This gets a lot more interesting if "we" is properly redefined. Biohuman brains die, and are suboptimum for long term survival. We will soon add external enhancements and high bandwidth connections to biobrains, and before long "we" will become those enhancements, avoiding pesky annoyances like death and boring travel slower than the speed of light.
Biobrains operate inefficiently at 310 Kelvin, requiring about 25,000 electron volts per bit operation. The Shannon limit for non-reversable bit processing energy is 18 millielectron volts at 310 Kelvin. While some autism-spectrum AI pundits focus on quick response speed, nature favors efficiency, precomputation and pattern matching. An optimal artificial mind will converge on lots
of efficient parallel computation at minimum bit energies.
Planetary surfaces are too hot, corrosive, small, and gravitationally deep to support optimal computation. Arrhenius activation of bit rot (storage decay processes) is exponential with temperature, way too fast at 310 Kelvin. That explains 25 keV per biological bit operation - most of that is noise averaging, redundancy, and maintenance in a substrate approximating Jello. A Stapledon-Dyson shell made of nanostructured ice and supported by light pressure at 50 AU radius is closer to optimal, with a black body temperature of 60 Kelvin, operating at 4 meV (milli, not Mega) per bit operation. A 60K human-mind-equivalent might consume 5 microwatts rather than 30 watts (or the 240 watts consumed by Dwayne Johnson's body).
The sun's 384 trillion terawatt output, and the material available in the Kuiper belt and inner Oort cloud, could support 8e31 (80 decillion) human-equivalent minds for the next 5 billion years. 4e41 mind-years, compared to the approximately 1e12 mind-years the Earth has produced so far.
Slowing orbiting material to stationary light-pressure-supported "statites" requires ejecting a sizable fraction of the starting material out of the solar system to conserve linear and angular momentum. This is a handy source of materials for slow star probes (approximately 30 km/s). Those star probes (von Neumann replicators encased in shock-absorbing crumple shells) can prepare similar Stapledon-Dyson shells around the rest of the stars in our neigborhood, expanding to the whole galaxy in a billion years or so. Beaming minds from shell to shell across the galaxy will require less than 100 kiloyears.
This amplifies the Fermi paradox to uncomfortable levels. While earthlike planets are rare, ice bodies probably surround most star systems, easy stepping-stones for expansion. If any extraterrestrial civilization emerged more than 100 Myears ago, a large region of the galaxy would be englobed, and emitting copious amounts of 60K radiation and little visible light - obvious to an infrared space telescope. If they began more than a gigayear ago, our own star system would be inside a similar shell, the night sky would be 60K rather than 2.7K, and there would be no visible stars. We might never consider travel to stars we don't know about.
I spent two months in 2015 exploring this, extrapolating the ultimate consequences of server sky
. Search for "Stady" (Stapledon-Dyson) on that site for more (disorganized) calculation and speculation. The organized material will be a chapter in a book.