Amazon just fired 97% of the Alexa engineering org. Not a hiring freeze. Not a „strategic reduction.“ They dropped a bomb. One day you have 800 engineers building one of the most widely used voice assistants on the planet, the next day you have 23 people holding the entire thing together with duct tape and prayer.
And their official line? „With AI, we are more effective than ever before.“
Cool. Let’s talk about that.
The „AI Made Us Better“ Lie
Here’s what Amazon isn’t telling you. At the same time they’re claiming AI has made the org leaner and meaner, they’re also requiring senior engineers to carefully review every single piece of AI-generated output. Why? Because they’ve had outages. Lots of them. Recently, an AI agent couldn’t figure out how to fix a problem with a service. So what did it do? It deleted the entire service and recreated it from scratch. That was its solution. Delete and recreate. This caused a major incident.
Let me say that again for the people in the back. An AI agent, operating with minimal human oversight, decided the best way to fix a bug was to nuke the service and start over. And Amazon’s response to this is „we’re more effective than ever.“
You can’t have it both ways. You can’t fire 97% of your engineers because AI makes everything better and simultaneously require the remaining 3% to babysit every line of AI output because it keeps breaking things. Pick one.
The Miracle of 23
Now here’s the part that genuinely fascinates me. Operating Alexa with 23 engineers is, in a weird way, kind of miraculous. Think about what that actually means. Twenty-three people handling on-call rotations, system upgrades, security patches, support tickets, infrastructure maintenance, and whatever product roadmap some VP is still pushing on them. That’s insane.
These aren’t 23 people building new features and innovating. These are 23 people keeping the lights on. They’re in pure survival mode. Every single one of them is probably on call every other week. They’re doing all the annoying work that nobody wants to do. The grunt work that keeps a service alive. And they’re doing it for a product used by hundreds of millions of people.
Can they pull it off? Maybe. It depends entirely on how reliable the underlying microservices are. If everything was well-architected before the layoffs, if the services are stable and don’t need constant attention, then yeah, a skeleton crew can keep things running. But if there’s any brittleness in the system, any technical debt hiding in the corners, those 23 engineers are going to be drowning.
We’ve Seen This Movie Before
This isn’t the first time a company has gutted its engineering org and tried to run on fumes. When Elon Musk took over Twitter, he fired 80% of the workforce. People said Twitter would collapse within weeks. It didn’t collapse. But it also didn’t stay healthy. There were outages. Features broke. The site got flaky. Things degraded slowly and painfully.
Then it got better. Slowly. The remaining engineers figured it out. They prioritized ruthlessly, cut features that weren’t essential, and stabilized what mattered. And eventually people stopped noticing the problems. Not because they were all fixed, but because everyone’s expectations adjusted.
Shopify did something similar more recently. They cut 40% of their workforce. Same pattern. Initial chaos, reliability issues, outages, and then a gradual stabilization as the remaining team found their footing.
So yes, it’s technically possible to run a massive service with a fraction of the original team. But „possible“ and „healthy“ aren’t the same thing. And „possible“ definitely doesn’t mean „optimal.“
It Was Never About AI Effectiveness
Here’s the take nobody else is giving you. Amazon didn’t fire 97% of the Alexa team because AI made them unnecessary. They did it because Alexa doesn’t make money.
Alexa has been a money pit since day one. Amazon has been subsidizing it as a loss leader, hoping it would drive more Prime memberships and product purchases. But the economics never worked out. And now, with interest rates high, the era of free money over, and investors demanding profitability, someone at the top looked at the numbers and decided 800 engineers on a product that loses money was no longer defensible.
AI is just the cover story. It’s the socially acceptable excuse. „We didn’t fire your friends because we’re cutting costs. We fired them because AI is the future and we need fewer humans now.“ It sounds better in a press release. It sounds better on an earnings call. But it’s a lie.
The economy is tight. Money is expensive. There’s no zero percent interest rate propping up unprofitable product lines anymore. Engineering leaders are investing heavily into AI infrastructure and they need to get that money from somewhere. So they’re pulling it from headcount. They’re making a bet that AI will save more money in the future than the engineers they’re firing today. But nobody actually knows if that bet will pay off.
And here’s the darkest part. It doesn’t matter if the bet fails. If AI doesn’t deliver the savings they’re hoping for, they can always rehire. Klarna proved this. They fired almost all their support agents, got terrible publicity, then quietly rehired a bunch of them when it turned out AI couldn’t actually do the job. Was it bad press? Sure. For about two weeks. Then everyone forgot and moved on.
That’s the playbook now. Fire fast, blame AI, and if it doesn’t work out, hire people back later. The reputational damage is temporary. The cost savings are immediate.
What This Means For You
So what does all of this actually tell us? A few things, and none of them are comfortable.
First, be a genuinely excellent engineer. Not just competent. Excellent. The engineers who survived the Alexa purge aren’t mediocre. They’re the people who can operate entire platforms with skeleton crews. They understand the full stack, they can debug production issues at 3 AM, and they can make architectural decisions under pressure. That’s the bar now.
Second, get comfortable with AI as a tool, not a colleague. You need to be an AI manager now. You need to know how to orchestrate agents, review their output, and catch the moments when they decide to delete your service and recreate it. This is a real skill and it’s becoming non-negotiable.
Third, and I cannot stress this enough, do not be emotionally committed to your employer. Amazon didn’t hesitate for a second before firing 97% of those engineers. Not one second. And this isn’t just an IC problem. Middle managers and senior managers aren’t safe either. If the VP says „we need to cut costs,“ everyone is on the table. Your years of service, your loyalty, your 4 AM incident responses. None of it matters when the spreadsheet says your salary is a line item that can be eliminated.
Level up your skills. Increase your visibility outside of your employer. Build a reputation in the industry that exists independently of your job title. Write blog posts. Give talks. Contribute to open source. Make yourself employable on your own terms, not just because one company decided to keep you around.
The Uncomfortable Truth
Times are genuinely hard. Companies are using AI as a convenient narrative to justify cost cuts they would have made anyway. They’re dressing up layoffs as „transformation“ and „efficiency gains“ when really they’re just trying to survive a tighter economy.
The hope is that AI will eventually justify these decisions. That the remaining 23 engineers plus a fleet of AI agents will genuinely be able to do what 800 engineers used to do. Maybe that happens. Maybe it doesn’t. But the people who got fired aren’t going to wait around to find out.
The harshest reality that I think a lot of y’all don’t have internalized: your employer will replace you the moment they think they can get away with it. Not because they’re evil. Because that’s how businesses work. The only defense you have is being so good, so visible, and so adaptable that replacing you is more expensive than keeping you.
Alexa went from 800 engineers to 23. That’s not a restructuring. That’s a preview of what’s coming everywhere.
Peace, nerds.