I would not count on Gitlab saving Cursor. If anything, the other way around.
Cursor has the install base to get very good (and is probably doing this!) at being smart about context window management when decomposing larger problems that deal with complex codebases. One thing to remember is that the context window rag optimization stuff has to be pretty tuned on a per LLM basis. Given they already are working on "let Cursor pick the model", they could hyperspecialize towards the ones they have the most experience with and not have to solve for all of them each release which really can slow things down.
This is similar to what I was thinking, a match made in heaven. Gitlab as it is right now is likely doomed with no one to support them from the AI side. They are manifestly bad at it.
Microsoft has so many advantages, and there is nothing disruptive about Cursor—meaning Microsoft can copy them and has no reason not to.
For Cursor to win, it needs to find a way to make it so that when Microsoft tries to copy them, they have to kill a multi-billion dollar business. This is what Android did to Windows Phone — they took the entire OEM business away, and the only way for MS to complete was to make the OS free (and open), which was not an easy choice to make.
I think this will turn from a tools battle into a platform battle from here, and at that point Cursor is toast. Meanwhile gitlab is having trouble innovating with AI, similarly have half of the puzzle, seems like a match made in heaven :).
I would not count on Gitlab saving Cursor. If anything, the other way around.
Cursor has the install base to get very good (and is probably doing this!) at being smart about context window management when decomposing larger problems that deal with complex codebases. One thing to remember is that the context window rag optimization stuff has to be pretty tuned on a per LLM basis. Given they already are working on "let Cursor pick the model", they could hyperspecialize towards the ones they have the most experience with and not have to solve for all of them each release which really can slow things down.
This is similar to what I was thinking, a match made in heaven. Gitlab as it is right now is likely doomed with no one to support them from the AI side. They are manifestly bad at it.
Microsoft has so many advantages, and there is nothing disruptive about Cursor—meaning Microsoft can copy them and has no reason not to.
For Cursor to win, it needs to find a way to make it so that when Microsoft tries to copy them, they have to kill a multi-billion dollar business. This is what Android did to Windows Phone — they took the entire OEM business away, and the only way for MS to complete was to make the OS free (and open), which was not an easy choice to make.
I think this will turn from a tools battle into a platform battle from here, and at that point Cursor is toast. Meanwhile gitlab is having trouble innovating with AI, similarly have half of the puzzle, seems like a match made in heaven :).
I think that would help, but maybe as a better acquisition target. In the end, they either need money or a business model that MS can’t copy.