The extreme difficulty of scaling production of new technology is not well understood. It’s 1000% to 10,000% harder than making a few prototypes. The machine that makes the machine is vastly harder than the machine itself.
Musk said this in connection with the battle he is now having to mass produce batteries.
This is a point that Matt Ridley, in his recently published book about innovation, also makes very forcibly.
Another illustration of this that I am a little bit familiar with (in the sense of having read about it) is that getting from a prototype to the mass production of the famed World War 2 bomber, the Avro Lancaster, took more than a year. This at a time when pressure to get the Lancaster flying and bombing in numbers could not have been more intense or more nationally prioritised. A prototype Lancaster first flew in January 1941, with great success, so there was not, on the face of it, a lot of trouble that needed shooting, so to speak. Yet it was over a year after that before the first RAF bomber squadron was able to switch to flying Lancasters in anger.
LATER: As I should have mentioned earlier, I got to the above quote by being told that Starlink (mentioned in this Michael Jennings comment here), which is intended to supply cheap and speedy internet connection in unpromising places, like disaster areas and very poor countries, is now being rolled out, despite whatever difficulties they had to overcome before they could start mass producing (and mass launching into space) all the kit for that.