price down - the going forward demand theme seems like one of the only stories they have left with how well Tesla did in Q3 and will likely do in Q4. Based on what I am seeing and hearing where I live and with my friends in other parts of the country, seems like a false narrative to me.
- Status
Similar threads
- Locked
- Locked
- Locked
- Poll
