Increasing the median internal energy gradients in any such vastly distributed information and energy-processing (i.e. computational) system is bound to have consequences. Complexity and self-referential combinatorial structure of sufficient sophistication is likely to acquire new and perhaps relatively stable phases.
If energy levels continue rising, then perhaps the “stable” phase state is itself likely to become one of accelerating, adaptive metamorphosis along something resembling predictable trajectories. Uncertainty as to the roles of positive feedback mechanisms and any inverse of dampening homeostasis in systems possessing so many variables and degrees of freedom make successful forecasting into a game of half-blind probability, although coarse-grained simulation may assist to some extent.
One thing I think we can be certain of and as a consequence of logic as much as of physics is that whatever change is now “locked in” will most certainly adopt the dynamical symmetries of a system that seeks sustainably continuous persistence; and all quite regardless of the qualitative desires or existential needs of its human inhabitants.