Trade-in values, cars paid in full but not yet delivered and Chinese import duties paid by customer at import time all count toward that number as well and are generally much larger than $2.5/car so they have an outsized influence. Just subtracting the 3 deposits and then dividing by $2.5k to estimate S/X backlog is so wrong it isn't even funny. Currently Tesla sells about 22k cars/quarter. Assuming only 2k being inventory sales (this is a severe underestimation) means that a backlog of 40k comes down to a wait time of 6 months for a new S/X order placed today. That is clearly not the case.
Here is my take. Let's assume M3 orders rose by 30k in Q2 to 500k. That means 'other deposits' dropped $43M this quarter, from $149M to $106M. Assuming energy related deposits remained constant (some powerwalls installed, but some solar roofs ordered) taking maybe $15M of the 'other deposits'. So S/X related deposits dropped from $134M to $90M. I explained before that the number includes things like trade-ins, cars paid in full not yet received and chinese import duties. But those all scale more or less together with the overall backlog (if the backlog is bigger, more people are doing a trade-in etc). So it is right to say that S/X backlog dropped by a third in Q2. The average wait times seem to confirm that as well.
