Why not? What's the limiting factor?
Heat, cost, die size, power consumption, all of those things combined.
A 5700XT reaches about 220w~230w of power consumption, reaches a 100c on the standard cooling solutions from AMD, and it's only 9TF (at it's gaming clock of ~1.75GHz).
imagine adding a 30% more powerful GPU than that, with Ray Tracing and more features on top, and adding that GPU next to a big CPU die operating anywhere between 2.2GHz and 3.5GHz.
We are talking about a GPU that is quite possibly 280w alone, if we assume linear power scaling with TF scaling (which could be the same or not), or possibly more. And that before we add in the CPU. It will also require a serious cooling solution to cool two GPU + CPU beasts next to each other. Then having to sell all of that at a reasonable 500$ price.
I think we all should be a little more serious and realistic here, Microsoft wording is clear cut to me, old CPU vs new CPU is 4 times, so different architectures are being compared, 8 threads vs 16 threads, low IPC vs massively high IPC, different clocks, yet Microsoft rounded up their total compute power numbers and chose to present it in a metric of 4 times more powerful than Xbox One X.
So it's only logical they do the same with their GPUs, old GPU vs new GPU is also different architecture, different clocks, IPC, and also compute/texturing/geometry capabilities, so it's only logical that Microsoft would round up their math and express it the same way they did CPUs: two times more powerful than Xbox One X.