Labor Compression is Coming

I’ve been a software engineer for 11 years. Recently, I built an entire startup solo—design, development, lead generation, sales—in weeks instead of months. AI made it possible.

Meanwhile, I’m still sitting through interviews where they ask if I know how to code.

The Math We’re Avoiding

Let’s be concrete about what AI-augmented productivity actually looks like.

A typical 10-person product team breaks down something like this:

With AI as an amplifier, that same output now requires 4-6 people. Not in theory. Not eventually. Now. The work still gets done. The features still ship. You just don’t need ten people anymore.

This isn’t about AI “taking jobs” in some autonomous robot sense. It’s about AI making individual humans so much more productive that you need far fewer of them.

And if a team’s work can compress into fewer people’s output, what happens when this becomes widespread?

The Calm Before the Storm

Right now, everyone’s loving ChatGPT. And they should—it’s genuinely helpful. People are using it to write emails faster, generate code, polish their work, research topics. Productivity is up across the board.

There’s talk about an AI bubble, and people are waiting for it to pop. Here’s the thing: yes, there’s an inflated financial bubble around AI investments. Valuations are frothy. Lots of money is being thrown at “AI startups” that won’t survive. That bubble will pop.

But the technology itself? AI as a productivity amplifier? That’s not going away. That’s not a bubble—that’s a fundamental shift that’s here to stay.

The Economic Reality

Competitive pressure doesn’t allow it. Some startup runs with 5 AI-augmented people and matches the output of your 50-person team. Their unit economics are 10x better. They can undercut you on price, outspend you on marketing, or move faster on features. Your company’s options are: match their efficiency or die slowly.

Shareholders won’t allow it. When boards realize they’re paying for 40-60% more labor than they need for equivalent output, they’ll demand the savings. This isn’t evil—it’s how markets work.

Performance measurement will catch up. Right now, managers can’t really tell the difference between modest productivity gains and transformative ones. Everyone’s “using AI” and getting more done. But once companies develop real metrics—actual shipped features, revenue per employee, cycle time compression—the gap becomes obvious and quantifiable.

The capability gap is real. There’s an enormous difference between “person who uses AI” and “person who knows how to orchestrate AI effectively.” Not everyone achieves the same productivity gains. Some people achieve 2x improvements. Others achieve 5x or 10x. That gap matters when headcount decisions get made.

For everyone to keep their jobs with AI productivity gains, you’d need:

None of these are true.

This Isn’t Just Engineering

Right now, AI is primarily text-based. But the compression isn’t limited to work that’s already text-native.

Once text can edit videos, create presentations, manipulate images, generate complete documents—and it’s already moving fast in that direction—it’s coming for whatever you’re doing next. The interface is expanding. The modalities are multiplying.

Companies just need to spend less to earn more.

Why This Time Feels Different

Every technological revolution displaced workers, and eventually new jobs emerged. Agriculture to industrial. Industrial to service. Why should this be different?

The timeline compression is genuinely different.

The adaptation mechanisms—education systems, social safety nets, new industries emerging—operate on 10-20 year cycles. AI capabilities are advancing on 6-month cycles. The mismatch in timescales is what makes this particularly challenging.

We’ve always adapted before, but we’ve never had to adapt this fast.

What Happens to Everyone Else?

This is the uncomfortable part. If AI truly enables 4 people to do the work of 10, then simple math says you need 60% fewer people for the same output.

If a team of 10 now needs 4-6 people, where do the rest go? The job market is tough as it is. Now we’re competing with more displaced people.

The displacement is real. People who are skilled, experienced, and competent in the current paradigm may find themselves economically unnecessary through no fault of their own.

Institutional Lag as Savior

The institutional lag is real, and it’s currently protecting jobs. Here’s why companies are still hiring like it’s 2019:

But this lag is temporary. Consulting firms are selling workforce optimization. Startups are running lean by default with AI tooling. The first companies that restructure around AI capabilities will demonstrate dramatically better unit economics, and competitive pressure will force everyone else to follow.

Interviews will shift from “can you code?” to “can you ship?” It’s a matter of when, not if.

When they update their evaluation criteria, they’ll also update their hiring models. And they’ll hire far fewer people.

I have begun hiring based on value and thought, rather than memorized algorithms or task-based work.

What I’m Actually Concerned About

Systemic instability. Too many qualified people competing for too few positions creates real social problems.

Lack of proper policy. No one’s solved this yet, anywhere. Being in Singapore triples the concern—meritocracy just means we’re on our own.

No Easy Answers

I’m not sure what to think about this. I’m not sure what the outcome or solution looks like.

Labor compression is coming whether we like it or not. Many will be affected. More value will be created with less people.

You can decide what to do with this information.