The layoffs came with a story. The story went like this: AI can write code now, so we don't need as many engineers. Especially the expensive ones.
It was a clean narrative. It fit on a slide. It made the numbers work for one quarter, maybe two.
And then everything started breaking.
The logic seemed sound if you squinted. AI tools were getting better fast. Junior developers with Copilot were producing code at rates that looked, on a dashboard, like senior output. The gap between a £50k hire and a £150k hire appeared to be closing.
So companies did what companies do. They optimised for the metric they could see. They kept the juniors, cut the seniors, and called it a strategic investment in AI-first engineering.
Some of them even wrote blog posts about it.
Here's what didn't show up on the dashboard:
The senior engineer who knew why the authentication system was built that way. The one who'd say "we tried that in 2019, here's what happened" before anyone wasted a sprint. The one who could look at an AI-generated pull request and know, in seconds, that it was technically correct and architecturally catastrophic.
That person wasn't writing much code anymore. They hadn't been for years. They were doing something harder to measure: they were preventing mistakes. They were making decisions. They were the reason the system held together.
When they left, the system kept running. For a while.
It didn't break dramatically. That's the cruel part. If the production database had exploded the week after the layoffs, someone might have connected the dots.
Instead, it was slow. A design decision here that created a dependency nobody noticed. A shortcut there that saved a week and cost a quarter. An AI-generated service that worked perfectly in isolation and created a distributed systems nightmare when it met reality.
The juniors weren't incompetent. They were doing exactly what they'd been hired to do: write code, ship features, hit targets. They just didn't know what they didn't know. And the person who used to catch those things was updating their LinkedIn.
About six months in, someone noticed the pattern. Not the engineers — they were too busy firefighting. It was usually a product person, or a VP who'd been around long enough to remember what "normal" felt like.
"Why does everything take longer now?"
"Why do we keep rebuilding things we already built?"
"Why does the AI keep generating code that doesn't fit with anything else?"
The answer was always the same: because nobody in the room has the context to direct it.
AI is an extraordinary tool. But a tool without judgment is just a fast way to make mistakes. And judgment was exactly what they'd laid off.
Here's the part the companies didn't expect: the people they let go didn't sit around waiting to be rehired.
Some of them did what senior engineers have always done in downturns — they went independent. But this time was different. This time they had AI too.
A senior engineer with fifteen years of context and AI-assisted tooling is not a freelancer. They're a small team compressed into one person. They can architect a system, implement it, test it, deploy it, and iterate on it — all in the time it used to take to get a project through sprint planning.
They're not writing code the old way. They're directing AI the way they used to direct junior developers, except the AI doesn't need onboarding, doesn't take holidays, and doesn't misunderstand the ticket.
These people are building things. Real things. Products, consultancies, tools. They're doing it fast, and they're doing it well, because they have the one thing AI can't provide: the judgment to know what's worth building and how it should fit together.
I've started calling them hyper engineers. Not because they work harder. Because the combination of deep experience and AI leverage produces output that looks, from the outside, like it shouldn't be possible from one person.
Now the companies want them back.
The job postings have changed. "Senior Staff Engineer — AI-Native Development." "Principal Engineer — System Design and AI Orchestration." The salaries are higher. The titles are grander. The desperation is palpable.
But here's the thing about laying someone off and telling them AI made them redundant: they believed you. Not the part about being redundant — they knew that was wrong. They believed the part about not being valued. And they acted accordingly.
They built their own thing. They found clients who understood what they brought. They discovered that the combination of their experience and these new tools made them more capable than they'd ever been inside a company.
Why would they come back?
The layoff narrative got the causation backwards.
It wasn't: AI replaces senior engineers.
It was: AI makes senior engineers more valuable than ever, because someone has to know what to point it at.
A junior developer with AI can produce code. A senior engineer with AI can produce systems. The difference between code and systems is the difference between having ingredients and knowing how to cook.
Companies that kept their senior people and gave them AI tools are now running circles around the ones that cut them. The math was never about headcount. It was about judgment per dollar, and judgment doesn't come from a model.
The lasting effect of the layoffs isn't the talent shortage, though that's real. It's the psychological shift.
A generation of experienced engineers learned, in the most visceral way possible, that employment is not safety. That being good at your job doesn't protect you from a spreadsheet. That the company will choose the narrative over the person, every time.
So they stopped relying on companies. They built their own leverage. They learned that with AI tools and hard-won experience, they don't need a team of twenty to do meaningful work. They need a laptop and the knowledge of what to build.
This isn't bitterness. It's pragmatism. The same pragmatism that made them good engineers in the first place.
The companies laid people off to save money on engineering. Those people used the same AI tools to become more productive than they'd ever been. Now the companies are paying more — in consulting fees, in recruiting costs, in the slow tax of systems built without sufficient judgment — than they would have spent just keeping the people who knew what they were doing.
The layoff didn't reduce the cost of engineering. It redistributed it. From salaries to consequences.
And the engineers? They're fine. Better than fine. They learned something the companies are still figuring out:
AI doesn't replace experience. It multiplies it.
The people who have it are worth more now than they were before the layoffs. And they know it.
That's the part that isn't going back to normal.