Airbnb Says AI Writes 60% of Its New Code — Are Developers Becoming Managers?

Last week on their Q1 2026 earnings call, Airbnb dropped a number that made the tech world do a double-take: 60% of their new code is now written by AI.

CEO Brian Chesky didn’t just slip it in as a footnote. He spent a good chunk of the call talking about it, explaining how AI has fundamentally changed how Airbnb builds software. And here’s the thing — Airbnb isn’t alone.

Google’s Sundar Pichai revealed that over 30% of new code at Google is now AI-generated. Microsoft CEO Satya Nadella said up to 30% of their code was written by AI. Spotify went even further — back in February, they said their best developers “haven’t written a line of code since December” thanks to AI tools.

Something’s shifting. Big time.

What’s Actually Happening at Airbnb?

Chesky was refreshingly specific about where AI is making the biggest impact. It’s not replacing their senior engineers — it’s supercharging them.

“API partners say they want to be better hosts and need better tools,” Chesky said. “Where you might have needed a team of 20 engineers before, an engineer can now spin up agents to do a lot of work under supervision.”

Translation: one good developer with AI agents can now do what used to take a small team. The bottleneck isn’t coding anymore — it’s supervision and architecture.

This aligns perfectly with what we covered in our earlier piece on Harnessing Agentic AI for Enhanced Workflows — where we talked about how AI agents are moving from simple task automation to orchestrating complex multi-step processes. Airbnb’s engineering team is a living example of that shift.

And it’s not just coding. Airbnb’s customer support AI bot now handles 40% of all issues without escalating to a human — up from 33% earlier this year. That’s real efficiency gains, not just hype.

How Other Tech Giants Are Using AI for Code

Airbnb’s 60% is the highest number we’ve seen publicly, but every major tech company is racing down this path:

Google: Over 30% of new code is AI-generated. Their internal AI coding tools help developers write, review, and debug code faster. Sundar Pichai has called it “one of the most significant productivity shifts we’ve seen.”

Microsoft: GitHub Copilot serves millions of developers worldwide, and internally, Microsoft says up to 30% of their code is AI-assisted. They’re using AI not just for writing code but for automated testing, documentation, and code review.

Spotify: The most dramatic shift — their best engineers, the ones building core features, simply don’t write raw code anymore. They describe what they want, review what AI produces, and focus on architecture decisions.

Anthropic: Their Claude Code tool lets developers use natural language to build, test, and deploy software. They recently partnered with SpaceX to provide AI coding capabilities for the aerospace industry.

What’s common across all of these? The winning companies understand that AI is becoming a natural interface layer — just like how we moved from command lines to GUIs, we’re moving from code to natural language instructions.

What AI Still Can’t Do

Despite the impressive numbers, Chesky was refreshingly honest about AI’s limits, especially in travel and e-commerce. He listed four specific problems with chatbot-based AI interfaces:

1. Too much text — e-commerce is visual, not text-heavy
2. No direct manipulation — you type instead of adjust sliders
3. Poor comparison — try comparing 1000 hotels in a chat thread
4. Single-player — most travel bookings involve multiple people

This is actually refreshing. A CEO who’s all-in on AI but knows exactly where it breaks. It means AI for code works great — it’s a deterministic, reviewable output. But AI for consumer-facing interfaces? Still a long way to go.

What This Means for Developers

If you’re a software engineer reading this, you’re probably wondering: am I cooked?

Probably not. But the job is changing.

Think about it this way: when calculators became common, mathematicians didn’t disappear. They just stopped spending time on arithmetic and started focusing on what the calculations meant. Same thing here.

The developers who’ll thrive are the ones who:

• Understand system architecture deeply — AI can write functions, but it can’t design coherent systems
• Can review and debug AI-generated code — AI code is fast but not perfect; human judgment is the bottleneck
• Focus on product thinking — the “what should we build?” question is more valuable than “how do I write this?”
• Use AI as a lever, not a crutch — the best developers use AI to go faster, not to skip understanding

And here’s the interesting flip side: AI is also making coding accessible to non-programmers like never before. Tools like Copilot, Claude Code, and Cursor let people who’ve never written a line of code build functional software. This isn’t just disrupting what developers do — it’s expanding who can be a developer.

The Bigger Picture

None of this should be surprising. Every major platform shift — from the web to mobile to cloud — has followed the same pattern: automate the routine, elevate the important.

AI coding is just the latest iteration. The companies winning are the ones understanding that AI doesn’t replace judgment — it amplifies it.

The developer’s job in 2026 isn’t to write more lines of code. It’s to decide which code matters, what to build, and how to guide the AI toward building it right.

So if you’re building software today, the question isn’t “should I use AI to write code?” — that ship has sailed. The question is: what do you want to build with the leverage you now have?

What do you think — excited or worried about AI writing your code? Drop a comment below.