To me it’s clear adding the ability to express intent to parallelise is the Right Thing. This is the only way the compiler can actually know what you want it to do.

One of the smallest sculptures in the universe.

I believe the rationale is that they are so much denser than they will compensate the price difference over the entire lifetime of the system. A 10 TB install is a full rack otherwise and this is 5% of the space. Colocation for HFT tends to be expensive and using 5% of the space for the same amount of data might make sense.

IIRC, metals aren’t great shields because of secondary particles emitted after the first one hits the shield.

What is the designed lifespan of a Starlink satellite?