To me the biggest problem has been since the 80s the shift to make this technology funded by Americans nd ten capture by PE then farmed out to cheap cost labor. Say you want to use CADENCE and even down to MATLAB you are forced to pay enormous SW cost. The hardware is so expensive while in other countries it’s incredibly cheap to get into IC/FPGA etc design. I can walk down the streets of these other nations and purchase a $50k/month SW what someone in China it’s for $5. Our elites priced us out. That’s the end of it.
On #5 , many of the documentation issues stem from politics and ego too. This is on the higher side in the semi industry. It becomes worse where the culture itself is tribal. Very few are "deep, experiential knowledge that can’t easily be replicated" issues. It is some proprietary information or document that gets withheld by someone which causes these issues.
Also poor tooling and workflow causes enormous amounts of stress in #6. There are so many crappy tools and the logging is extremely antiquated. Also due to poor documentation, patch work done on earlier parts the problem keeps coming up in various ways in the later chips. So essentially it becomes a game of someone planting a problem due to their carelessness and later down the road, someone spending a lot of time hunting it down and creating patch work and finally it breaking elsewhere.
Due to all these, why would a passionate engineer waste his time in such dinosaur semi companies which is where all these problems usually lie? A better proposition would be to work in a startup semi company where you are not bogged down by baggage. Unfortunately startups in semi are few and far between.
A lot of it is kicking the can down the road. My first reaction when I joined the industry was — “wow that’s it? How do billions of chips even work?” I had an impression that people really have thorough ways before making chips. Turns out it is much more ad hoc than that.
Smaller companies is where I learned the most. It was where my work had most impact and was most satisfying. That’s for sure. In Big companies you feel like the cog that you actually are.
"My first reaction when I joined the industry was — “wow that’s it? How do billions of chips even work?” I had an impression that people really have thorough ways before making chips. Turns out it is much more ad hoc than that. "
I had the same experience. It is like finally finding out how the sausage is made and it is brutal. I joined the semi industry coming from experience as an embedded engineer where I worked a majority of time in the medical devices domain and also a little bit in military and transportation domain. So a significant experience in the regulated domain where kicking the can down the road is simply not an option and mistakes will get people killed. Looking at how semi chips are developed in the non regulated domain, with horrible processes, it was too much for me. I then understood why getting support as an end user from the chip manufacturer was such a pain or the why turn around time to fix a bug is so slow.
Luckily the situation in regulated semi industry seems to be much better where things are slowed down I guess based on some interaction with some of my colleagues on the other side.
Also worse thing than a big company is a big company with lots of acquisitions. It is where the left hand does not know what the right hand is doing where processes, documentation and EDA tools are all over the place with some critical engineers cashing out and in semi retirement mode. It is a nightmare to say the least.
All these problems also stem from the fact that there is not much competition in the EDA industry and the licenses are prohibitive for young competitors and therefore job opportunities are limited for engineers. This is why you will find a lot of boomerangs happening in this industry and people are too scared to burn bridges or shake the boat.
Luckily FAANG companies are investing in OSS tools and it is picking up. Hopefully we see more and more competition and a lot of companies bloom.
There was an interesting take in the Oxide podcast recently by Bryan Cantrill where hopefully Intel opens their fabs and makes many things open source to gain customers for their foundry. Hopefully something like this happens.
Your reasoning in #2 doesn't really convince me. It doesn't make sense why you would IEEE survey as a comparison when Levels collects data on hardware engineering salaries as well.
If you navigate to Level's hardware engineering section you can see that new grads make 70-80k less than software engineers, and a staff hardware engineer at Qualcomm for example makes half of what a FAANG engineer makes.
If you are a student what sounds better: make 250k/yr in 3 years, or make 250k/yr in 8 years?
My point is I’m sure a converse example can be found as well. And not every salary is reported on levels. I know first hand that engineers in certain “hot” hardware fields are paid a lot in stock for retention purposes so they don’t go to the competitor. It’s not cut and dry.
Location location location! As a 20 something where are you going to want to live? Malta, NY? Round Rock, Texas? Hillsboro, Oregon may be reasonable, if you don't mind a commute from Portland. Compared to software jobs semi work is in "lame" places.
On top of that who wants to work in the clean room for 8+ hours a day? A cushy desk chair in the office or at home is much nicer.
I think fewer people are willing to compromise on location when pay isn't as nice and the working environment is less cushy
Personally, I left the semi industry after 10 years because I couldn't find work with in reasonable distance of where my wife was living. It's a shame because I loved the work, but it just was too inflexible to allow for the family life I wanted.
#5 is a really good point. I've faced issues where the lack of documentation and a proper documentation format really inhibit productivity. The amount of time I spend on an issue increases drastically due to bad documentation, which means more cost, etc.
I'm currently pursuing a masters degree in EE because after finishing my CS undergrad studies I got deeply interested into semiconductor manufacturing. Certainly, I am finding it very hard to navigate the space and how to pivot into getting a job there, it's such a vast field, and as you point out, much more different than the software world I was used to.
"Culture of secrecy" is a way to weaponize information. I've seen a lot of times in multiple companies where senior engineers weaponize tribal information in order to degrade perfectly capable junior engineers in order to make them feel stupid, and less likely to challenge the status quo. And why wouldn't they? Hardware engineering as you say can be niche and it is difficult. So knowledge hoard what little you know to as a means of job security. If it's an adversarial culture where companies pit teams against each other, you spend the majority of your time trying to untangle layers of obfuscation to get at the root of the real ask and motivations. Sooo much easier to work with people who you can trust.
