SpaceX and the Intelligence Economy: i = er²

Research evidence excerpt

SpaceX and the Intelligence Economy: i = er²

IdeaMmost value. Can SpaceX achieve higher intelligence per watt vs. its peers?

Drivers include access to more efficient chip design, more efficient gateways/

routing/software/architecture from DC to the edge. In manufacturing Elon

Musk is fond of saying “the best part is no part.” In AI compute, is the best

token no token?

• Intelligence-per-watt-per-dollar: ((i/w)/$). Measures both the energy

density and capital density of intelligence. Imagine you had 6 companies with

the identical intelligence-per-watt. All else equal, the one that consumed the

least amount of capital to produce their (i/w) would create the most value.

Can SpaceX achieve superior capital efficiency vs. its peers? Drivers include

design, scale, data, manufacturing and an ability to capture arising data to be

fed back into the manufacturing machine in self-reinforcing recursive loops.

Data informs the software, which informs the hardware, which informs the

manufacturing plant.

• Intelligence-per-watt-per-dollar-per-second: (((i/w)/$)/t). Measures the

energy density, capital density and temporal density of intelligence. Imagine

you had 3 companies with identical intelligence-per-watt-per-dollar. All else

equal, the one that could deliver their ((i/w)/$) in the least amount of time

would create the most value. Can SpaceX achieve superior time to convert ‘e

to i’ vs. its peers with the least amount of capital deployed? Drivers include

vertical integration, pervasive physical and digital networks. As traditional

compute moves towards quantum we see each second of improvement of

time to power and ‘time-of-flight’ computing setting some companies apart

from others. The most energy efficient intelligence for the lowest price at