Compensation Bands and Promotions - by Philip Su
Molochinations
SubscribeSign in
Compensation Bands and Promotions<br>Income-maxxing has some unintuitive subtleties
Philip Su<br>Jun 05, 2026
51
Share
I previously gave an overview of how to max out your compensation, which summarizes to: do something rare and valuable. Today I’ll talk more specifics about how promotions affect compensation.<br>Fireside True Story™ Time: I’ve never accepted a job due to compensation. Well, actually, I did once in college: living near Washington DC, there were things called “line services” where lobbyists could pay college kids to stand in lines outside government buildings in order to save them spots in public hearings. At a time when minimum wage was $4.25, each hour of standing in line was $19. It was some of the easiest money I’ve ever made.<br>But I’ve never accepted a job based on compensation as a key factor. When I started at Microsoft, it was the second lowest offer I received. I had loved my internship and knew I’d learn a lot by joining, so I did. Similarly, when I joined Facebook, I took a lower salary and less RSUs because I wanted a chance to update my skills at a time when Microsoft was still stuck in shipping boxed enterprise software every 3 years.<br>Just because I haven’t based career decisions on compensation doesn’t mean you can’t. I’m empirical proof, though, that compensation is often a natural side effect from building valuable skills.
Mid-Career Dominates Long-Term Compensation
I lead a seminar every year in University of Washington’s computer science program around career topics. Each year, understandably, seniors ask me a lot of questions as they mull over how to respond to job offers. Compensation differences between offers often cause them uncertainty.<br>My biggest suggestion to students contemplating different compensation packages is not to overthink them, because mid-career compensation far outweighs early-career compensation. If you want to be compensated well, the most productive thing you can do is build yourself into an amazing mid-career engineer.<br>This is the spiritual opposite of a child star going nowhere in their adult acting career. You’re like a child star when you get hired out of college: people haven’t seen much of you yet, but they think you have promise. By the time you become an adult, the best actors have been separated from many who never progressed beyond their childhood potential.<br>How dominant is mid-career compensation over early career? In my case:<br>Started at Microsoft at $42k a year, with perhaps an additional $12k RSUs a year.
Ten years in, went from L66 → L67, where base pay was around $150k with around $50k RSUs.
Fifteen years in, my pay at Facebook was ~$250k with $1M RSUs. Note this is after taking a pay and RSU cut to join Facebook from Microsoft.
Twenty-five years into my career, at age 48, my base pay at OpenAI was around $450k with $2M RSUs.
Several things might stand out:<br>The magnitude of difference in what you might earn in your 40’s is hugely different from your first job out of college. In looking at the numbers above, it’s clear that optimizing for maximal mid-career compensation is the wisest strategy while nearly disregarding early offers. Remember: Microsoft was my second lowest offer, but I took it believing it would build skills which would make me more valuable later — which it definitely did.
Companies compensate differently. OpenAI has a different comp philosophy from Microsoft. As a gross oversimplification, harder-to-get jobs tend to compensate more because they generate higher applicant demand. Getting yourself that harder-to-get job is all about what skills you’ve built earlier in career.
Your compensation growth will be lumpy. It’s more a punctuated equilibrium with large discontinuities than some sort of monotonically rising line graph. Expect years where nothing changes much. In fact, expect to sometimes take jobs offering less in order to invest in yourself. And if you work at startups, of course, pay cuts are a real thing. I’ve omitted experiences from the above timeline, like my three years starting a nonprofit where I was paid less than a tenth what I made the years before.
Compensation growth is geometric. Students in my seminar have a hard time intuiting that compensation in their 40’s will dwarf compensation in their 20’s because they don’t quite see that it grows geometrically. For instance, companies tend to increase pay by fixed percentages (e.g. “Everyone who gets promoted this year gets a 2% raise.”). Since those percentages apply on top of whatever your compensation happens to be, growth over time is geometric.
Some promotions have outsized consequences. Back when I worked at Microsoft, a promotion from L65 → L66 had a fixed percentage increase in RSUs, but going from L66 → L67 doubled your RSUs. There are similarly large discontinuities when going from L67 → L68 (“Partner” or E8). This tends to be because the market of qualified...