Spotify’s CEO told Wall Street this week that his best engineers haven’t written a single line of code since December. They just prompt AI and review what comes out.
This isn’t a startup flex. This is a company with 600 million monthly users telling investors that code-writing is officially a machine’s job. And the implications go way beyond Spotify.
The Supervisor Era Is Here
Gustav Söderström didn’t hedge. His most senior engineers “only generate code and supervise it.” They’ve gone from architects to inspectors — and he’s thrilled about it.
The wildest part? He says the bottleneck isn’t talent, compute, or AI capability. It’s how fast users can absorb change. That’s a constraint no tech company has ever operated under before. The factory can now produce faster than customers can consume.
He also dropped a line that should make every PM nervous: “The things you build now may be useless in a month.”
AI Made Engineers More Productive — and More Miserable
The same week Söderström was celebrating, software engineer Siddhant Khare published an essay that went viral for a different reason: “AI fatigue is real and nobody talks about it.”
His experience is a paradox. He shipped more code last quarter than ever before. He also felt more drained than ever before.
The problem isn’t output — it’s cognitive overhead. When every task “only takes an hour with AI,” you end up juggling six tasks instead of one. The AI doesn’t get tired between problems. You do.
Harvard Business Review backed this up with an eight-week study: AI tools didn’t reduce work. They intensified it. Once the novelty fades, workers discover their workload has quietly ballooned, leading to burnout and worse decision-making.
One Hacker News commenter joked about doing squats while waiting for Claude Code to finish generating. Others described their deep focus skills atrophying in real time. It’s funny until you realize it’s not.
Half of Big Tech Engineers? Gone.
Steve Yegge — Amazon veteran, 12 years at Google, author of the vibe coding book — made his prediction on The Pragmatic Engineer podcast: 50% of Big Tech engineers will be laid off.
His logic is pure economics. Companies are pouring money into AI tokens, GPU clusters, and enterprise licenses. That capital has to come from somewhere. Labor budgets are the answer. You cut engineers to fund the tools that make the survivors productive enough to compensate.
Meta’s Zuckerberg says one engineer now does the work of a whole team. Microsoft’s Mustafa Suleyman claims most white-collar work will be “fully automated” within 12-18 months.
When the podcast host pointed out that 50% cuts would dwarf pandemic-era layoffs, Yegge didn’t blink: “It’s going to be way bigger. It’s going to be awful.”
Vibe Coding Sounds Fun Until You Need Fire Suppression
“Vibe coding” — Collins Dictionary’s 2025 Word of the Year — conjures images of casually describing apps into existence. The reality is messier.
As The Verge put it, AI coding is like “pair-programming with a savant intern — competent yet oddly deferential, still a tad too eager to please.” It crushes isolated tasks. Parallelize a dozen operations? Flawless. Build a complex integrated feature? You might get “a cockpit-shaped death chamber with a nonfunctional dashboard.”
The gap is integration. Large codebases are cities. Vibe coding builds great pop-up shops. But connecting fire suppression systems across all floors so they trigger in the right sequence? That still requires deep engineering intuition AI can’t replicate.
This creates a brutal irony: mid-level engineers who write the most code are most at risk. The survivors will be senior architects who understand systems well enough to direct the machines — exactly what Spotify’s top devs are already doing.
Three Stories That Can’t All Be True
CEOs see liberation. Veteran engineers see a culling. Working engineers see burnout. The AI keeps getting better, compressing the timeline for all three.
What makes this moment different from previous automation waves is speed. Manufacturing automation took decades. Software eating the world took a generation. The transformation of the people who build the automation? Months.
If you write code for a living: learn to work with AI or get replaced by someone who does. But set boundaries — the research is clear that unchecked AI-assisted productivity leads straight to burnout.
If you’re in any other knowledge work: software engineering is the canary. If the people who build AI tools are being disrupted this hard, your role is next.
If you use Spotify to listen to music: you won’t notice a thing. The playlists still play. But behind the scenes, the people who built the app are living through one of the most profound professional shifts in modern history.
The question isn’t whether vibe coding transforms development — it already has. The question is whether we can find a version of this that doesn’t burn everyone out on the way there.