One way to think about it - if there is 50% dropout (which I think is dramatically too high), and Tesla builds 100k Model 3's in the first full year (2018) of production, then the first year's production is already spoken for. Right now, today, about 24 hours from when the first reservation was taken.
If they then climb to 200k production in 2019, and 400k in 2020, all while producing 100k Model S/X alongside of Model 3, then they meet Tesla's guidance of 500k cars in 2020 (that has mostly been panned as ridiculously over the top).
Meanwhile on the demand side, it looks like if they can ONLY build 100k, 200k, then 400k Model 3, then the first THREE years worth of production will all be spoken for before the first car is delivered.
Or maybe Tesla needs to find a manufacturing plant and start ramping plant #2, and GF #2, approximately today
Another way to think about it. If 35% of cars delivered in the US are 35k and above, and Model 3 (and later derivatives) achieve the same 25% market share that Model S has in it's large luxury sedan segment, then 1/3 of 1/4th sounds to me like 1/12th market share in the US for the Model 3 (and derivatives) when they are 4 or 5 years into delivery. That's more like 1.5M/year demand (1/12th of about 17M vehicles/year), and consistent with 100k/200k/400k/800k/1600k annual deliveries (just for the US). Can Tesla grow annual manufacturing of Model 3 at 100% per year? I suddenly believe the demand supports it.
Oh - and Model S is STILL growing market share in the US. We still don't know what the actual market demand is for Tesla vehicles, at least in the US, because we've never seen the vehicle plateau or drop on an annual basis. Short term fluctuations looks like they have as much to do with incentives expiring, and prioritizing shipments to other markets - not demand limitations.
(Important disclaimer - I think we'll see Model S demand plateau and drop before Model 3 delivers, as most all of the Model S buyers that have never stretched so high for a car, will instead "settle" for a loaded Model 3, and be happy. I believe that Model 3 will prove to be the primary competitor for Model S, and a significant source of substitution for Model S buyers. Meanwhile, Model 3 will be doing what Model S has been doing, and bringing a noticeable number of shoppers from the 10-25k range up into the 35k range.)
I've been suggesting 1M reservations by the time the first Model 3 is delivered. Now I'm thinking I've been grossly low.
I'll be amused if we see the Model 3 competitor market in the US shrink (or grow more slowly than otherwise expected), this year and next, as future buyers in the segment instead decide today to wait. Tesla can start taking market share, in a market they've only shown a concept for