Justin Smestad makes a blunt argument on his blog: if you stop hiring junior engineers because AI can handle their work, you're not saving money, you're giving your senior engineers all the power. Juniors aren't just cheap labor. They're the pipeline that becomes your mid-level and senior staff in two or three years. Cut that pipeline, and when a senior engineer demands a 40% raise, you pay it. Or you spend six months finding a replacement at an even higher rate.
The power shift is already happening.
Smestad draws a parallel to the retirement crisis hitting small businesses right now. Companies aren't closing because they failed. They're closing because the owner retired and nobody was trained to take over. He sees the same pattern playing out in tech, just faster. Stop hiring juniors in 2026, and by 2030 you've got a team of expensive seniors with no succession plan.
Many senior engineers in tech are also financially independent. They don't need the job. They work because they want to. That changes the bargaining position entirely.
Shopify is the counterexample. They're investing in early-career hiring while pushing hard into AI. Smestad argues that AI doesn't eliminate the need for a junior pipeline. It just changes what "junior" means. Future juniors might spend less time writing boilerplate and more time reviewing AI output, learning system design, and building the judgment that makes seniors effective. That takes real management attention, and it won't be cheaper in the short term.
The alternative is worse.
Hacker News commenters pushed back on the power balance argument, noting that in regions with high unemployment or heavy immigration, employers hold more cards and can replace seniors at market rate. Another commenter raised what they called the "low-background steel" problem: engineers who learned to code before AI assistance may become more sought-after, similar to how steel forged before nuclear testing is prized for sensitive instruments.
The real worry is a lemon market where employers can't distinguish between genuine skill and AI-assisted competence. That drives away good candidates and leaves only those who depend on the tools.