But Nintendo was involved in the development of the Tegra X1 very early in some way or another despite the Tegra X1 not being designed specifically for Nintendo.the tegra x1 was not originally made for Nintendo, so it's not a good example.
The more advanced a process node is, the more difficult the process of scaling down the size of SRAM (e.g. L1, L2, L3 cache, etc.) becomes due to the laws of physics.A 4x leap seems extremely modest to project unless mobile hardware development has just completely stalled out, not sure why Digital Foundry are so skeptical of it reaching PS4 level then.
Most leaps are like 6x?
In fact, SRAM scaling has practically stopped with the introduction of TSMC's N3 process node family. And that means cramming more hardware components, whilst also increasing the amount of SRAM, and whilst making the die smaller, is becoming exponentially difficult.
And despite Apple having really good engineers, Apple's not immune to the problems with scaling down SRAM with more advanced process nodes. In fact, Apple actually increased the die size of Apple Silicon for iPhones, iPads, and Macs in order to be able to increase the amount of SRAM (e.g. the Apple A14 Bionic's die size is ~88 mm², the Apple A15 Bionic's die size is ~108 mm², the Apple M1's die size is ~120 mm², the Apple M2's die size is ~155 mm², etc.).
So Nintendo's new hardware's probably going to have one of the biggest performance increases with respect to Nintendo hardware thanks to Drake. And I don't expect future hardware after Nintendo releases hardware equipped with Drake to have as big of a performance increase.