So, it seems like a lot of people want to give you an opinion based upon their usage scenario. We don't really have huge volumes or price stabilization yet. As such, the general rules and conclusions apply to these cards. As such, there are 6 usage scenarios, and each of these cards slot into a different one.
1) The 9070 at MSRP is for you if you are budget strapped, will not overclock heavily, and want to explore basic RT.
2) The 9070 at a mark-up but under the MSRP of the XT is for you if you want to clock this up yourself, think you can overcome the binning on the cards, and are willing to end somewhere between the 9070 and the XT's performance with either some choice aesthetics or mild performance improvements.
3) The price of a 9070 being above the MSRP of the XT is a dead zone...where you should not buy
4) The XT at MSRP is you choice for a set it and forget it card with good performance. It's not really designed for a better OC, but it's stable coming out of the box.
5) The XT above MSRP but below street value of the 5070ti is probably the performance sweet spot that will allow good overclocking, underclocking, and not break the bank.
6) An XT above the street price of a 5070ti is probably a non-starter. At that price you maybe purchase a 4080 or similar step up. The 9070 and XT are good....but not infinitely priced good.
My personal goal, an XT with about a $50 upcharge for a much better cooler. The extra price would not be bad, I'd likely slightly underclock the thing to decrease temperatures to about the 85 range at 100% to preserve performance and save energy, and running QHD resolution on two monitors should still be well within the capability of any modern game. It wasn't long ago that a 3080 with 8 Gb was normal. The 9070 XT should be significantly better than that in every way, with double the VRAM. All of that in a package that...if pricing matches what they quoted, would be half or less than what the 3080 sold for most of its lifetime. I'd call that a win for value and performance...if nothing else.
