Can Cursor Disrupt GitHub Copilot?

I switched from Copilot to Cursor because it’s better and was easy to adopt, but it’s not a disruption, and I expect Microsoft to catch up.

Rick Minerich just wrote a piece about how Microsoft is trying to compete with Cursor on AI-assisted coding.

Microsoft also clearly has a major potential advantage with GitHub itself. Github hasn’t yet been substantially exploited. Clearly AI for the rest of the dev lifecycle is coming, and there are throngs of AI startups foaming at the mouth for a piece of that. Will Github be able to deliver in a major way there, or will they have to open up yet another front in the AI dev tooling wars?

This is true and what I was getting at two years ago when I wrote Large Language Models are a Sustaining Innovation:

In the Innovator’s Dilemma, Clay Christensen made a distinction between “disruptive” and “sustaining” innovations. Disruptive innovations favor new entrants and new categories of products because incumbents are harmed by adopting the innovation, and so they resist. Sustaining innovations favor incumbents because it improves their products and margins, and they have the resources and incentives to adopt them.

The problem for Cursor in competing with Microsoft is that Microsoft has no disincentive to follow them. This is in contrast to when Google killed their phone OEM business by releasing Android for free. To follow Google, Microsoft would have had to give up a multi-billion dollar business. This is what disruption looks like—the incumbent can’t simply copy the new entrant.

But, in this case, Microsoft can easily follow and because Cursor is easy to switch back from, there is actually no advantage to Cursor’s land grab. I went from VSCode with Copilot to Cursor in 20 minutes, and I could go back faster. I can run them in parallel.

Here are Microsoft’s other incumbent advantages:

1. Infinite money

2. Azure (gives them at-cost compute)

3. Experience with AI engineering (built up from years of working with OpenAI)

4. The relationship with OpenAI which gives them low-cost models

5. 50 years of proprietary code (could this augment models?)

6. Developer Tools expertise (and no one is close — maybe JetBrains)

7. GitHub (as Rick pointed out)

8. Control of Typescript and C#

9. Control of VSCode (which they are flexing)

Rick has also written about how the programming language matters for AI-assistance and specifically pointed out how poorly they do at C#. Microsoft is uniquely qualified to fix that and has every reason to do so.

To see how these advantages compound, imagine Microsoft offered every GitHub private repo owner money to use their code for training purposes and were able to overcome the language popularity problem. Access to private or proprietary data is a moat for building LLMs.

For Cursor to compete, they would need to do something that causes Microsoft to kill a multi-billion dollar business when they follow. In my piece two years ago, I suggested that would be no-code solutions. But Microsoft clearly has an interest here (e.g. Excel and the Power platform) because the non-programmer knowledge worker market is orders of magnitude bigger than the developer market. So even that might not work.

Right now, I don’t see how Cursor can win unless they have truly unique technology that Microsoft can’t copy. The only argument that I can see for why this might be true is that it’s taken so long for Microsoft to match them.