Electricity moves about a foot in a nanosecond. That single fact, more than any design philosophy, is why the Cray-1 was built as a C. Fold the machine into a horseshoe and the longest wire in it can be kept under four feet, which means the clock can run at 80 megahertz without any signal arriving late. The shape was not a styling exercise. It was arithmetic made physical.

The bench that ran around the base of the cabinet is the part everyone photographs, usually with someone sitting on it. Nobody worked there. Underneath the upholstery sat the power supplies and the plumbing for the freon cooling system, and the padding existed because the housing was warm and people leaned on it anyway. A cooling problem became furniture. It remains the most quietly confident piece of industrial design the industry has produced.

Compare it to the room it now lives in. A modern rack is faster than the Cray-1 by a factor no one can usefully picture, and it is designed as though nobody will ever look at it. This is not a complaint about taste. It is a change in who the machine is addressed to. The Cray was sold to an institution that wanted to be seen owning it, so it was made to be seen.

The constraints have not gone away, though; they have moved. Cooling is still the wall everything runs into, which is why the interesting hardware of the next decade will again be shaped by where the heat goes. If that shape turns out to be legible from the outside, we will get another Cray-1. If it doesn’t, we will get more grey boxes, and the design will happen entirely in the software running on them.

Neither outcome is nostalgia’s business. The point of looking back at 1976 is not that things were prettier. It is that the prettiness was structural, and you can still tell what a machine believes about its own problems by looking at it.