While the point is valid, the upside still isn't practically infinite; humans intuitively do something close to hyperbolic discounting for future returns (I assume evolution did this because of the possibility of black swans but it isn't talking).

The easy version of this is to value a company over the next 20 years of profits. The first half of that time horizon, I don't see much changing at all outside of Starlink subscribers. They've only got ~5 Mars launch windows in any given decade, and while their design philosophy is great for LEO where launch windows are daily, launching rockets to see what fails is exactly wrong for Mars.

OTOH, now to 2036, there's at least a chance AI goes boom. Perhaps even one of Musk's AI. Great for the overall economy (at least right up until it changes the economy so much that capitalism looks as obsolete as feudalism looks today, which may or may not be in that horizon), but even if it's the next industrial revolution, there's so much competition there that I expect everyone's profit margins to be almost nothing. And conversely, if any one of them does appear to be massively dominant, in comes an anti-trust case (who remembers Standard Oil?)

Same with rockets, though a "rocket boom" is perhaps an unfortunate phrasing. He finds a way to make more money than expected, just like he's already doing with Starlink? Great. Industrial espionage is a thing, and even if it wasn't, China's not got a problem with the state funding businesses to simply re-discover everyone else's secret sauce. US companies and residents may find themselves restricted from using the fruit of that work, but most of the world will just buy what's cheapest.

Still, as I said, your point is valid.