In 2025, the AWS region US-EAST-1 went dark for 15 long hours.
Estimated losses of about $72 million per hour. That’s the annual salary of 288 software engineers.
Wasted. Per hour.
To put that into perspective, Valve makes blockbuster games 130M+ people play.
With only about 180 devs.
So let’s imagine what a few days of downtime would cost.
Think of bank transfers that start bouncing. Planes that stop flying. Empty food shelves in supermarkets.
Like a zombie apocalypse.
But in real life.
All because some brain-dead CEO thought the average Joe did nothing except to look at a screen and copy-paste. All while having too much coffee.
AI can do his job. Without the free coffee. Right?
Sure, Joe was “just” looking at a screen.
But he also knew why the checks were bouncing. Why the shelves were emptying. And why the planes won’t take off.
He also knew HTTP. Database migrations. Memory usage.
And a gazillion other details about this thing called the Internet. The same platform on which the CEO who fired him is using to run their business.
Which 6 billion people, or 73.6% of the world population, also uses.
And which no one seems to care about anymore.
They are all on a new drug.
Generative AI.
(Do you know you still use the internet, even after you die? Yes, your data will still be out there on some server. They might even train an AI on it.)
They don’t teach those gazillion details in business school.
Which is why disasters must happen.
So the “business idiots” up in the financial department finally wake up. And learn one simple lesson:
Input = Output.
Shitty AI slop gets you shitty, buggy software.
You can’t make good money selling shitty software.
Unless your name is Microsoft. But there can only be a few Microsofts (even they are getting their ass kicked, with a record of people switching to Linux).
Then, it will become obvious.
To write reliable software, you need competent engineers.
When you start hearing about web infrastructure crumbling like a house of cards. About people being sent home because the “computers are not responding”.
That’s the sign the hiring spree is coming.
And what do you need to do to be ready?
No, you don’t need to “learn AI”.
Because despite the hype, there isn’t much to learn about AI coding. LLM companies haven’t integrated most “hacks” into their products.
You also don’t need to “transition from coder to architect”.
Can you architect systems that you can’t code?
If you think yes, then please share your drug dealer’s address. I also want to try the stuff you are on.
Such levels of psychosis confirm that yes, collapse is about to happen.
But I can’t tell you exactly when the bubble will burst. Or what will trigger it.
And honestly, it doesn’t even matter.
Because you are a developer, not a private credit investor. You write code and build software for a living.
Luckily, the only one thing you have to do is to survive the next few months.
Position yourself to win.
Ramp up your technical interview skills. Polish your fundamentals.
Then, hit the market when it’s getting hot. At full speed.
Negotiate hard. Job hop.
Go to the most desperate company paying the highest salary (hint: it’s gonna to be one of those big fellas pushing AI in each of their products).
The software industry is a cycle.
Senior devs are just developers who survived the dark times. So they can be around when the sun rises again.
It always does. But you need to stick around.
We can help you make the most out of this “bad market”.
Check if you qualify here.
Stay sane,
Dragos
P.S. The average “Joe” can be, and is most likely, a “Jane”. Women represent an ever increasing number of developers. Blame it on maternity leave, sexism, or else, they’ve been affected the most by AI layoffs. Hang in there, ladies. Revenge is best served cold, and with a pinch of empty data centers for LLMs that “failed to meet investor expectations”.
P.S. No AI has been used to write any part of this email because life is too short for that sh**t.
