Digital is gaining a lot of ground, but to make an analogy, it seems to me a bit like the thought of Niccolò Cusano, who said that human knowledge is like a polygon inscribed in a circle; no matter how many angles we add, the polygon will never reach the circle, which represents universal knowledge of everything. This comparison is very beautiful because if you think about it, digital yearns to tend more and more towards the analog, even if its domain of existence makes such a thing impossible. In digital, we find ourselves in a discrete domain by definition; acoustic belongs to the continuous domain. Those who have played on digital for a long time, upon returning to acoustic, will not struggle to notice this: the different sensations under the fingers, the enveloping sound produced by the vibration not of a membrane immersed in a magnetic field, but of a large soundboard that vibrates in its entirety, filling environments and bodies with sound. We do not listen only with our ears, but with our whole body. Filling ourselves with sound gives us a sense of satisfaction; what we derive from digital, we always perceive as 'missing something'.
Digital will continue to grow and probably one day the differences will be difficult to hear. For the moment, we are limited by our technology and by data processing and transport speeds.
The first problem to solve (easily solvable) will be creating a new serial protocol to replace MIDI which, although universally used, has major limitations, such as the 128 representable dynamic levels that cause us to lose some nuances that could be important from a psychoacoustic perspective.
Increasing this number by 7 bits, bringing it even one bit further, would allow this level to reach 256 levels, but we would have major problems with data management involving very large libraries and a substantially heavy workload for the CPU.
We must wait for new architectures, new systems. Today, FPGA architecture is promising, although little work is still being done in the field of libraries. Probably, if the CPUs of future computers do not find significant improvements—given the almost reached atomic limit in the dimensions of tracks on silicon wafers—one will have to opt for dedicated hardware. A sort of stand-alone electronics to be connected to the PC that will only handle processing samples and sending them to the computer's sound card to be played.
There will be many solutions; time will confirm this. The fact remains that, by definition, acoustic will be unreachable, both at the level of mechanics and sound... My modest opinion.