Comment by jopsen
4 days ago
I suspect the kinds of salaries that's possible in Silicon Valley only happens because:
(A) Skills are fairly transferable. (B) There is a lot of employers competing for workers. (C) An awful lot of value is created along the way.
If you specialize in some tiny part of chip manufacturing, there aren't many places you can transfer your skills.
Even if, in the future, you have multiple chip vendors. They won't all use the same processes, and you might only fit into one role at each of these businesses.
Maybe it's not that simple. But few chip companies have to compete against startups for workers. And that probably won't change.
Not saying the jobs can't be well paid, just that it's not unlikely that it won't be absurd SV level salaries.
> Maybe it's not that simple. But few chip companies have to compete against startups for workers. And that probably won't change.
It seems like what EE needs is something similar to open source, so that does happen.
The way things like Google or AWS got started is they started with Linux and built something on top of it, so it could be a startup because they don't first have to build the world in order to make a contribution, and they're not building on top of someone else's land.
There isn't any reason that couldn't inherently work in EE. Get some universities or government grants to publish a fully-open spec for some processors that could be fabbed by TSMC or Intel. Not as good as the state of the art, but half as good anyway.
Now people have a basis for EE startups. You take the base design and tweak it some for the application, so that it's a startup-sized job instead of a multinational-sized job, and now you've got EE startups making all kinds of phone SoCs and NVMe drives and Raspberry Pi competitors and whatever else they think can justify a big enough production run to send it to a fab and sell it to the public.
An interesting license for this could be something along the lines of: You can make derivative works, but you have to release them under the same license within five years. In other words, you get five years to make money from this before it goes into the commons, which gives you the incentive to do it while keeping the commons rich so the next you can do it again tomorrow.
I believe you’ve just described the RISC-V project, though I could be mistaken.
RISC-V is the ISA, which is a solid first step. What you need is a production-ready fully open source whole device, so that someone who wants to fork it only has to change the parts they need to be different instead of having to also re-engineer the missing components.
There were a ton of chip making startups in the 1970-1980's. Now the processes are much harder to access so you have fabless.
It's just maturity. You can't invent the op amp twice.
The same analysis makes me doubt those wages are likely to prevail for software engineers. They are the result of a particular time and place.