Has anyone attempted to fab a chip in an older process? I have wanted to build a CPU for kicks, but it would be great to have real silicon rather than being stuck on FPGA. I'd consider it reasonable if you could fab a CPU for <$2000 or so.
Many years ago I worked on an electron beam lithography system that was a converted electron microscope. You need to control the X/Y position of the beam and need some sort of shutter/deflection mechanism which should be within the realm of what a very dedicated hobbyist or group can do. I wrote the software and designed some hardware. I wonder if you could pick up an old electron microscope for cheap somewhere if that would be a starting point for a home made chip fab ;) probably too many difficult parts though, wafers, resist, bonding, packaging, clean room etc. But who knows, sometimes these things can be done on the cheap if you're willing to make some compromises...
But here's another idea, if you live close to a university that teaches semiconductor physics or chip design check with them, the system I worked on was used by graduate students and for general research...
I don't think any fab will talk to you. A "real" chip is usually millions of dollars even for older processes.
If you don't care about the actual fabrication then you can use programmable logic like FPGA and get something very close to an actual chip design running on an actual chip. You can probably just get a design to run in a simulator ...
If you can couple up with a local academic institution, and get them to submit your chip as a research or teaching project, MOSIS will fabricate your chip for free [1]. They will package up to 5 chips, though it's not clear whether the user or MOSIS pays that cost.
That's probably just about feasible, but if you want it to actually work you'll need to be or find the services of someone very familiar with the design process.
(My own design experience is a tiny corner of a Dialog Semi chip. I don't think you can do this kind of thing on your own, but with the right startup team and market you might be able to crowdfund it.)
Unfortunately your talking about something like $100,000 to get custom silicon made, even on an older node size. And you cannot get just 1, your minimum is about 100. If you insist on just 1 then they will still make a 100 and just throw the other 99 away, so you might as well get a 100 of them.
You cannot just throw some Verilog at the fab and get silicon back. You need to use some very expensive software that generates and validates the actual format they need. Hence you would go through a third-party company that specialises in taking your Verilog and does the work needed to get it to the point that the fab can make it.
There are some open source toolchains, such as Yosys [1] and Qflow [2] for synthesising Verilog into ASIC designs that can be fabricated. And with multi-project wafer fab services like CMP and MOSIS, fabrication can be had in the tens of thousands of dollars for a few hundred ICs. So things aren't quite as bad as you make it sound, but it's still a big investment and requires a lot of fairly niche skills and experience.
This is an example of the syntax it has. https://github.com/hansihe/ttl_cpu/blob/master/microcode/mic...
It, along with the assembler is configured through a yaml file. https://github.com/hansihe/ttl_cpu/blob/master/microcode/map...
If anyone is interested, there is an unfinished article about the project located here. http://hansihe.com/2016/11/25/i-built-a-computer-from-scratc...