It'll decrease, just not fast enough to overcome the cost of throwing away your initial battery. Some of the calculations on battery cost have put it pretty high (can't remember the article now, but it's something like "back of napkin battery calculation cost"). I'm going to invent some numbers for the sake of illustration.
Today:
160 price = $10000 (granted, rolled into initial car purchase, but you are paying for it).
300 price = $30000 (Remember here I'm talking "go buy a new battery", not the $20k option upgrade price)
Let's say you upgrade from 160->300 3 years later and battery costs have dropped 25%:
160 price = $7500
300 price = $22500
So, if you bought the 300 mile straight off, you'd pay an extra $20k more than if you bough the 160.
If you buy the 160 right off, then upgrade to the 300 later you're paying $22.5k to upgrade. The key question then becomes, how much can you sell the old 160 battery for? Remember, it's old technology now and it's had degradation. Can you get enough that it was really worth waiting 3 years to upgrade to the 300?
The more expensive the base battery (say costs were 20/30/40k instead of the 10/20/30k I used), the better it is to upgrade right away rather than wait. If the battery cost curve drops more rapidly over time, say 20% a year rather than 8%, then it swings back the other way.
So, all that was my thinking. From what I can tell with the knowledge available today, it's better to just go big on battery right from the start.