It is a pet peeve of mine that journalists are so (willfully) economically illiterate regarding sunk costs. If the DoD orders a new airplane design, a huge R&D cost precedes serial production. Once that money is spent it is gone.If the first prototype crashes early in the test program, as happened in the F-14 program for example, it can be said that it cost, in current terms, probably billions of dollars. But whether or not the first plane crashed, the sunk costs are gone, and irrelevant to the cost of the crash. The cost of the crash is the schedule time impact of it and the cost of building one more plane, the marginal cost of the product.
But if the DoD considers developing a new plane to replace the one whose development costs are sunk, DoD will claim that it will buy an infinite number of the new model, to make the huge R&D cost seem negligible - while continuing to divide the R&D cost of the old model by the (obviously finite) number of planes it actually has bought. But the difference in reality is that the R&D costs of the old model are sunk, whereas those of the proposed model are money still in the bank (or, realistically, not yet borrowed).
It is perfectly true, in that context, that the marginal cost to Apple of one more iPhone is a lot less than the accounting cost you describe in your post. OTOH the R&D costs of the next new model of iPhone, while not perhaps yet sunk, are an ongoing cost of staying in business because in a competitive market such as tech, standing pat with the old model is tantamount to going out of business. In tech, R&D costs are, realistically, and expense rather than an investment. Not that the IRS would agree . . .
Whereas in the military plane example, continuous improvement of the old model is SOP and development of new models is, by comparison to tech, sporadic.
No, these are not true "sunk costs" because they are made in anticipation of making return. I am not referring to all previously made R&D for all iPhones for the past 7 years prior to the release of the iPhone 6, but rather the R&D made solely for the design and creation of the iPhone 6.
It is indeed appropriate to amortize the unique R&D costs for each model into the expenses attributable for each model phone over the life of that phone. If a company does not do this, there is no way to get a true cost accounting of each product it makes to establish a selling price. Under GAAP, one does a best estimate of the number of anticipated units to be manufactured over a specific period of time and amortizes that R&D over those units for that time. Once that is reached, the rest is fully amortized and that expense is no longer accounted for per unit. Apple does not release what they use for their period in which they account for these costs.
Back in 2003, when many companies were adopting this approach, Apple like many others were being forced to charge a fee for "free" updates instead of just including them under GAAP rules. Apple chose to use two years for that GAAP period. That requirement was later dropped under public pressure and the fact the accounting costs to keep track of the fees were greater than the fees.
I really doubt that Apple uses two years to amortize the unique R&D for its products, but it might be as long as the one year product cycle before they bring out the next iteration of that product.
General R&D for new products not associated with a specific product are a general overhead expense of the company. Apple does report their operating expenses in their Annual 10-K, and their overall R&D budget.
One also accounts for a certain amount of company overhead per unit and applies that to cost per unit. Under non-GAAP accounting principles, smaller companies can just lump everything together and use a general percentage.
This idea that the "cost" of product is only the raw components dumped into a box is far too simplistic.
By the way, "sunk cost" generally refers to costs spent on specific projects that have failed and the money spent trying to get them off the ground will never be recovered, not ones that go on to be successes and DO make money to pay back their initial investments. That money spent on R&D to develop those successful projects IS recoverable and then some in the form of profits over the life of the successful products that were developed with the money spent getting it off the ground. That money is CERTAINLY recoverable and not sunk. One their of the origin of the term "sunk cost" came from the oil industry where nine times out of ten wildcatters drilled dry wells and sunk their investments into the ground for no return, money ill spent for no return; a dry hole in the ground where they sank their money. Another popular theory on the origin of the term is that it came from investors who would put their money in shipping a cargo to the orient in hopes of a more valuable cargo being returned during the age of sail when the ships did not come back having "sunk" on the journey taking their investment with it.