The End of Software Engineering as We Know It

I remember when being a software engineer felt like having a golden ticket. In 2021, the world was software's oyster - every company needed engineers, the work was plentiful, and I knew I could do this for decades. Fast forward to 2026, and I'm staring at an industry that might not survive another ten years in its current form.

Tasting Our Own Medicine

There's something almost poetic about this moment. We software engineers spent the 2010s automating away other people's jobs - that's literally why programming became so lucrative. Now the automation train has finally circled back to us, and it feels like cosmic justice.

The question keeping every working engineer up at night is simple: what will be left for humans to do once AI agents have fully diffused into our industry?

And before you think I'm being dramatic, consider this: I'm probably among the last to go. As a staff engineer, my job has already looked like supervising AI agents for years - communicating in human language, making sure things stay on track, coordinating work. The junior and mid-level engineers? They'll be the first casualties. Why hire a team of "hands" when you can rent Claude Opus instances for a fraction of the cost?

The Overshoot Question

My next decade will likely be defined by one critical question: will the tech industry overshoot or undershoot AI capabilities?

If companies undershoot - continuing to hire engineers long after AI can replace them - I'll keep my job longer. But "my job" will increasingly mean supervising AI agents, reviewing their code, reading their outputs. The ratio of human-to-machine work will keep shifting.

If companies overshoot, things get weird but potentially better for me in the medium term. Tech companies realize they stopped hiring too soon and scramble for technical talent to manage their sprawling AI-generated codebases. With the junior market drying up, experienced engineers become more valuable - until the models get good enough to replace us entirely.

Why Past Doom Prophecies Don't Comfort Me

Sure, people have been predicting the death of software engineering before. High-level languages were supposed to let non-technical people write code. Outsourcing was supposed to kill demand in high-cost countries. None of it happened.

But here's the thing: industries do die when technology makes them obsolete. Eventually, a crisis comes along that the industry can't ride out.

The most optimistic scenario is the Jevons effect - demand for software engineers increases because the total amount of software rises so rapidly that even with fewer engineers per line of code, we need more engineers overall. Some engineers tell me "I'll always have a job cleaning up AI-generated code."

I just don't buy it. AI agents can fix bugs and clean up code as well as they can write new code - better than many engineers, and improving monthly. Why hire humans to manage AI-generated code when you can just throw more and better AI at it?

The Maintenance Myth

The argument that AI can write code but not maintain it sounds plausible. Every engineer knows maintaining code is harder than writing it. But my experience using AI tools tells a different story.

Over the past year, I've asked almost every codebase question to an AI agent while simultaneously looking for the answer myself. I've watched them evolve from hopeless to "sometimes faster than me" to "usually faster than me and sometimes more insightful."

There's still room for competent engineers in the loop today. But that room is shrinking. I don't see any genuinely new capabilities AI agents would need to take my job - just incremental improvements in reliability and quality at tasks they already perform.

Final Thoughts

It sucks. I miss feeling secure, missing the days when my biggest career problems were internal struggles like burnout rather than external existential threats.

But it's a bit silly for software engineers to complain when the automation train finally catches up to us. At least I recognized the good times while I was in them. Even when zero-interest rates ended and the industry got less cozy, I still felt incredibly lucky to be a software engineer.

I'm in a better position than many of my peers, especially those just entering the industry. And hey, maybe I'm wrong! I hope I'm wrong - that there really is some ineffable human element required to deliver good software.

But if not, my colleagues and I will need to find something else to do. The industry that gave us so much leverage over others is about to experience that same leverage turned back on itself. Cosmic justice indeed.