Spotify's Co-CEO confirmed it on the earnings call: the company's top engineers haven't manually written a single line of code since December 2025. This isn't hype — and it isn't a one-off.
Picture a Spotify engineer on their morning commute, phone in hand, typing into Slack: "Fix the bug in iOS navigation and add the new feature-flag logic." By the time they reach the office, a new version of the app is waiting on that same phone — ready to merge to production.
That's not science fiction. It's a concrete example that Spotify Co-CEO Gustav Söderström used on the February 10, 2026 earnings call to describe the company's new development reality. The headline that followed: Spotify's best engineers haven't manually written a single line of code since December 2025.
What's Actually Happening at Spotify
Behind that statement is an internal system called "Honk." It enables remote code deployment in real time, powered by generative AI — specifically, Anthropic's Claude Code. An engineer describes what they need. The system plans, implements, and tests. The engineer reviews, approves, and merges.
Klingt interessant?
The results speak for themselves: Spotify shipped over 50 new features and changes in 2025. Recent additions include AI-Powered Prompted Playlists, new audiobook features, and "About This Song" — all within weeks of each other. Söderström was direct on the call: "We don't see this as the end of AI-assisted development — only the beginning."
What's notable is what Spotify isn't saying. They're not talking about fewer engineers. They're talking about better work. More output. Higher velocity. The engineers aren't writing code — they're making decisions, defining requirements, and validating outcomes. The cognitive work that most developers actually find interesting stays with humans. The rest is delegated.
Spotify Isn't Alone
It's tempting to write this off as a quirk of a billion-dollar streaming company with its own AI research team. That would be a mistake.
In April 2025, Shopify CEO Tobi Lütke published an internal memo — and posted it himself on X. The core message: anyone requesting more headcount or resources now has to explain first why the task can't be handled with AI. "I have seen people tackle seemingly impossible tasks with reflexive and brilliant use of AI that has done 100x more work," Lütke wrote. AI usage now feeds directly into performance reviews.
Duolingo announced its "AI-first" strategy in late April 2025 and began systematically replacing contractor work with AI agents. Amazon, Google, and Microsoft have all confirmed internally that AI systems now generate significant portions of their newly written code.
The GitHub numbers back this up: GitHub Copilot had over 15 million users as of early 2025, is deployed across 90% of Fortune 100 companies, and internal studies show 51% faster coding and 84% more successful builds after rollout. This isn't an experiment anymore. It's infrastructure.
The Wrong Question
Most of the online debate circles around the wrong question: "Will developers be replaced?" Spotify has answered that implicitly — the answer is no. Their best engineers aren't writing code, but they haven't gone anywhere. They're operating at a higher level of abstraction.
The right question is: Who decides what level of abstraction a developer works at?
At Spotify, that's a deliberate company-wide choice backed by infrastructure (Honk), defined processes, and a real cultural shift. At Shopify, it's a CEO mandate. At Duolingo, it was a public strategic pivot. None of these companies stumbled into AI-first development by accident — each one built, trained, configured, and tested until it worked.
That's the invisible labor behind the headline "our engineers don't write code anymore."
What This Means for Companies That Aren't Spotify
Here's where things get uncomfortable for most established businesses. The principles behind Honk — specialized agents, human-in-the-loop validation, integrated quality gates, clear architecture guardrails — aren't technically exclusive to publicly traded tech giants. But building them takes time, deep expertise, and serious investment in AI infrastructure.
A mid-market manufacturer, a growing SaaS company, a digital services firm with 30 engineers: none of them can afford to spend 12 months building and calibrating an internal AI development system. But opting out of that system increasingly means watching competitors who have it pull steadily ahead.
That's the moment to stop admiring what Spotify built — and start thinking about how to access the same infrastructure without building it yourself.
That's exactly what nopex.cloud provides: a managed platform that makes this kind of agent infrastructure directly accessible — no need to assemble your own AI team, no vendor lock-in to a single model, with EU data residency and quality gates built in from day one. Not a prototype. Production-ready.
What Spotify's best engineers have been experiencing since December — more focus on the interesting problems, less time on boilerplate — isn't a question of budget anymore. It's a question of having the right infrastructure in place.
What this looks like in practice — without building your own AI infrastructure team: nopex.cloud
