To me, bullet trains and ghost cities are speculation in the most traditional sense.
"If we build a 350kph train from Los Angeles to Phoenix, people will ride it" or "If we build 30,000 slapdash condos in a modern-day Brasilia, people will want to live there" is not really that much different from "If build an app for people to sub-lease their spare kitchen cupboard space to strangers, they'll watch advertisements to gain access."
The nice thing is that tangible-world speculation has a lower downside potential, because you're actually building something that has real-world utility to anchor its value. The ghost city might have to mark down their apartments until they finally find a selling price, but it's probably not going to be zero. The bullet train can fit into a larger economic development scheme, or worst case be cut up for scrap steel. In the process, you probably created a nice long cycle of blue-collar jobs, and worst case, you exit by selling out at 25 cents on the dollar.
When they turn out the lights at a random "tech startup" for the last time, you've distilled a few million investor dollars into a few unloved snippets of code released to the dark corridors of Github, a handful of obsolete MacBooks, and ugly trade-show gift items. There's nothing left to flog to get your money back because there was nothing there of real-world value in the first place.
I feel like this could be a welcoming alternative for some more cautious money-- risk-tolerant but not completely gambling. You know you're not going to get a "call up AWS and order more severs" growth moment, but you're not going to completely lose your shirt, and if it's done with competent people, the process is slow and relatively surprise-free.