Well, if you want nerdy...
The thing is that the growth curve is exponential at the beginning and since the derivative of an exponential function is still an exponential function. That means that being X weeks late also means growth lower by x weeks. So as long as both graphs, projected and real are in their exponential phase the factor by which the real production graph is lower should not change too much.
View attachment 251596
Sorry for the very cheaply edited photo, it should show a 4 week delay and the consequences on deliveries. Look at the end of September and the end of October. See that the factor between those two doesn't really change too much? That of course changes once the growth stops being exponential and even by the end of October it should be lower than ten, but it will still be quite big. By the end of November it should only be two and in late January it should be the same.
But if they were real exponential graphs the factor should always be the same. Simple example 10^n is always ten times more than 10^n-1, the same is of course true for 1.3895^n and 1.3895^n-7. And if n would be weeks and 1.3895 the growth factor per week, a seven week delay would be the same as always missing projected deliveries by a factor of ten.
Now since the S curve is only nearly exponential to a certain point, Tesla has the chance to catch up.