My interpretation of “more expensive to shut it down” includes an implicit assumption that it’s not a permanent closure and there will be a need in the near future to start it up again.
Obviously, it costs money to operate something and if you stop the operation permanently you’ll save that money. But if you shut it down and then try to get it back in gear a year or two later, you have a bunch of additional expenses you weren’t previously expecting. At minimum you have to try to chase down and rehire as many people as possible, so you don’t need to retrain. Any remaining gaps, that’s all fresh recruiting. Then, in a manufacturing facility, you have to inspect and clean up machinery that’s been sitting idle; large-scale mechanical automation generally prefers to be kept running, and an extended shutdown will require tons of hands-on work to make sure everything is safe to reactivate. You need to renegotiate and reestablish supply chains. And so on and so on.
In that context, I can see how it would be less expensive, from a bottom-line perspective, to just keep things running in the short term.
(In my own job history, there was an on-again-off-again software project that a certain executive kept initiating and then abandoning and then resurrecting, which wound up costing three times as much as it should have due to all the stop-and-start overhead. That’s the kind of example I have in mind that’s informing my interpretation.)
Obviously, the flip side of this is the sunk-cost fallacy, the idea that as long as you’re spending the money now you should just keep grinding forward because you’re already on the road. The only way to avoid it is to perform the difficult analysis and try to project which path is more efficient. I presume that’s what’s been done here.
The reason it’s funny, though, is that the analysis assumes demand for the Cybertruck will rebound, and this backlogged inventory will move, and eventually the market will catch back up with the factory’s output. That’s the part that I find amusing — Musk is utterly delusional about the future of the Cybertruck and doubtless would not accept any analysis that says “shut down the line and we’ll re-evaluate in two years.” So if the article is accurate and if my interpretation above is correct, the whole thing is skewed by faulty assumptions.
Unless Musk thinks he has some kind of secret deal with Trump to buy the whole kit-and-caboodle and slap, I don’t know, postal service logos on them or something. That could explain his perspective as well.