To me the biggest problem has been since the 80s the shift to make this technology funded by Americans nd ten capture by PE then farmed out to cheap cost labor. Say you want to use CADENCE and even down to MATLAB you are forced to pay enormous SW cost. The hardware is so expensive while in other countries it’s incredibly cheap to get into IC/FPGA etc design. I can walk down the streets of these other nations and purchase a $50k/month SW what someone in China it’s for $5. Our elites priced us out. That’s the end of it.
On #5 , many of the documentation issues stem from politics and ego too. This is on the higher side in the semi industry. It becomes worse where the culture itself is tribal. Very few are "deep, experiential knowledge that can’t easily be replicated" issues. It is some proprietary information or document that gets withheld by someone which causes these issues.
Also poor tooling and workflow causes enormous amounts of stress in #6. There are so many crappy tools and the logging is extremely antiquated. Also due to poor documentation, patch work done on earlier parts the problem keeps coming up in various ways in the later chips. So essentially it becomes a game of someone planting a problem due to their carelessness and later down the road, someone spending a lot of time hunting it down and creating patch work and finally it breaking elsewhere.
Due to all these, why would a passionate engineer waste his time in such dinosaur semi companies which is where all these problems usually lie? A better proposition would be to work in a startup semi company where you are not bogged down by baggage. Unfortunately startups in semi are few and far between.
A lot of it is kicking the can down the road. My first reaction when I joined the industry was — “wow that’s it? How do billions of chips even work?” I had an impression that people really have thorough ways before making chips. Turns out it is much more ad hoc than that.
Smaller companies is where I learned the most. It was where my work had most impact and was most satisfying. That’s for sure. In Big companies you feel like the cog that you actually are.
"My first reaction when I joined the industry was — “wow that’s it? How do billions of chips even work?” I had an impression that people really have thorough ways before making chips. Turns out it is much more ad hoc than that. "
I had the same experience. It is like finally finding out how the sausage is made and it is brutal. I joined the semi industry coming from experience as an embedded engineer where I worked a majority of time in the medical devices domain and also a little bit in military and transportation domain. So a significant experience in the regulated domain where kicking the can down the road is simply not an option and mistakes will get people killed. Looking at how semi chips are developed in the non regulated domain, with horrible processes, it was too much for me. I then understood why getting support as an end user from the chip manufacturer was such a pain or the why turn around time to fix a bug is so slow.
Luckily the situation in regulated semi industry seems to be much better where things are slowed down I guess based on some interaction with some of my colleagues on the other side.
Also worse thing than a big company is a big company with lots of acquisitions. It is where the left hand does not know what the right hand is doing where processes, documentation and EDA tools are all over the place with some critical engineers cashing out and in semi retirement mode. It is a nightmare to say the least.
All these problems also stem from the fact that there is not much competition in the EDA industry and the licenses are prohibitive for young competitors and therefore job opportunities are limited for engineers. This is why you will find a lot of boomerangs happening in this industry and people are too scared to burn bridges or shake the boat.
Luckily FAANG companies are investing in OSS tools and it is picking up. Hopefully we see more and more competition and a lot of companies bloom.
There was an interesting take in the Oxide podcast recently by Bryan Cantrill where hopefully Intel opens their fabs and makes many things open source to gain customers for their foundry. Hopefully something like this happens.
Great description of the semiconductor industry overall.
Your reasoning in #2 doesn't really convince me. It doesn't make sense why you would IEEE survey as a comparison when Levels collects data on hardware engineering salaries as well.
If you navigate to Level's hardware engineering section you can see that new grads make 70-80k less than software engineers, and a staff hardware engineer at Qualcomm for example makes half of what a FAANG engineer makes.
If you are a student what sounds better: make 250k/yr in 3 years, or make 250k/yr in 8 years?
My point is I’m sure a converse example can be found as well. And not every salary is reported on levels. I know first hand that engineers in certain “hot” hardware fields are paid a lot in stock for retention purposes so they don’t go to the competitor. It’s not cut and dry.
I think that Adding Qspice would be a great thing for another tool as it is really rtaking off in the Mixed Signal space.
Location location location! As a 20 something where are you going to want to live? Malta, NY? Round Rock, Texas? Hillsboro, Oregon may be reasonable, if you don't mind a commute from Portland. Compared to software jobs semi work is in "lame" places.
On top of that who wants to work in the clean room for 8+ hours a day? A cushy desk chair in the office or at home is much nicer.
I think fewer people are willing to compromise on location when pay isn't as nice and the working environment is less cushy
Personally, I left the semi industry after 10 years because I couldn't find work with in reasonable distance of where my wife was living. It's a shame because I loved the work, but it just was too inflexible to allow for the family life I wanted.
Not all semi jobs are in undesirable locations. Still, I get your point. If the job isn’t compatible with the life you want to live, time to move on.
#5 is a really good point. I've faced issues where the lack of documentation and a proper documentation format really inhibit productivity. The amount of time I spend on an issue increases drastically due to bad documentation, which means more cost, etc.
I'm currently pursuing a masters degree in EE because after finishing my CS undergrad studies I got deeply interested into semiconductor manufacturing. Certainly, I am finding it very hard to navigate the space and how to pivot into getting a job there, it's such a vast field, and as you point out, much more different than the software world I was used to.
"Culture of secrecy" is a way to weaponize information. I've seen a lot of times in multiple companies where senior engineers weaponize tribal information in order to degrade perfectly capable junior engineers in order to make them feel stupid, and less likely to challenge the status quo. And why wouldn't they? Hardware engineering as you say can be niche and it is difficult. So knowledge hoard what little you know to as a means of job security. If it's an adversarial culture where companies pit teams against each other, you spend the majority of your time trying to untangle layers of obfuscation to get at the root of the real ask and motivations. Sooo much easier to work with people who you can trust.