Yeah, this thread brought me back from my lurking
Anyway, a new Amiga needs to be cheap, and powerful.
Take an ARM quad-core (the cheap as peanuts chinese ones), a powerful DSP for musicians, a VPU for processing 4K videos for video specialists, a dedicate GPU for gaming... and you have a "custom chipset" of a sort, along with cheap as peanuts components.
Slap AROS on it (I can't see 4.x as a real alternative for the future of Amiga. You can't do closed development when your customer base is probably only a few thousand of users), and there you go.
Sell it for $200 as a powerful PC alternative for professionals (that's why you have a DSP and a VPU in there) and gamers alike and boom, sales everywhere!
You don't even need a lot of money to do that. I mean, you can see projects like FPGAArcade which sell roughly for the same money AND actually reach production, so I'm sure a small company could manage to do that.
Also make it easy to port Android stuff to it. You need to have software to let the devs come onboard. I'm sure nobody in here has forgot the A1000 sidecar and its ability to load PC software, right?
After that you create a less powerful version (cheap dual-core?) directed at offices, you know, one integrated in a keyboard with just a few USB ports, a mail client and office suite. You call it ARM500 and sell it for 99$. Small business and big business would be all over a cheap thin client
My 2c. It's easy, it's doable, and it would be the best chance for Amiga to return as a USEFUL, real alternative.