Computronium it is

There does not seem to be any reason that the super-intelligences of the future would not place an exigency on computronium.

Even if super-intelligences are not strictly goal-oriented (see "Goal - a temporarily useful construct"), they will likely be active in attempting to unravel whatever is not yet understood about the universe, including how to relocate off the Earth prior to the implosion of the sun (luckily there is some time for this thorny problem, about 4-5 billion years).

Super-intelligences could follow three routes to try to understand what is not yet understood about the universe: 1) directly by their own experimenting, experiencing and thinking, 2) indirectly by perceiving the experiences and thinking of others (with varying levels of sentience) living biologically and electronically and 3) by simulation.

To leverage their resources and inject a diversity of new ideas faster, super-intelligences would likely take advantage of Options 2 and 3, creating or otherwise engendering wetware and software lifespaces/simulations.

To generate and assimilate all of the new information, super-intelligences will clearly require AMCAP (as much computronium as possible).