Amazon cut about 14,000 corporate jobs yesterday. Biggest layoff round in the company's history, and the memo that went out to employees leaned hard on a word I've been chewing on ever since: efficiency. Not "we overhired." Not "the market changed." Efficiency, like it's a virtue you build, not a euphemism you reach for when you need to explain thousands of people losing their jobs a week before the holidays.
I don't work at Amazon and never have. But I've got a friend in Seattle, mid-level PM on the devices side, who's been there six years, and she found out at 6:40am when her badge stopped working before the email even fully loaded. Not fired for performance. Not part of some struggling division. Just... efficiency. Andy Jassy has been saying for months that AI would let the company do more with fewer people, and now here's the receipt. Reuters and a few other outlets put the number at roughly 14,000 corporate roles, with reporting suggesting the total could climb toward 30,000 across additional rounds before the end of the year. HR, devices, some of the AWS-adjacent teams. The stock, last I checked, didn't really flinch.
Here's the thing that bugs me about the framing, and I say this as someone who genuinely likes what AI tools can do for a solo developer or a small blog like this one. When a company says "AI-driven efficiency," it's doing two jobs at once. It's telling investors "look how modern and lean we are," and it's telling the laid-off employee "this wasn't personal, this was inevitable." Both of those things can't really be true at the same time. If it's inevitable, it's not a strategy, it's just gravity, and nobody applauds gravity on an earnings call. If it's a strategy, then somebody chose it, and the language exists specifically so nobody has to say "we chose this."
I've been running this blog since November 2011, and I've watched enough corporate-speak drift through this industry to know the pattern by now. "Right-sizing." "Restructuring for the future." "Building a leaner organization." Every cycle the euphemism gets a fresh coat of paint and every cycle it means the same three things: some VP's budget got cut, some product got deprioritized, and a few thousand people who were doing fine work last Tuesday are doing something else by Friday. What's new this time isn't the layoff, it's that AI gets to be the noun doing the layoffs, grammatically speaking, instead of the humans who signed off on the org chart.
My friend's severance, for what it's worth, is decent by tech standards, something like 20 weeks base pay plus the standard COBRA subsidy, and she's not panicking exactly, she's annoyed. Mostly at the timing. Late October, right when a lot of hiring freezes for the year, right before the exact quarter when everyone's budgets reset in January and nobody's actually posting new roles yet. She'll be fine, she's good at what she does and she's got savings, but "fine eventually" and "fine right now" are different sentences and the corporate memo doesn't distinguish between them.
I keep thinking about the guy who has to write these memos for a living, some poor comms director drafting the sixth version of "we're grateful for your contributions" this year across the industry. There's a whole subgenre of corporate literature now that exists purely to make a bad thing sound like a competent thing, and I'd honestly respect it more if a company just said "we hired too fast in 2021 and 2022 and now we're paying for it, sorry." That's a sentence a human being could say out loud without flinching. "AI-driven efficiency gains" is not.
Anyway. If you're reading this and you got one of those emails yesterday, from Amazon or anywhere else, I hope the severance math works out better than the memo made it sound. It usually takes longer than anyone wants, and it usually goes better than the first panicked week suggests. Small comfort, I know, but it's the honest kind.