Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Tesla, TSLA & the Investment World: the Perpetual Investors' Roundtable

This site may earn commission on affiliate links.
Most MR were sold at $45k in Q4, so really it’s more like a $1k price reduction on those models QoQ.

Model 3 base production costs were down c.$9k in Q3, likely $2-3k in Q4 and are probably going to be down another $2-3k in Q1. So you can’t compare profit impact to a static Q3 base. Margin will be up in Q4 and up again in Q1.
The easier math is just like this

Sell 100,000 cars and give a 2000$ per car discount that’s 200,000,000$ in LESS top line revenue. Compare it to overall net profits as you will.
 
  • Like
Reactions: lklundin and EVNow
I am not concerned about 2K discount, but still freaking out about EU homologation status.
Do you have any reliable source that indicates there is an actual problem? Tesla has never commented on homologation and also never had a delayed EU launch. I still haven't seen anything that leads me to believe this is anything but made up.
 
So, this image library is not a "tested one image, then all others will work." I believe every image needs entered with numerous scales/resolution, viewing angles variations, lighting conditions. Probably hundreds or thousands variations for a single sign. So, if stop sign is recognized ok, it doesn't mean that all other signs will be recognized with the same precision. There was some article about someone from the Tesla FSD team describing how difficult it is to assemble all this data, including different road signs for different countries.

Yes - my guess is that Tesla's training set is probably in the "billion labeled images" size category.

My point is that NNs are fundamentally defined via differences, i.e. "corner cases", and that due to this their "speed of convergence" towards a production quality version is non-intuitive: "too slow" in the initial stages, and "mind-blowing fast" in later stages.

I get the impression that they are in these "later stages" already.
 
Do you have any reliable source that indicates there is an actual problem? Tesla has never commented on homologation and also never had a delayed EU launch. I still haven't seen anything that leads me to believe this is anything but made up.

nope no source, probably it will be fine, just a bit scary.

I think eventually in February first production EU shipment will get the car for homologation. I think it was like that with model S.
 
So, this image library is not a "tested one image, then all others will work." I believe every image needs entered with numerous scales/resolution, viewing angles variations, lighting conditions. Probably hundreds or thousands variations for a single sign. So, if stop sign is recognized ok, it doesn't mean that all other signs will be recognized with the same precision. There was some article about someone from the Tesla FSD team describing how difficult it is to assemble all this data, including different road signs for different countries.

As far as corner cases, I'm thinking, for example, traffic light colors made invisible by a sun glare - recently saw such case when you could see the color only from a certain angle. Snow on the road - we currently have some untreated roads with no visible lane markings where people can drive ok, but I doubt an FSD would know what to do.
A lot of times "wrong way" signs are perpendicular to the road you're on and totally confusing; you can barely figure out they relate to a different road. It just feels like a lot of these need to be hit specifically in testing, not just a single quarter of generic testing.
Not really. Modern deep learning techniques handles the scale and angle issues very well.
The street sign under glare issue is however a 'maybe' concern because if the glare is so strong to overwhelm the CCD. There is physically no way to see the sign. Then again, the algorithm may just handle it like a human does: look for unlit lights or just slow down as a caution.
The edge cases will be endless and Elon and others all acknowledge this. That doesn't mean the software will not be extremely useful.
 
One thing many analysts are not talking about is how the 2000 price cut will effect sales of other EVs and ICEs. Are they also gonna cut their prices with $2k or lose sales? Tesla is using the Amazon strategy of undercutting the competition and winning by volume. With FSD around the corner the strategy could turn out to be very clever.
Well just to be clear there’s not really any undercutting here. At the moment and for the foreseeable future Tesla is the high price leader.
 
The easier math is just like this

Sell 100,000 cars and give a 2000$ per car discount that’s 200,000,000$ in LESS top line revenue. Compare it to overall net profits as you will.

If you run with the assumption that cogs are fixed.

That and the entire market is US domestic.

The only known quantity is ASP is down by $2000 per car.

Any loss in ASP is bad, but how bad needs context.
 
One thing many analysts are not talking about is how the 2000 price cut will effect sales of other EVs and ICEs. Are they also gonna cut their prices with $2k or lose sales? Tesla is using the Amazon strategy of undercutting the competition and winning by volume. With FSD around the corner the strategy could turn out to be very clever.

Exactly. Tesla wants to move units and grow. Gotta keep those lines moving while they wait for the shuttered GM factories to hit the market ;)
 
The move to reduce price is an interesting one:
1. Entice more buyers in US
2. Avoid idle lines and keep improving efficiency
3. Reduce profit margin by a significant amount in the US
4. Help Tesla to reach deeper into the US market
5. Gauge the market response.
6. They can afford to. I am sure the management didn't make this decision without knowing the consequences and circumstances.
7. Don't forget if the US demand is too strong because of this Tesla can always increase the price :D
 
The easier math is just like this

Sell 100,000 cars and give a 2000$ per car discount that’s 200,000,000$ in LESS top line revenue. Compare it to overall net profits as you will.
Overall net profits will be same if they have managed to reduce their production cost by $2K/car, which may have already occurred on the way to the $35K SR version.
 
If you run with the assumption that cogs are fixed.

That and the entire market is US domestic.

The only known quantity is ASP is down by $2000 per car.

Any loss in ASP is bad, but how bad needs context.
Well I think we know that 65k of m3 was NA with probably the majority in USA for q4. My hope is that overall sales will only increase and keep the USA sales at or exceeding q4/18.
 
Tesla knows what its margins are. Their price adjustments will be in line with meeting or exceeding guidance, which is a 20% Model 3 margin for Q4.
I’m sure M3 margins are increasing and that is well known from last ER. Nevertheless, it signifies some small demand saturation at current price points which is why they reduced by $2k and this is new information. It does not change long term thesis, but short term might cause some pressure on stock until earning call.
 
analyzing the $2000 cut is going to drive everyone crazy until they forget about it. One thing I'm sure of, everyone will either 1) forget about it or 2) remember it as an example of something we should have just forgotten about.

Maybe it means less revenue, maybe larger addressable market (so maybe not less rev.), maybe lower demand (but EU China, so prob not), maybe a million other things. It does create uncertainty, though! But certainly, it's not that big of a worry.