One word of caution on the cost of TE. Although we know the pack cost on the cars is below $190. But the cells are all imported from Panasonic. Starting Q4 this year, the cell and pack would all be coming from the GF. So it additional to the raw material costs, it will bear the full depreciation of the GF. I forgot how much CapEx is on GF now but I think it would be over 800M by the end of year and maybe 1.2B up till 1H 2017. Assuming they would sell 500M worth of TE in 1H 2017, that's about 1.25 GWh with $400/kWh of selling price. If assume depreciation of that 1.2B asset is over 10 years like Tesla does with most other asset. That would make additional cost of $200/kWh, this is just from depreciation of GF, not including labor, raw material etc. So before we see tremendous ramp up of TE, I think a 25% premium cost over battery pack in cars is way underestimating it. I think this is also why they said "even early phase, TE had positive margin" in Q1 ER.
But for the record, I may doing this depreciation wrong too.
I'm looking at depreciation a bit differently. We can't be sure how Tesla will depreciate the Gigafactory (unless I've missed them telling us), but a pretty obvious method for the Gigafactory would be on a per unit basis. See this example:
Cost: $5,000,000,000
Annual Production (GWh) 35
Annual Production (kWh) 35,000,000
Life of Factory 10
Total kWh Production 350,000,000
Per unit depreciation $14.29
You can tinker with the formula all you want, but even assuming a life of the factory of only 5 years and reducing annual production to 20 GWh (pretty low assumptions), it only raises the per unit cost to $50 kWh. Of course, annual production could be much higher and that would lower costs further.
Tesla alternatively could depreciate simply on the basis of time (as you've assumed). In fact, that is how Tesla depreciates its existing property and equipment (30 years for the property and 3-12 years for the equipment). I think here's how that would work (using your example) -- I think you may have an error in your calculation, but I admit that I might too.
Cost: $1.2 billion
Per Half Depreciation (10 years at $30 million/quarter): $60 million
2017 1H Production: 1,250,000 kWh
Depreciation per unit: $48
Either way, depreciation is not that big of a deal and completely manageable. If they choose straight line, you can see they are keeping prices so high for the time being. As they ramp, they automatically cut depreciation/unit and then they can lower prices by the same amount of decreased depreciation and even raise their profit margin.
The above per unit example is still helpful, as it helps show how depreciation works on a per unit basis and therefore the outer limits of how much extra pricing power they can find. If Tesla decides to do straight-line depreciation, the good news is that later years will have hardly any depreciation on per unit basis and as a result profitability would be much higher.
(Obviously, estimates of profitability of TE can matter a whole lot in the short term, so I'm glad we're keeping this conversation going).