The Future of Software Engineering (Meta Engineer's Take)

The future of software engineering comes down to one paradox: if AI writes code this much faster, why isn't software shipping any faster? I've been a software engineer for 15 years, currently at Meta, and I've watched this play out from the inside — the promises, the actual delivery, and the exact places it falls short. Big tech hiring is down 65% since 2019. CS grad unemployment just went from 6.1% to 7%. Every month another CEO announces AI-driven layoffs, and then quietly walks it back and rehires. It's a lot of noise, and noise makes it hard to take a step in any direction. So this video is the model I use to cut through it: the theory of constraints. A pipeline only moves as fast as its bottleneck. AI made one step — writing the code — enormously faster. It did nothing for design, alignment, verification, or shipping. Speed up a step that was never the constraint and you don't get output, you get inventory: a pile of almost-right code waiting on a human. Once you see that, the job market stops looking random. You can see where the field is going over the next three years, and what's actually worth learning now — whether you're a student, a junior, or fifteen years in. WHAT YOU'LL LEARN • What the numbers actually say: 7% CS grad unemployment, big tech hiring down 65%, and the 10% growth projection nobody mentions • Why software engineering turned from a ladder career into a threshold career — and what that means if you're trying to get in • The study where developers predicted AI would make them 24% faster, felt 20% faster, and measured 19% slower • The prediction scorecard: Amodei, Klarna, Amazon, and Altman's "I'm delighted to be wrong" • Theory of constraints applied to software: why the bottleneck moved from writing code to verifying and shipping it • Why "almost right" AI code costs more time than it saves • Why SaaS didn't die — and why rolling your own version of the software you pay for is usually a bad trade • Four ways to prepare: wear multiple hats, learn to ship, train verification like a skill, and stay in the game CHAPTERS 00:00 Is Software Engineering Still Worth It in 2026? 01:54 The Job Market Numbers: CS Grad Unemployment and Big Tech Hiring 05:14 Ladder Career vs Threshold Career: Why Junior Roles Disappeared 08:10 The AI Productivity Paradox (The METR Study: 19% Slower) 11:13 AI Predictions vs Reality: Amodei, Klarna, Amazon, Altman 12:59 Theory of Constraints: Why Faster Code Doesn't Ship Faster 18:49 Why Side Projects Die: Shipping Is the Hard Part, Not Coding 21:21 Is SaaS Dead? The SaaSpocalypse That Didn't Happen 25:30 Strategy 1: Wear Multiple Hats (Pure Coders Get Replaced) 28:53 Strategy 2: Learn to Ship End to End 30:07 Strategy 3: Train Verification Like a Skill 32:02 Strategy 4: Stay in the Game (Why AI Skills Transfer Anywhere) 34:34 Are We Okay? The Verdict 35:05 Altitude and The Workshop Community RESEARCH MENTIONED The METR study I promised to link — developers predicted +24%, felt +20%, measured -19%: https://metr.org/blog/2025-07-10-earl... Paper: https://arxiv.org/abs/2507.09089 The follow-up study (Feb 2026, 57 developers, 800+ tasks) — the gap narrowed to -4%, still no measured speedup: https://metr.org/blog/2026-02-24-upli... The theory of constraints comes from "The Goal" by Eliyahu Goldratt — the management book behind the pipeline model in this video. VIDEOS MENTIONED ▶️ The 4 Levels of Coding With AI — the verification systems I build to stop being the bottleneck: The 4 Levels of Coding With AI (Most Peopl... ▶️ How to Learn To Code In 2026: How to Learn To Code In 2026 ▶️ Don't waste 2026 learning the wrong tech skills: Don't waste 2026 learning the wrong tech s... LINKS 🚀 Altitude — my app that teaches you to code and work with AI by building a real project inside a real coding agent. It tracks what you've actually learned and teaches you what you don't understand yet. Join the waitlist: https://learnaltitude.com 💬 The Workshop — my free community for people building with AI. Our next 30-day challenge starts in a couple of weeks: https://jasonk.us/workshop #SoftwareEngineering #AICoding #TechCareers #FutureOfWork