Considering their track record, you should take this with a basketball size grain of salt.
Ampere GPUs will allegedly feature bigger frame buffers across the board as compared to Turing. They will also run at anywhere between 100-200 MHz higher vs Turing and do it all within an even tighter power budget. Higher-end parts ( RTX 3080 & RTX 3080 Ti ) will allegedly cost slightly less than the RTX 2080 and 2080 Ti.