I think the day of the 8255 has come and gone. For the cost and footprint of a 8255, you can have another micro. Waferscale had their attempt at peripheral add-ons which was fine for a while, but now just about all micros have inbuilt flash thus freeing up port pins and iic/spi/uart interfaces to inter-cpu comms. Personally, if I need more i/o I choose a micro with more i/o! The other reason is EMC - not having high speed signals routed around the board makes compliance easier - having on-chip flash has made this easier. For special logic, we have CPLDs and FPGAs, otherwise one would use standard chips configured as a shift register (there's even parts with SR and high current drive). With all this in mind, doing a new design even with a new & improved 8255 would not even register on my radar. There would have to be something extremely compelling for me to do this - and I can't think of anything compelling at the moment.

Their might be a requirement as a spare part - I'm not sure who still makes the 82(c)55 and it might have to be military qualified to make it worthwhile.

Speak to your customers that buy lots of ics and ask them for a wish list. I don't know what your break-even point is, but I would expect you'd have to sell 100,000+ devices to make it feasible. Some time ago, I would have suggested a simple usb host chip, seems quite a few companies are doing this now. How about a simple iic/spi to video chip. There's a few on screen display chips avail, but they have limited resolution - I'm talking something that can do decent graphics (320*200) and on-chip VRAM. I have no use for something like this at the moment but with the proliferation of cheap lcd monitors, there seems to be a requirement for such a device. If you could put the 8051 with a slab of flash on it, you might have a winner. Who else does such a thing - apart from the DVD chip makers?