The world needs millions of AI engineers. It barely has any.
So why does everyone feel like they’re already too late?
A developer I spoke with this week, Farah, gave me the sharpest answer I’ve heard.
Her argument, roughly: AI engineering is a brand-new field. It didn’t really exist until 2022. People didn’t take it seriously until 2023 and the first half of 2024. People like her are only starting to learn it now.
Sit with what that means. An entire global workforce still has to be trained in this. That doesn’t happen in a year — it’s a whole industry retooling, one person at a time, over years.
Everyone is talking about AI. Very few can actually build it. That gap does not close quickly.
So even after the hype cools, demand for people who can genuinely do AI engineering stays high — because the supply is tiny and slow to grow. Anchor in now, she said, and you can farm a decade of opportunity from it.
You might push back with the obvious objection: won’t AI just write the AI? Won’t coding agents like Claude Code do the engineering?
Her answer, and I agree: AI will do part of it. But someone has to work alongside it — and someone has to be accountable. When AI ships broken code, a company can’t blame the model. They need a human who understands the system end to end.
So no, this isn’t a train you missed. It’s one that’s barely left the station.
Get on early,
Kirill
P.S. Every giant field looked empty from the outside right at the start. AI engineering is about three years old. We’re still at the beginning.