Current estimated pack prices for the Model S 85kWh come at around $15,000 ~ $17,500. This price is actually very low for the energy density (~$200/kWh.) Tesla aim for at least 25% margin on every Model S sold, so they've got a large margin to play in.
So it's plausible that we could see a $12k to $15k pack in a few years time. The electric drivetrain -- excluding battery -- is very cheap to make. Only a few thousand dollars. (Motor+Inverter cheaper than all but smallest ICE.) So $25k to $35k with 10% margin appears doable. Obviously, you'd be looking at a rather basic vehicle: probably around 200 to 300 mile range, 150 to 200bhp, and probably no significant fancy touchscreen like in the S, nor sporty performance. Maybe it won't be under the Tesla badge either; they may wish to use that for their luxury vehicles only.
So it will still be a compromise compared to an ICE, but it will likely pay for itself due to the low cost of electricity and free supercharging -- if it remains free for GenIII.
Something else interesting; it was recently announced each swap station costs around $500,000 to build and will have around 50 battery packs in rotation. That makes for less than $10,000 per pack, before you even exclude the costs of construction and machinery -- so how are they going to achieve this feat? Are batteries a lot less expensive than we've been led to believe? Perhaps they're not included in that cost, but it seems strange to not include that.
Apparently, while idle, these packs will be used as local grid storage. If they use half of each pack for grid storage, 43kWH and have 50 packs and 5 swap stations in the first few years, this gives the grid an effective 10.75MWh of energy storage. Useful for solar/wind power, would make it one of the largest such storage stations.