When AI Becomes the #1 Reason for Layoffs: A Calm Observation from an Engineer

Preface

In March 2026, AI overtook "operational restructuring" and "market adjustment" for the first time and officially became the top reason for corporate layoffs, accounting for 25% of all announced cuts — corresponding to 15,341 positions.

That number made me pause. As an engineer who writes code with AI tools every day, I'm honestly not sure how to process it. On one hand, I do feel the productivity gains AI brings; on the other, when "productivity gains" starts to equal "your colleague got laid off," it stops being a purely technical question.

This is a collection of the data I could find and a few phenomena worth thinking about — purely personal observation notes.

Just how big is the layoff wave?

Look at the numbers first. In 2025, global tech layoffs hit 245,000 — already a historical high. 2026 is more aggressive: the first quarter alone cut more than 59,000 jobs, and at that pace the full-year total could reach 265,000.

A few representative cases:

Company	Layoffs	Notes
Oracle	~30,000	Largest in 2026
Amazon	~16,000	Continued "streamlining"
Dell	~11,000	Hardware-transformation pressure
Meta	2,000+ (potentially 15,000)	Big gap between public number and actual scale
Block	~4,000	Fintech isn't spared either

There's a key detail here: a CFO survey cited by Fortune indicates that companies privately admit AI-related layoffs are 9x larger than the public numbers. In other words, what we see may only be the tip of the iceberg.

Those hit hardest aren't just customer service and admin support — they include junior engineers and middle management. A Brookings study points out that nearly 11 million workers without degrees are in the high AI-exposure zone. These are the so-called "Gateway Jobs" that many people use to enter the tech industry.

The productivity paradox: is AI really that magical?

This is the part I find most interesting.

A study by Duke University in collaboration with the Federal Reserve found that the average self-reported AI productivity gain by companies is 1.8% — not much, even though companies often pitch it sky-high externally. More crucially, there's a clear gap between this self-reported number and actual statistical data.

This brings to mind what economist Robert Solow said back in 1987:

"You can see the computer age everywhere but in the productivity statistics."

Nearly 40 years later, swap "computer" for "AI" and the line still seems to apply.

From my own experience, AI does make some work faster — writing boilerplate code, looking up documentation, organizing notes, generating test cases. The efficiency gain on these is tangible. But if you ask me "what percentage has overall productivity improved by," honestly I can't give a precise number. A lot of the time, the time saved gets spent going back and forth correcting AI, checking for hallucinations, or dealing with new problems it created.

"AI washing": when AI becomes a convenient excuse

A San Francisco Standard piece used the term "AI washing" to describe how some companies use AI as a shield, when the actual reasons for layoffs are more complex — declining revenue, strategic pivots, or simple cost pressure.

It makes sense when you think about it. Telling investors and the board "we're embracing AI transformation and streamlining our workforce structure" sounds far better than "we lost money last quarter, we have to cut people." The AI-layoff narrative gives companies a forward-looking, plausible package.

This isn't to say AI has zero impact on employment — the impact clearly exists. But when "AI replacing labor" becomes a universal reason that doesn't need much explanation, we may need to add a few question marks.

Reversal signals: laid off, then what?

If you only look at layoff numbers, it's easy to fall into a pessimistic trap. But there are several signals worth noting:

Gartner forecast (February 2026): by 2027, 50% of companies that laid off staff due to AI will rehire workers for similar roles — just under different titles
CEO attitude shift: 67% of CEOs expect to increase entry-level hiring in 2026
New job postings emerging: LinkedIn data shows 1.3 million new AI-related positions added
Upskilling becomes the theme: 87% of business leaders rank workforce reskilling as a top priority

In other words, many companies' approach is "cut first, restructure, then rehire under new titles." This sounds absurd, but from an organizational change perspective, sometimes a clean reset is actually easier to push through than internal transformation — though for those laid off, the process is anything but pleasant.

Another data point: roughly 80% of the global workforce needs to acquire new AI skills within 12-18 months. Whether or not the exact figure is precise, the direction is right.

How engineers can think about this

As an engineer, my observation is this:

Short term, AI is genuinely redefining what work is "worth having a human do." Tasks that are highly repetitive, rule-based, and don't require much judgment are getting automated sooner or later. This isn't a trend unique to AI — AI just accelerated the process.

Medium term, "knowing how to use AI" by itself isn't a competitive edge, because soon everyone will. The real differentiator is whether you can judge the quality of AI output, whether you can define the problem, whether you can step in and fix things when AI screws up. To some extent, this is the same logic as "knowing how to Google" not being a competitive edge.

Long term, what concerns me more is how newcomers will enter the industry once those "Gateway Jobs" disappear. If junior engineering openings keep shrinking, the pipeline for growing senior engineers will be broken. This is a structural problem the entire industry needs to face.

A few things I'm doing myself:

Treat AI tools as part of the daily development workflow, not occasional new toys to play with
Deliberately practice the skills AI is bad at — system design, technical decision-making, cross-team communication
Stay curious about new tools, but don't blindly chase every hype cycle
Keep writing technical articles and shipping side projects so my abilities stay visible

Conclusion

The AI layoff wave is real, but a lot of it is bubble. Some layoffs really are because AI changed how work is done; others use AI as a pretext. The productivity paradox tells us AI's actual impact may not be as dramatic as the headlines suggest, and Gartner's forecast hints that many of these layoffs will end with "same role, new title, rehired."

For us engineers, rather than worrying about whether AI will replace us, it's more productive to focus on the abilities AI is bad at in the short term — understanding business context, making system-level judgments, and turning vague requirements into running architectures.

These abilities won't depreciate as AI advances; if anything, they become more valuable.