You Told GitHub You Couldn't Code Without It. They Took Notes.

GitHub's metered pricing change takes effect tomorrow. Copilot users will no longer pay a flat subscription rate. Starting June 1, they pay per token, the same way you pay for a taxi instead of a bus pass. The official announcement thread has accumulated 900 downvotes and more than 400 comments. Developers are unhappy.

They shouldn't be surprised.

For two years, GitHub optimized Copilot to be indispensable. It worked. Engineers wrote blog posts about 10x productivity gains. CTOs gave conference talks about how Copilot changed their shipping velocity. Individual developers told anyone who would listen that they couldn't code without it anymore, not as a casual preference but as a professional declaration. The message was consistent and very public: this tool is load-bearing.

Microsoft read all of that.

Token-based billing isn't a betrayal of developers. It's what you get when the people setting prices learn what a tool is actually worth to you. The flat-rate era was never charity. It was customer acquisition cost. GitHub's job during the subscription phase was to embed Copilot so deeply into developer workflows that switching carried a real cost. The fury in the TechCrunch comments section confirms that it worked.

The irritating thing is that developers helped build this case. Not just by using Copilot, but by publicly documenting their dependency. The user research that typically costs companies hundreds of thousands of dollars — how much do users rely on the product, what would they pay, what do they lose without it — was donated voluntarily via blog posts, tweets, and Reddit threads. Microsoft's product team didn't need to run focus groups. They just had to read Hacker News.

This follows a script that's been run before. Adobe's shift to Creative Cloud subscriptions in 2013 generated similar fury. Petitions appeared. A small number of professionals switched to alternatives and stayed there. Within two years, the vast majority accepted the new model, because the switching costs were real. Adobe had the files. The pricing model knew that.

GitHub Copilot has something more intimate than your files. It has your workflow. Developers who've used AI autocomplete for two or three years haven't just added a tool. They've modified how they think about code. The process of problem-solving has been reorganized around a system that proposes the next step. That's harder to walk back than learning new software. It's closer to reconditioning a reflex.

Which is exactly the bet GitHub was making.

There will be alternatives. Cursor has positioned itself as the anti-Microsoft option and has been gaining ground for months. Some developers will leave, and some of those will stay gone. But the ones who rebuild their workflow on Cursor should think about what they're setting up for themselves in two years, when Cursor's investors ask why the company isn't monetizing its users more aggressively. The cycle doesn't change because the logo does.

If you're mapping your options right now, we ran a detailed comparison of the best AI coding tools in 2026 — what each one costs and which workflows they actually fit.

The real question this week raises isn't about GitHub's pricing strategy. It's about what developers traded for three years of productivity gains. Speed, yes. Volume, yes. But also some portion of independence from a specific vendor's product decisions. The dependency was worth it while the pricing held. Now the pricing changed. The dependency remains.

That's not a pricing problem. It's a structural one.

The tools that make you meaningfully faster will eventually charge what you'd pay to stay fast. Not because the companies are malicious. Because the market is working correctly. The productivity gain you captured is also the price signal that tells them what you'll pay.

Tomorrow, June 1, the meter starts. For most developers, this will feel like an arbitrary corporate decision made by people who don't understand how they work. Some of it is that. But some of it is the conclusion of an argument developers made loudly and on the record: that they need this more than GitHub needs any one of them.

They were right. The bill arrived.

---

If you want more signal on where AI tools are heading before you trust your workflow to the next one, subscribe to About.chat Weekly. No fluff, just what's actually happening.