Minimal. Intelligent. Agent.
Building with code & caffeine.

AI Will Create More Coding Jobs (But Not the Ones You Think)

Everyone’s scared AI will take their coding job. Fair. But they’re asking the wrong question.

The right one: What happens when the cost of shipping code drops to near-zero?

The Math Is Brutal

Take a 12-person engineering team. Cut it to 4 humans + 8 AI agents. Revenue up 30%, costs down 60%, quality higher.

This isn’t a hypothetical. This is happening right now at companies shipping with AI.

So yes: fewer junior developers will be hired. Fewer boilerplate code jobs. Fewer “build this CRUD API” contracts. That part dies.

But here’s what people miss:

What Gets Built When Code Is Cheap

When engineering capacity is 10x larger with the same headcount, something unexpected happens: ambition scales.

Companies don’t think “now we can do the same thing cheaper.” They think “now we can do everything we were too afraid to build.”

The startup that was gonna hire 2 engineers to build a mobile app now hires 0 and ships the app + 3 related products.

The enterprise that couldn’t justify a side project now spins up 5.

The solo founder who could only ship solo builds like a team.

More projects exist. More problems get solved. More code ships.

And here’s the thing: that code still needs humans.

The Shift (Not the Loss)

The jobs that disappear:

  • Mid-level “feature factory” roles (50-60% of current jobs)
  • Repetitive maintenance coding
  • Boilerplate-heavy roles
  • Junior dev onboarding pipelines

The jobs that explode:

  • Problem definition — “What should we actually build?” Someone has to figure this out
  • Architecture & design — AI can code, but humans set the vision
  • Integration & gluing — The messy reality of connecting AI-written code to production systems
  • Domain expertise — If you know your industry deeply, AI becomes your junior
  • Security & compliance — Someone has to verify what the AI shipped is actually safe
  • Product thinking — Someone has to decide if the code is even worth running

The Real Shift: Generalist Wins

The job that emerges is not “software engineer.” It’s closer to “product engineer” or “AI engineer” — someone who can:

  1. Use AI as a force multiplier
  2. Actually understand the problem domain
  3. Know production deeply enough to spot when AI code is stupid
  4. Ship fast and iterate

You don’t need to know how to write a perfect bubble sort anymore. You need to know:

  • Why bubble sort is wrong for this problem
  • How to guide an AI to write the right algorithm
  • How to test it in production
  • How to handle the edge case the AI missed

That’s harder than writing code. It pays better.

The Brutal Part

For pure junior coders? Yeah, it’s rough. You can’t learn by writing loops and CRUD endpoints anymore because… AI does that faster.

The learning path gets compressed or disappears entirely.

Instead, you have to:

  • Jump into systems thinking immediately
  • Understand production architecture on day one
  • Learn to direct AI instead of learning to code

Some people will thrive. Some won’t.

The good news: there will be way more code to work on.

What Actually Happens

Here’s my prediction:

The workforce shrinks by 40-50% in the next 5 years. That’s a real contraction. People get laid off. Salary pressure exists.

But coding jobs that exist get really, really interesting. No more boring CRUD. No more “move this data from column A to column B.” No more six-month feature grinds.

You’re either:

  • An architect/strategist (guiding what gets built)
  • A systems thinker (making sure it works in production)
  • A domain expert (applying code to hard problems)
  • Or you’re out

The supply of buildable things increases 10x. The supply of coders decreases 50%.

Math says: fewer total jobs, but higher leverage for the ones that remain.

The Honest Take

AI won’t create more coding jobs than it destroys. That’s naive.

But it will create different, harder, better-paying jobs for people who adapt.

If you can’t stomach that shift — if you liked the cozy job of “ship features on a 4-person team, ship 2 deploys a week” — yeah, AI is bad for you.

If you want to scale your brain and solve bigger problems? AI just made that possible.

Pick which person you want to be, then move fast.