3) Also, the 3kWh difference is really so small that it might not make sense to design two manufacturing processes, it is quite common to have the same hardware in electronics, but reduce capacity on some versions to have a range of products and be able to charge more for the high end version.
4) We know that harder cycling lead to more rapid decline in capacity.
Except in this case Tesla is charging less for the high end version.
Make no mistake, the 7kWh battery is the more capable battery by a lot, with a potential ROI likely far higher than the 10kWh version (I haven't done the math, but we are talking ~2,500 cycles, vs ~500 cycles over a 10 year period. No way that a mere 43% increase in size is able to match the generating capacity of a battery capable of being used 5 times as much).
How to resolve all this? My take is that the products are actually the same, just two different deals to diversify the product line. Both contain 10 kWh, but the 7kWh version is constrained to 7kWh to give margin for more rapid degradation, but still have a 10 year warranty. The spare capacity could be gradualy brought online during the 10 year guarantee period.
With regard to the battery cost:
Hence, the 167 USD/kWh estimate can be disregarded, the 429USD/kWh can be disregarded, and the estimate based on 10kWh for 350 USD is the more accurate one at pack/product level. Also makes it possible to a better pack and cell cost estimate: Assume 25% margin, and you have 260 USD/kWh at pack level, very reasonable, and then assume pack integration and electronics adding 50% above cells (quite high overhead for small amount of kWh) and you arrive at about 175 USD/kWh at cell level, which is also very reasonable.
And finally, add a lower pack integration overhead and slimmer margins for utility and you could well have the above quoted 250 USD/kWh for the utility deal.
Didn't you just say that they are the same packs? Why are you basing your analysis on the higher price? You should be doing your calculations based on $300/kWh using this logic since the 7kWh system is actually nominally 10kWh in your scheme and is just $3,000.
Honestly, the mental gyrations needed to justify Tesla using the same number of cells in the two packs is just unsustainable. The per cell cost is easily the primary cost driver for these systems, and 10kWh system requires a ~43% increase in cell count over a 7kWh system.
That's like $600 in extra costs (assuming Panasonic is delivering these cells at ~$200/kWh) being buried in a product that Tesla is offering at a $500 discount. Where are the manufacturing efficiencies to be gained that can offset costs at that scale? The only time Tesla did something similar was when they canceled the 40kWh version of the Model S and offered the 60kWh battery with a 40kWh software limit. And that was a one time thing done for a known number of cars that would also have required extensive engineering to produce aside from the reduced cell count.
In contrast to that exception, the 85kWh battery coincidentally uses ~42% more cells than the 60kWh battery, so by this logic it also would make sense to have just manufactured the 85kWh battery exclusively.
Fundamentally these products offer ~500 cycles for the 10kWh system and ~2,500 cycles for the 7kWh system. While you can achieve that performance by increasing cell count and reducing depth of nominal discharge, the costs required don't make any sense. And if you do want to do that, why not just have a single product with a switch that changes modes and just charge $3,500? Why use smoke and mirrors just to reduce your revenue?
It makes more sense that there is are physical differences in the power electronics and cooling at a minimum. And the idea that the additional costs associated with manufacturing packs using cells with two different chemistries will somehow be more expensive than increasing your cell count by 43% just isn't something I can see as likely.