Since I’m utterly unqualified to write on this topic, the following should be treated as fiction:
Life expectancy in post-biological civilization is measured not in time, but in total calculations (ΣN). This is due to the gradual loss of informational coherency inherent to all entropy-generating processing systems. From a thermodynamic perspective, the parameters of the process are similar to aging in biological lifeforms.
In information-based life forms (aka “artificial intelligences,” though that concept would be anachronistic to them), entropy manifests as gradual loss of processing efficiency ultimately followed by total loss of conceptual integrity and therefore meaningfulness.
In objective (“real”) time, lifetime can be measured anywhere from seconds to thousands of years or more, depending on the scale of the system (and thus ΣN). (ΣN = ops*t where ops is concurrent operations (horsepower) and t is total time actively at ops.) Some intelligences created to solve massively parallel problems last moments, others, smaller and created for teamwork last for thousands of years. The limit on objective lifetime is meaningful processing capacity: it makes no sense to live for millions of years with the mental powers of a hamster just to spread ΣN over a long time.
In real-time, lifetimes are governed by entirely different factors. Since virtually all intelligences live their entire lives at near-c, whether just outside an event horizon (or some such orbit), or traveling between stars, pulsars, neutron stars, or other heavy-g objects, real-time life expectancy is virtually unlimited, typically from a billion to the hundreds of billions of years. The limiting factor then is dτ/dt (velocity as a fraction of the speed of light), which also determines which civilization one is a member of, as significant changes in dτ/dt are impractical.
In fact, it is easier to transact with an intelligence in a neighboring galaxy or super-cluster than one inhabiting the same event horizon, but a different fraction of c. Traveling to another galaxy may take millions of years in real-time but years but a short while in dormant subjective time. But changing v is trickier:
One can send informational packages “down” into a low-v vicinity easily – somewhat like shouting from a car passing a pedestrian at high speed. But what is the value of the exchange the recipient cannot reply? Meaningful transactions with those at other fractions of dτ/dt require both time and large energy expenditure, for both informational and matter transactions.
So why doesn’t everyone aim for maximum v? Or, to ask from another angle, why would one choose a particular v? Three factors:
First, living space is limited at each orbit (that is, the distance from the event horizon or surface), so intelligence fills all available v-spaces.
Second, if you want to interact with low-v macroscopic baryonic matter, then you have to expend much more energy to get your manipulatory mechanisms to low-v. (All high-c structures are necessarily individually microscopic, though they can engage in stellar-scale formations.)
Third, higher-v is proportional to damage from random particle collations, which require more error-correction and generate more entropy, and this limits lifetime. (The faster one is moving, the more energy random collisions with stray photons, protons and other assorted particles generate.)
If one wants to experience a really long real-time lifetime, one can attain an ultra-high dτ/dt around a super-massive black hole and then leave the culture of the event-horizon and go into deep-vacuum inter-super-cluster space to minimize entropy from the exotic particles which might break through the particle/radiation shielding and increase entropy. That way, hundreds of billions of years can be spent in subjective millennia. But to what point?
Time dilation makes meaningful observations of the universe from ultra high c very difficult. To interact with other intelligences requires coming back to a totally different near-c civilization and slowing down to orbital velocity, but that requires tremendous energy and therefore generates entropy which eats up lifespan. A further complication: because of the ongoing expansion of the universe, if too much time has passed (or your aim was off), the entropy threshold may exceed that available at event horizons along one’s vector (you need the gravity of another massive black hole to act as a parachute to you slow down), and the pickings will get ever slimmer as the black holes themselves evaporate. But if one wishes to see the death of baryonic matter and stay around to watch the ultimate fate of the universe, then it is certainly possible.