If we manage to get through this chip shortage and then China decides to invade Taiwan and we lose TSMC, I'm gonna fucking lose it. Everything feels so fragile right now.
I'm not convinced we'll ever see an end to the chip shortage while proof-of-waste crypto exists.<p>The interesting part of all this is that one company (ASML) won out in the race to build the machines for EUV lithography so for the cutting edge processes like TSMC's 5nm (and I assume 3nm?) the chipmakers rely on a monopoly where demand currently exceeds supply.<p>Now for GPUs and a lot of CPUs I see the obvious advantages of a smaller process: smaller chips, less power usage, less heat dissipation (although this is all complicated). What I don't understand is how older processes can't be used for things where this of primary concern. Cars have a lot of control units in them but I don't believe they all needs the billions of transistors a Ryzen does. Couldn't older processes like 14nm and 22nm be used to make these just fine?<p>This isn't my field but I'm also led to believe that a new process basically means an entirely new fab (vs retrofitting an old fab). Whatever the changes required are it's almost from the ground up except perhaps the shell of the building. If true a lot of these old processes must still exist right?
"In his personal opinion—not necessarily that of TSMC—it is going to take two to three years to bring new fabs online to resolve the situation."
Recent discussion:<p>Global chip shortage may soon turn into an oversupply crisis<p><a href="https://news.ycombinator.com/item?id=30489538" rel="nofollow">https://news.ycombinator.com/item?id=30489538</a>
Well, I'm banking on an oversupply which makes production cheaper by making my team build up chip design knowledge, so we have some designs ready when it happens. If not, FPGA dev skills will still be super useful for our work.