From "scratch," as in raw materials, absolutely not. Even the most dedicated home-fabrication hobbyists ( http://sam.zeloof.xyz/first-ic/ ) have come nowhere near achieving the transistor density required to construct the "OberonStation's" CPU or memory, whether taped out (laid into a chip design using actual silicon gates) or built using a programmable gate array.
From a circuit board etched by someone else and PCB components purchased online and soldered together by a hobbyist, certainly, without too much difficulty (the package used by the programmable gate array requires a bit of specialty equipment, but is within reach).
One other point about Oberon is that I _believe_ that because the design predates open-source FPGA toolchains, a closed toolchain is still required to convert the Verilog (high-level design describing the processor logic) into a netlist (FPGA configuration), and it hasn't been taped out into a chip, either. So, the book covers the design of a computer end-to-middle - that is, without covering how the block / HDL level design becomes logic gates, and without covering how said gates are then manufactured. This is probably fine - the scope has to end somewhere. The linked NAND-to-Tetris articles cover the other end - from logic gates up to typical computing constructs (adder, registers, shifters, etc.).
From a circuit board etched by someone else and PCB components purchased online and soldered together by a hobbyist, certainly, without too much difficulty (the package used by the programmable gate array requires a bit of specialty equipment, but is within reach).
One other point about Oberon is that I _believe_ that because the design predates open-source FPGA toolchains, a closed toolchain is still required to convert the Verilog (high-level design describing the processor logic) into a netlist (FPGA configuration), and it hasn't been taped out into a chip, either. So, the book covers the design of a computer end-to-middle - that is, without covering how the block / HDL level design becomes logic gates, and without covering how said gates are then manufactured. This is probably fine - the scope has to end somewhere. The linked NAND-to-Tetris articles cover the other end - from logic gates up to typical computing constructs (adder, registers, shifters, etc.).