But if at $350 AMD gains 10% additional marketshare for releasing a GPU first, or if they release at $270 AMD gains 20% additional marketshare, than clearly the $270 is better. It's more money, and more marketshare.
Actually, unit margin * sales is what matters.
If AMD spends $250 on that GPU and has 10,000 sales with $20 unit profit, they've only earned $200,000. However, 5,000 sales with $100 profit is $500,000 - more than twice as much profit at half the sales volume.
You can bet your gluts that AMD will only *barely* aim for being the price/performance leader until nVidia has something on the market.
Let's assume four total new GPUs to cover all of AMD's new GPU offerings, and they have the following performance equivalents:
P10 - GTX 960 - 5% (~380X)
P10x - GTX 960 + 10% (~380X+)
P11 - 980Ti + 50%
P11x - 980Ti + 65%
Expected pricing (for ideal market position, acceptable sales, and profit):
P10 : $220~$250
P10x: $300~$340
P11 : $650
P11x: $800
AMD will then drop the prices on existing GPUs to help flush inventories (mind you, these are target market prices, not MSRP):
380: $145 (from ~$185)
380X $185 (from ~$225)
390: $220~250 (from ~$300)
390X: $300~$330 (from ~$389)
Nano: $409~$449 (from $$499)
Fury: $459~$479 (from $$529)
Fury X: $529~$550 (from ~$600)
All of the above would eventually be EOL'd, but not right away - AMD would want to fill the market with options and allow AIBs to turn a profit (you have to keep them happy!).
I'd expect AMD to keep production like this for as long as they can, only dropping prices after they needed to do so. Their halo products would draw more people to AMD, and they would be providing what would be considered great performance at a great price.
nVidia will then need to drop the price on their existing GPUs to remain objectively competitive until they have Pascal out.
If AMD's lead on the market is more than a quarter (*doubtful*), then AMD can even program in a price-drop just before nVidia releases Pascal, playing havoc with nVidia's market positioning and targeting plans, while looking like the good guy to most consumers (so long as they do so before news of Pascal's release is widespread... timing is critical).
AMD will probably not do the same with Zen, though. There, they'd be best served completely supplanting the FX line, relegating it to the trash bin. They will not price any lower than the equal performing Intel CPUs, and they will be best served following Intel's market segmentation scheme exactly - people are very accustomed to it. Here, AMD can simply charge lower premiums for SMT and unlocked CPUs, which will win them some much needed public opinion.
EDIT:
I'd also like to add that my pricing structure takes into account possible production and supply issues, which AMD will certainly be considering as well. Here, AMD will likely have a significant advantage over nVidia for some time - as AMD can easily dual-source all of their GPUs, with only a couple WEEKS lead time (by moving certain backup/spare equipment between GloFo/Samsung while producing replacements).