It is really a silly thread. Do we celebrate AMDs genius in the CPU section of competing in the mid to low end for the past decade? AMD launched what they had available. No idea why they went with a 480 over something more performance oriented. Only they can answer that.
https://www.techpowerup.com/224149/sk-hynix-to-ship-hbm2-memory-by-q3-2016
AMD announced a long time ago that Vega was designed to be paired with HBM2. That means there is a real technical limitation why Vega isn't even out. Throwing more $$$ wouldn't have resulted in AMD releasing Vega HBM2 earlier since they are at the mercy of SK-Hynix.
The question should be asked instead why AMD didn't foresee such a lengthy HBM2 delay? Probably because when they started the engineering of Vega, it was 3 years ago. It just comes down to the fact that NV anticipated HBM2 issues better than AMD did; as well as NV's memory compression and perf/watt are superior. This means there could be a real technical memory bandwidth bottleneck if the flagship Vega were to be paired only with GDDR5X 10Gbps (all that's available now in volume). Also, any die size and wattage savings are more crucial for AMD than NV since AMD needs a bigger die and more power to compete.
Another reason is that a larger die size has more defects. It's possible that AMD will try to undercut the $649-699 1080 when they launch, but we don't know. Let's say if that is their goal, knowing they cannot get Vega out today even at $699-999 since there is no HBM2, possibly higher yields in the future could let them. Either way, since AMD's roadmap already showed Vega for 2017 before P10/11 even launched, this suggests a real technical limitation, more so than a financial one.
I wouldn't call their strategy genius though because they failed on the execution:
1) Couldn't meet launch demand with RX 480, implying that the competitor will get sales that AMD should have gotten simply because there are no RX 480 cards in stock
2) Couldn't launch RX 460/470 alongside
3) Couldn't achieve the required perf/watt needed to win a lot of mobile GPU design wins.
On paper, AMD's strategy was good by targeting 85% of all PC gamers but P10/11 are nowhere near the class leaders AMD needed to really start taking market share.
One could even argue that by the time Vega launches, a large chunk of high-end gamers who wanted to finally upgrade will have moved on to a 1070/1080. Even if Vega beats those cards, NV will have Big Pascal.
In the past AMD had very solid cards in the $100-350 range. This time 1060 is a technically superior chip, and RX 480's real value are DX12 and price/performance but it is no market leader like the 5850 or the 7850 were in their $250-260 price brackets.
AMD needs to get the $149 RX 470 out ASAP. This card actually has the potential to be a hit since it's much cheaper than the 480 8GB and 1060 and yet still likely very good for 1080p gaming if it comes close to the R9 290's level of performance.