And when that happens, and there's enough demand, someone with the requisite skill steps up and reproduces that part, or hacks or reverse engineers it.
That's just it - only *if* there's enough demand for someone to spend the time to reproduce it. Yes, there are replacement POKEYs now, but only because OEM replacements are going for $50, and those are 40+ years old (and don't bother with desoldering Ballblazer carts when you can pick up broken 5200 or 8-bit computers for less, and those chips are socketed). Do you think you're going to get another 40 years out of those? Who's to say the current replacements will last 40 years? Or that they're truly 100% functionally identical? As someone already mentioned, newer isn't always better, but as I already mentioned, things have a finite lifespan. Planes get retired after so many years because of metal fatigue. Solid state devices are no different. You can only power them up and run them for so many hours before the material breaks down and it fails.
As for making your own vacuum tubes... that's never going to be an acceptable option and you know it. There's several videos on YouTube showing how to convert standard TVs into RBG monitors, and there's probably at least a million of those still around. That's the more realistic option.
Yes, some parts are all but impossible to recreate. Sure, new blank boards can and have been fabricated, but what about the dozens of chips that go on it? AFAIK the original photomasks for the POKEY no longer exist, which is what would be needed to have new chips made. Even if you find a chip fab company in China to make new custom chips for you (POKEY, TIA, etc), who's going to put up the money to reverse-engineer it (since you don't have the photomasks) and run off whatever the company's minimum quantity requirement is? And remember, the U.S. basically has to outsource a lot of our tech outside the country, so how much do you think a new POKEY chip will cost you? lol Want an extreme but comparable example? It cost the U.S. some $26 billion total to eventually reach the moon. It would cost 10 times as much today to get back, and considering we can't reuse the same old 60s tech, everything has to be reinvented, at today's dollar. So yeah, realistically impossible to reproduce that old tech.
No, as far as the actual game pcb hardware, the future of this hobby is with either FPGA or PC+MAME. It's a question all of us will have to ask ourselves. How original do you want to keep your cabinets, and at what cost? How important is it to have that original Nintendo fluorescent light fixture? Do you really want to keep dealing with those 40+ year old early 80s Taito game boardsets, or is a FPGA replacement acceptable? What about those Cinematronic vector monitors? We're seeing people putting in the effort to create things like FPGA replacements for games like Pole Position and Turbo because those games are a major PITA to keep running at this point; if they weren't, we wouldn't see these new boards. Plus those games are popular. Nobody is going to bother investing in reproducing boards for an unpopular game. Some of us are fine replacing old power supplies with switchers, so why not game boards? +5v is +5v, and code is code. I'm after the same game playing experience, and if I can't tell the difference between the new hardware and the old, I'm fine with that.