Everyone’s panicking about the wrong number
When Google CEO Sundar Pichai announced that 75% of all new code at Google is now AI-generated, the internet did what it always does — it spiraled. Developers worried about their jobs. Tech commentators declared the beginning of the end for human programmers. But here’s what most of those hot takes missed: this number is actually a sign that AI is working with engineers, not replacing them.
Let’s slow down and look at what’s actually being said.
What Google Actually Said
On the first day of Google Cloud Next, Pichai shared that about 75% of all new code at Google is now AI-generated and approved by engineers. That’s up from 50% just last fall. Alphabet has also set its 2026 capital expenditure between $175 billion and $185 billion, a figure that signals just how seriously the company is betting on AI infrastructure.
Two words in that announcement are doing a lot of heavy lifting: approved by engineers. The AI isn’t shipping code on its own. Humans are still in the loop, reviewing every line before it goes anywhere. That’s not a minor detail — that’s the whole story.
Why Non-Technical People Are Reading This Wrong
If you’re not a developer, “AI writes 75% of the code” probably sounds like “75% of programmers are about to lose their jobs.” That’s an understandable read, but it’s not quite right.
Think of it this way. A lawyer who uses AI to draft a contract still needs to read it, edit it, and sign off on it. The AI speeds up the drafting. The lawyer’s judgment is still what makes it usable. Google’s engineers are in a similar position. The AI handles a lot of the repetitive, structural, or boilerplate writing. The engineer decides whether it’s actually correct, safe, and worth keeping.
What’s changing is the ratio of thinking to typing. Engineers are spending less time writing syntax and more time making decisions. For most developers, that’s not a threat — it’s a relief.
The Jump from 50% to 75% Is the Real Story
Going from half of new code being AI-generated to three-quarters in less than a year is a fast move, even by tech standards. That kind of acceleration tells us a few things:
- Google’s internal AI tools are getting good enough that engineers are actually trusting them more
- The workflows around AI-assisted coding are maturing — this isn’t experimental anymore
- Other large tech companies are almost certainly watching this number very closely
The speed of adoption inside one of the world’s largest engineering organizations is worth paying attention to. Google employs tens of thousands of software engineers. If three-quarters of their new code is now coming from AI tools, that’s not a pilot program. That’s a new way of working.
What This Means for People Learning to Code
This is the question I get most from readers on this site, and it’s a fair one. If AI can generate most of the code, should beginners even bother learning?
Yes — and the Google stat actually supports that answer. The engineers approving that AI-generated code still need to understand what they’re looking at. You can’t review something you don’t understand. The skill set is shifting, but it isn’t disappearing. Knowing how to read code, spot errors, and make judgment calls about quality is arguably more valuable now, not less.
AI tools are raising the floor for what one person can build. A solo developer with solid AI assistance can now produce what used to take a small team. That’s genuinely exciting for anyone just starting out.
The Part Nobody Wants to Say Out Loud
There is a real conversation to be had about what happens when that number climbs higher — to 85%, 90%, or beyond. At some point, the ratio of human review to AI output becomes a question worth asking seriously. How much can one engineer meaningfully review? What happens to code quality if the review process gets rushed?
Google hasn’t answered those questions publicly, and the verified facts we have don’t tell us. But they’re the right questions to be asking, and they’re a lot more useful than panicking about a statistic taken out of context.
For now, what we know is this: AI is writing more of Google’s code, humans are still approving it, and the company is investing enormous resources to keep pushing that forward. That’s not a crisis. That’s a company figuring out, in real time, what the next version of software development looks like.
And honestly? It’s pretty fascinating to watch.
🕒 Published: