You probably know about the demographic collapse that's happening in places like Japan, South Korea, Italy, Spain, and China[1]. A drop in the fertility rate is taking the population age over a cliff. The effects of this have taken decades to play out, making them extremely hard to reverse or address.
The tech industry is running the same experiment with junior engineers.
Companies are discovering that a small team of seniors armed with AI tools can outperform a much larger team. So they're cutting junior hiring. Why bring in someone who needs mentoring when Claude can write the code and a senior can review it?
You could counter this easily if you took a longer-term view. The issue is that AI has introduced significant uncertainty. Taking a longer-term view is harder than ever.
One of my aphorisms for juniors experiencing their first big mess is "Everyone drops the production database at least once in their lives." Experienced engineers were all juniors once. It's a slow accumulation of judgment that only comes from years of getting things wrong.
Cut off that pipeline and you can't backfill it on demand.
Meanwhile, vibe coding is producing more software. All of it eventually needs to be maintained by people who know what good looks like.
The current pool of senior engineers in key domains could shrink over the coming years. COBOL engineers have been retiring for decades while billions of lines of COBOL still run the world's banks and governments[2]. The supply side of COBOL is shrinking while the demand side is stubbornly static.
The difference is COBOL had a 40-year runway of decline. AI could compress that timeline considerably.
There are hopeful counterarguments: if this AI wave shows us anything, technology moves faster than demographics. Vibe coders might gain the (new) experience necessary. The AI tools they're using might become capable enough that it doesn't matter[3].
I wrote a quip in Having a Senior Moment at the start of the year. I meant it tongue-in-cheek at the time. Now, less so.