I've been hammered by a work emergency and have been completely out of touch with the discussion. I don't have time to really engage, but the mention of the tweak to cell chemistry is interesting as that was a big unknown. However, based on my research so far, I doubt that minor changes to the chemistry will add more than a few pennies to the cost of the cell, and depending on the tweaks its possible it might reduce costs. The costs at that level are almost purely materials costs because the main variable is the mix you put in the blender on the day you are doing Tesla production.
As to production itself, it looks like Tesla batteries are produced at the big Panasonic fab in China, which had a planned eventual capacity of 600m/units per year for their NCR line of batteries. That means that Tesla represents something like 25% of orders for the main Panasonic product line.
As to the rest of the comments referenced (laptop batteries shorting out) that is all consistent with the design patents I posted in the battery IP thread, which means a battery that has the protective cover removed (and changes are made to the head of the battery to keep it from shorting out) and the cap is drastically simplified. The Model S pack requires that the protective cover be removed, and a conventional "laptop" battery will indeed short out (as I explained in the IP thread) without the other associated changes. Together, these changes reduce both materials costs, and assembly costs.
Anyways, as to why I am here. I had a brief break lunch break and decided to go back to Google to see if there was more current information on battery costs. Instead I found this 2013 study on moving 18650 manufacturing to the U.S. -
http://americanmanufacturing.org/files/1-s2.0-S0378775312018940-main%20%284%29.pdf
If you skip ahead to page 7 you will see a
total per cell cost of ~$1.58 for 18650 cells using NMC chemistry, while being built in China at a large 350m unit fab using fully automated processes. That includes shipping to the U.S. and all costs associated with production except profit.
The Panasonic fab is fully automated and with a capacity of 600m units per year it likely achieves significant economies of scale over the 350m unit fab that this data is modeled on.
Tesla uses NCA chemistry, which depending on the mix of Cobalt you choose for the NMC is either less expensive or more expensive. The worst case NMC assumption (that I could find) for the mix used is this study increase the materials cost that Tesla uses by ~$0.13 when using NCA. The most likely scenario is that the cost difference is more like $0.05.
Just their simplified battery design probably saves more than that, though its more uncertain. Total assembly costs are ~$0.25 in the 350m fab, and by far most of the assembly tasks are associated with building the cap. Every one of those tasks requires capital machinery and labor supervision. So some significant chunk of ~$0.25 is eliminated ($0.05-$0.10?). There might be some savings on materials and scrap as well (simplified cap, no protective cover), but its probably just a penny or two at most.
Bottom line, this data points yet again to the significant likelihood that Tesla is paying less than $2/cell, and even accounting for a profit margin for Panasonic it might well be under $1.70/cell.
If Tesla is paying $1.70/cell, that is ~$152/kWh. The 85kWh pack would have a cell cost of ~$13,546 and probably a total cost of under $16,000.
And again, this data is all quite consistent with a wealth of data posted earlier in this thread, including NCR18650B 3400mAh cells (which should be more expensive than the 3100mAh cells that Tesla uses) being advertised for $1.80 each -
Panasonic 3400mAh cells