In addition to agreeing with the general sentiments expressed here about software paying more, there is a lot more to it. Here are a few things that probably aren't getting enough attention. I love chip design, but I am totally disillusioned with the engineering side of the business. I am actively trying to get away from it, whether it is move to the business side or management or even switch into software.<p>- It is a job, not a career. Because the industry is so consolidated and highly specialized, there are actually very few jobs available. This is why I believe it pays so much less than software. Don't like the pay, the management, your coworkers, the career advancement? Where are you going to go? There are only a couple of options which may or may not be hiring for your specialty and you can only hop around so much since there are few employers. Having tons of jobs to choose from isn't just about getting more pay, it is another form of job security and freedom. This industry doesn't have that.<p>- I get the whole argument that software is more capital efficient and so more money can be spent on salaries, and it may be part of the reason, but not the majority. Many of these hardware companies are quite profitable. They don't pay more because there are no market pressures forcing them to. I don't expect this to change even with a looming talent shortage. I know someone with a PhD, probably ~20-25 years of experience, and has spent their entire career in a single area which gives them deep domain knowledge and is making slightly more than what an L4 SWE(4-8 years experience with a Bachelor's Degree?) at Google(assuming I the information I have seen on Google pay grades are accurate). This is total comp by the way and even adjusting further for geo differences doesn't change the comparison much. Yet this is considered pretty good for our field. It is almost a sense of entitlement on the part of the semiconductor industry that they shouldn't have to compete on pay. On a couple occasions long ago I heard from official HR communications and high level management at Intel say something to the effect of being proud about paying "around industry average". Weird how they don't want average employees or average work/life balance though.<p>- Salaries don't appear to have been influenced much by Google/Facebook/Amazon/Microsoft getting into the business. Disclaimer: I haven't looked deeply into this. Are their teams to small to make a difference in the labor market? Are they paying chip industry rates instead of software rates? Are the traditional semiconductor companies just ignoring it and refusing to match their pay?<p>- The industry like many others does not want to spend money to train people, so they only want unicorn candidates. Yet it is not enough to simply be an RTL designer, or a verification engineer, or a physical engineer, they usually want someone with the exact skills. e.g. USB experience, PCIE experience, power experience, etc. Things like tuition reimbursement are mostly a myth. It certainly was at Intel despite their claims of $50k for tuition reimbursement. The funds came out of the local discretionary budgets so managers never wanted to approve it because discretionary budget is also the same budget that pays for many other things like business travel and conferences. Discretionary was also the first thing to get hit in belt-tightening situations and was a good way to look good to upper management if a site manager wasn't using their full discretionary budget. Since tuition reimbursement is a several year commitment once approved, it couldn't be cut as easily once started.<p>- Managers are incentivized to promote execution over learning/growth or innovation due to schedule pressures and frequent hiring freezes. For the same reasons, managers are also incentivized to block transfers because they don't benefit from it, even if there are rules to prohibit this behavior. Combined with wanting unicorn candidates that will accept low pay, it is no surprise that it is getting hard to find employees. Yet if you outlive your usefulness or burnout they will have no problems laying people off and looking for a new unicorn candidate rather than retraining existing employees despite all the sacrifices they made. I saw this happen first hand. Layoffs one week, 2 new job reqs the next week for RTL jobs which are hard to get. Also since they were "silver bullet" hires to fulfill "critical" needs(false!) no internal candidates from places like verification would be acceptable. So not only did they not try to save anyone from layoffs, they also basically said no one inside the group was worth investing in either. Classy.<p>- Instead of a gradual talent pipeline, I think the industry appears to have had waves of talent progression. This leads to situations where if you are between waves, you will have a group ahead of you that is fairly young but in high level positions and which puts a real limit on how far you can advance regardless of how good you are because positions are just solidified. When I graduated most of the really senior engineers were from their mid 30s through their mid 40s. Grade level *distributions* get "maxed-out" as they say. Because there are few alternatives and people stay in jobs far longer than average, it becomes stagnant. People leave the companies or leave the industry altogether. When the current cohort that is now in their 50s starts to retire, there might not be a new wave to replace them.<p>- Chip design is far more demanding than software engineering because you have to get it 98,99% right the first time you build it or you get a brick back from the fab instead of a working chip. Simulations are slow, and post-silicon ability to debug while impressive, is very limited compared to pre-silicon simulations. Unless it can be fixed in firmware, you are stuck until the next batch of samples gets made. If you "go fast and break things", you are going to fail and get fired. Schedule pressures are constant: you need product on time for back-to-school, or Christmas, or competitive threats. Sometimes the schedule pressures are artificially created by lazy management that want to create a sense of urgency. Pushing to tapeout can be grueling because you don't want to miss your fab window by being late. During power-on people are expected to work in shifts 24-7. Power on can last 2-6 weeks on early samples and be in foreign countries. This grates on people. When you look at software which has fewer schedule pressures, is less difficult to make, makes more money, and has more employers to choose from, why would anyone want to do this if they possess the skills and intelligence necessary to do software. Even if you don't work for FAANG salaries you can probably get paid the same as the chip industry and have a better life and better job security and career options.<p>- There is nothing to get excited about and inspire kids to go into the profession. The startup space is minimal so no cool new ideas or riches to be had there. When I was in high school, hardware was the show, the performance improvements were huge with every generation. There was tons of coverage of new CPUs and GPUs talking about their architectures. Now there is little coverage. Between companies trying to be like Apple and minimize the importance of hardware and the chip companies releasing less info on their chip designs as performance gains get smaller it relegates hardware to being an after thought.<p>- Lastly, I don't know if this has changed with the addition of Google/Facebook/Amazon/etc making chips, but around the time when I got out of college, the total number of EE jobs was actually declining in the US. This pours cold water on the talent shortage narrative that had been used for many years falsely. If jobs are going down, you should have an excess of workers unless they are choosing to retire or leave the industry. It also acts as a discouragement for people to join or stay in the field. Who wants to be in a shrinking industry that is already highly consolidated?<p>The EE/ECE industry has no one to blame but themselves for the looming talent shortage.<p>That was a lot to write in a short time and I didn't do a ton of proofreading, so hopefully there aren't too many sentence fragments or typos.