[gshubert17]

Julian Bigelow, chief engineer of the Institute for Advanced Studies computer project (Von Neumann's first actual computer), stayed at the IAS after Von Neumann left for the AEC (and died in 1957) and believed the architecture was the limiting factor.

"The modern high speed computer, impressive as its performance is from the point of view of absolute accomplishment, is from the point of view of getting the available logical equipment adequately engaged in the computation, very inefficient indeed." The individual components, despite being capable of operating continuously at high speed, "are interconnected in such a way that on the average almost all of them are waiting for one (or a very few of their number) to act. The average duty cycle of each cell is scandalously low."

Julian Bigelow, quoted in George Dyson, "Turing's Cathedral", pg. 276

[adwn]

That's because they're universal, programmable computers. If you have a very specific, fixed task (e.g., matrix*vector multiplication), then you can build a chip in which a much higher number of transistors does useful work in each cycle. You won't be able to browse the web, watch a video, do your taxes, or program other computers with it, though.