CODE IN TOKYO

Where Did the $200K Prompt Engineers Go? — Why "Domain Experts" Are the Ones Still Standing

A few years back, you saw these headlines everywhere.

“Prompt Engineer — $200K salary” “Just type instructions into AI and earn big”

Stuff like that.

But in 2026, looking around, how many people are actually still pulling massive salaries purely by typing prompts?

In my own field of view, basically none.

To be fair, there were definitely high earners in the AI industry back then. But realistically, they were closer to “the people inside AI.”

They understood how LLMs worked, designed systems that embedded models, thought about GPU and inference costs, and translated all of it into actual business workflows.

In other words, they weren’t “prompt typists” — they were proper AI specialists.

The shelf life of a “new job title” is brutally short

The IT industry has always done this.

A new term appears. ↓ It gets treated as a rare skill for a moment. ↓ Everyone studies it. ↓ It becomes a normal skill.

Rinse and repeat.

Cloud was like this. Mobile apps were like this. Now AI is in that phase.

And prompting has a low barrier to entry.

In extreme terms, anyone who can type can do it today.

So naturally, a flood of people rushes in. Rarity value collapses fast.

As market dynamics go, pretty natural.

So what’s actually strong in the AI era?

We end up back at something people have said forever.

“People whose core specialty is strong, are strong.”

AI is useful. No question.

I use it every day. Honestly, I can’t work without it anymore.

But what I actually feel on the ground is — value rarely comes from AI alone. It almost always shows up as:

“Existing expertise × AI”

For engineers, that means not just outputting code, but thinking about:

You have to think about “real-world responsibility.”

AI speeds up how fast you write code. But “what should we build” and “is it safe to run in production” — those are still human jobs.

And this isn’t just engineers. For planners and managers too, AI can produce a first draft — but coordinating stakeholders, budgets, internal politics, and final accountability are still human work.

”Catch-up speed” alone doesn’t last

Of course, keeping up with new tech matters.

I’d say I’m one of the people who plays with new tools a lot.

But what I’ve felt strongly recently:

Making “chasing new things” itself your job is exhausting.

The hype cycle is too fast.

Last year’s “cutting edge” is normal six months later.

So eventually it comes back to:

“What am I actually a specialist in?”

AWS? Domain knowledge? Finance? Embedded systems? Data analysis?

The people who stack AI on top of that foundation are the ones who stay strong.

By the way

The same media outlets that hyped “Prompt Engineer, $200K/year” are now writing “Vibe Coding Artisan, $33K/month” articles.

Half a year from now, there’ll probably be another title.

If you jumped on every one, you’d run out of lives.

So today, again, I’ll quietly sharpen my actual craft, hand the boring parts to AI, and go to sleep early.

That’s probably the right call.

And a few years from now, while saying “remember when AI was a thing?”, I’ll probably be bookmarking the next “$XXX/year ◯◯ Engineer” article.

Humans don’t really progress that much.