Re: Gatekeeping
Not sure if you're trolling or just really far out of your depth, but you seem to have no clue what's involved in modern chip design and manufacturing, and a severe lack of critical thinking. As an analogy to describe how ridiculous your idea is: If you told me you wanted to design and build a road-legal car in your shed from scratch using only raw materials (aluminum ingots, rubber/plastic pellets, etc), I would give you higher odds of success than if you told me you wanted to design, manufacture and verify any logic circuit that's too complex for a 15€ off-the-shelf FPGA.
Let's ignore for a second that most people would struggle to even build a clocked flipflop from logical components, nevermind designing anything more complex based on physical components (as in, the transistors and resistors that make up the logical components such as a NOR gate). Let's further ignore the complexities of formally verifying that your design actually does what you want, translating the design to a working chip, and verifying that the resulting chip still adheres to the specification and can hit the target frequency and power levels. And of course we're ignoring the enormous requirements for creating and handling wafers and performing litography anywhere near the required orders of magnitute, which would include things such as a clean room and vacuum processing.
Even if some machine were able to magically turn an idea into an etched wafer with a circuit correctly implementing your idea, you would have lots of work in front of you before you could actually use it. You handwave away the cutting and packaging step (which are far from trivial and currently require machines bigger than your garden shed), so let's assume the magical device directly produces an IC. But unless that IC is a black box fixed-function component which you plug into an existing circuit, your work would still be far from finished. You would need to design and build a fitting PCB, including any further logical and analog components (e.g. power delivery). If the chip requires any sort of firmware, now you have a big coding project as well, and need a way to flash the compiled firmware to the chip before you could use it. For anything more complex you would also need some sort of OS on top of the firmware, which again complicates things, even if we assume you could re-use an existing embedded OS with minimal adaptions to its kernel.
Even if you managed to do all of that work in your garden shed, the next question would be why you'd want to go through all the hassle. If you need a custom chip implementing some digital logic, just buy an FPGA off the shelf - a random example chip from Lattice with a 6k LUT and 180Kb mem costs 15€ for a single chip and less in bulk. If your requirement is too complex for currently available FPGAs then even creating a correct chip design for it would be an enormous task measured in man-years, nevermind manufacturing and integrating that chip. And even then you would probably be vastly better off by simply using an available microcontroller and programming your logic in code.
This post ignores various further things, e.g. that your magical chip machine would be far behind in process technology compared to bleeding-edge nodes and thus would be laughably slow and inefficient. Just the light source for anything in the EUV and DUV ranges would be so vastly complicated and expensive that the idea of a private buyer owning such a device is about as realistic as privately owning a nuclear reactor, and that won't change unless someone achieves a whole lot of physics and engineering breakthroughs.
So TL;DR: Your hypothetical chip-making device would currently be a few orders of magnitude too expensive, big, and power-hungry for personal use. All of these issues are complicated enough that you'd get a nobel prize for shaving even one order of magnitude from one of these quantities. But even just designing and verifying a non-trivial chip is also far too complex for a hobbyist in their shed, integrating the chip into a circuit to do anything useful is not trivial either, and an off-the-shelf FPGA or microcontroller would probably be the better choice in all cases.
Biting the hand that feeds IT
