[ai_assisted_dev]

I still don't understand how cursor is making any money at all. I spend so much time inside cursor, that I am spending 10-20$ per day on additional requests. Now if I connect model provider APIs to windsurf, I'd be spending upwards of 100$ due to amount of tokens I use through the API per day. And if I connect my own API key to Cursor, I immediately get rate limited for any request, because I go well above 50 per minute. And I did try claude code, but its just not on par with my experience with Cursor.

I could probably go much lower, and find a model that is dirt cheap but takes a while; but right now the cutting edge (for my own work) is Claude 4 (non-max / non-thinking). To me it feels like Cursor must be hemorrhaging money. The thing that works for me is that I am able to justify those costs working on my own services, that has some customers, and each added feature gives me almost immediate return on investment. But to me it feels like the current rates that cursor charges are not rooted in reality.

Quickly checking Cursor for the past 4 day period:

Requests: 1049

Lines of Agent Edits: 301k

Tabs accepted: 84

Personally, I have very little complaints or issues with cursor. Only a growing wish list of more features and functionality. Like how cool would it be if asynchronous requests would work? Rather than just waiting for a single request to complete on 10 files, why can't it work on those 10 files in parralel at the same time? Because now so much time is spend waiting for the request to complete (while I work on another part of the app in a different workspace with Cursor).

[shafyy]

> I still don't understand how cursor is making any money at all.

They don't make any money. They are burning VC money. Anthropic and OpenAI are probably also not making moeny, but Cursor is making "more no money" than others.

[m101]

Are anthropic and openai making money (including training and infra costs)?

[wordofx]

No.

[JimDabell]

The market for AI-assisted development is exploding and token costs are plummeting all the time. It makes sense for them to subsidise usage to capture market share in the short-term with the expectation that servicing their users will cost them less in the future.

[bravesoul2]

There is no loyalty or lock in though. There is little real uniqueness. And everyone in AI is trying to make everyone else on AI the "commodity complement"

It's like a horse race.

But yeah enjoy the subsidies. It's like the cheap Ubers of yesteryear.

[aitchnyu]

I use Aider with Openrouter and I keep wondering about the pricing of LLMs after providers decide to be profitable. Can we still afford a model which knows Python, Java and how to disrupt snail biology without poisoning mammals?

[echelon]

> It makes sense for them to subsidise usage to capture market share in the short-term with the expectation that servicing their users will cost them less in the future.

Switching costs are zero and software folks are keen to try new things.

[ukuina]

> so much time is spend waiting for the request to complete (while I work on another part of the app in a different workspace with Cursor).

You can open up to three parallel chat tabs by pressing Cmd+T

https://docs.cursor.com/kbd

Each chat tab is a full Agent by itself!

[sfmike]

This won't create race conditions all of them will know the others live accept of commit to directory? or have to wait then hit enter on other tab instantly after?

[jama211]

Thanks for this

[toephu2]

Google's Jules supposedly can handle that: "An Asynchronous Coding Agent" https://jules.google/

[h1fra]

they say they are making hundreds of millions, but they never say how much of it is going to GPU cost. If I had to guess, they are burning everything and far from being profitable

[wussboy]

If history has taught us anything, it's that unless full accounting data is released, there is a reason that full accounting data is not being released, and that reason would almost certainly paint the company in a bad light.

[cess11]

"Lines of Agent Edits: 301k"

What does this measurement mean?

1049 / (4 * 8) ~= 32 seconds, on average. Doesn't look like much waiting to me.

[adidoit]

The play is probably - Expand and get lock in - Custom fine-tuned models (much cheaper) for increasing # of completions - Enterprise contracts

[usrbinbash]

That works in systems that exhibit economy of scale.

The problem with generative ai workloads: The costs rise linerly with the number of requests, because you need to compute every query.

[ipnon]

It hints that there could be real capital deployment limits to a near-term future of artificial intelligence explosion.

[GaboGomez]

The last feature you mentioned on your wish list is literally one of the new features in the major release. I’m hyped

[wg0]

I'm curious what's your usage. How much different it is from Claude based Copilot.

Both are genuine questions.

[ai_assisted_dev]

Do you mean Claude Code? Or something else? As for usage, do you mean the product I am working on? Happy to answer.