The Label Changed: Same Layoffs, New Excuse (For Now)
Layoffs.fyi tracks tech job cuts and flags the ones companies attribute to AI. In the first quarter of 2024, ~57k tech workers were laid off, and ~8k of those cuts were put down to AI. In Q2 this year, the total was almost the same at ~56k, but ~49k were attributed to AI.
Chart 1: Layoffs.fyi, AI-attributed vs other tech layoffs by quarter
We’re not really seeing a sudden wave of new layoffs, it’s much the same volume, but AI is now being blamed.
For most of 2024 and 2025, AI wasn’t really blamed at all. Then in the Q4 2025, around the time Amazon announced its cuts, it jumped to two-thirds of the total, and it has stayed there.
OK, things are moving insanely quickly, but this makes me suspicious. When Musk cut around 80% of Twitter in late 2022, a month before ChatGPT launched, he called it bloat. But today, that would almost certainly be blamed on AI.
I am not saying AI isn’t to blame at all, but it’s clear the label on its own tells us very little. As I flagged in my last piece, the market loves an AI story. Blaming AI is a win win. It allows you to cover for your over hiring while getting a pat on the back from the markets.
So, I sent Claude to have a look at the companies’ filings to look at RPE and see what the data shows.
Chart 2: Real RPE, Indexed 2020
Most of these cuts are too recent to show up in numbers yet. But the numbers do show the trend going in is positive. Adjusted for inflation, several of these companies follow the same, expected, shape. Efficiency dipped during ZIRP as they over hired, then recovered as they started to cut. Amazon’s RPE has risen steadily since 2022. Block’s RPE rose by about two-thirds in real terms between 2023 and 2025, before its AI announcement. So, if or when these companies post big jumps in RPE this year, not all of that will be AI.
Workday cut about 8.5% of its staff in early 2025, while continuing to hire in some areas and acquired a few AI companies. A year later it had around 21k employees, 600 more than before the cut, and its real RPE was only just back above where it started in 2020. Autodesk cut about 9% at the same time and did see a jump, though part of that came from a change in how it sells and records revenue.
It’s clear AI will make companies more efficient, and you can already see some examples. Salesforce says its support headcount has fallen from around 9k to 5k as AI handles more of the work. But right now, it’s difficult to attribute AI replacement while we deal with the hangover from ZIRP. I think that will take some time to flush through.
Most of what’s being called an AI layoff still looks like cost-cutting, whether that be undoing pandemic hiring or freeing up cash for other priorities. The companies making the largest cuts are betting on efficiency they expect to get but they’re claiming they already have it.
Block and others may post big jumps this year if revenue keeps growing while headcount stays down. If they do, then the bet is paying off and AI deserves the credit. But if the headcount creeps back, the label was doing the work.
I’ll check back in a year, and let you know!
Sources: