What doesn't really make sense is that the GFLOPS are exactly the same as a GTX 1080TI. The number is too much alike, so it's not accurate.

That 16GB of ram is certainly very very useful at even this current time. VFX renders and especially volumetric simulations need a ton of gpu ram, and since CUDA cores are used across almost all VFX and filmmaking applications the extra ram is excellent. Don't just think the sole purpose of a GPU is gaming, it's a recurring pattern to see people state so.

16GB of RAM would certainly not put a smile on my face at current costs... especially not as it's going to be functionally useless for a couple of years. It gives you a nice idea about how much they plan to charge for this, though.

All the data is speculation, many websites do it. I remember TPU made a Vega page that was completely out out of this world when the actual GPU launched. If you take a closer look at many GPUs you can still see wrong information.