That's quite a war chest to raise in such a short period of time. The market has concentrated with the couple of recent acquisitions. Sounds like likely uses of that cash are
- Buy a few million more users with more generous free tier, as models get cheaper the cost to acquire the marginal customer goes down over time anyway
- Build your own foundation model for coding. tbh I'm skeptical that a company can do this better than the Big 3 AI cos.
- Go to war over enterprise. Do a deal with Deloitte/Accenture and get every single one of their consultants spending 8hrs a day in Cursor. Another flavor: compete head-on with Accenture by making your own service firm that undercuts them and delivers ahead of schedule for once.
> Build your own foundation model for coding. tbh I'm skeptical that a company can do this better than the Big 3 AI cos.
They already have their own "tab" model which might not be a very large one but definitely better than most open weight models on short code snippet completion. And for larger agentic LLMs, they can totally start from a pre-trained base model (e.g. deepseek R1) and only do post-train/RL/finetune, which is doable with a small team and their cash reserve.
It's not hard to imagine a good base model (probably deepseek V4?) + cursor's user data leads to a model that surpass sonnet/gemini on coding tasks only.
Work in consulting. The entire industry has flipped on its head as rates(interest and consulting) have collapsed. Most of the non-techie and somewhat-but-dangerous people(those of us left...) are now spending most of their time vibe coding demos and early prototypes.
What is it with western white collar (tech) workers and not using first person pronouns? This sentence took me a couple tries before I realized the “I” was missing.
I see it a lot on this site as well as my own company.
> Go to war over enterprise ... consultants spending 8hrs a day in Cursor.
And students. Sun's Java push, especially its proliferation as "object-oriented programming language" in CS courses world-wide, might offer a lesson or two.
I've gotten an AI generated email back when I asked for support, which did seem to understand my request at a surface level, but was not actually helpful. That's the extent of support I've received from cursor.
Re: support, no luck for me either. I even responded to a personal message I got from someone about a glitch with my signup and never heard back. I have no idea if I’m actually paying them.
“Please don't hesitate to reach out if there's anything we could do to improve Cursor for you. Making Cursor phenomenal for our pro users is our number one priority.
I predict that in 6-12 months we'll all be back on VS Code. I would hate to have Microsoft as a direct competitor, especially in a space they care so much about (developers + AI).
This thesis has existed since Cursor first started, and the gap between them and VSCode has only widened since then. It’s worth spending some time thinking about why that may be before having such strong conviction about their demise.
You can't really name a list of features that cursor has that copilot doesn't. It's more like: Cursor appears to heavily dogfood their features, VSCode's copilot seems to check the feature boxes, but each one sucks to use. The autocomplete popups are jarring. The copilot agent doesn't seem to gather the correct context. They still haven't figured out tool calling. It's really something you have to try rather than look at a checklist of features.
I think your knowledge is a bit outdated? Cursor definetley still has an edge, but VSCode Github Copilot UI has come a long way and using the same underlying models for both the results are fairly similar and change only in ux niceties
I tried copilot agent like 3 weeks ago. If that much has changed since then, props to Microsoft.
Zed is very nice, it’s just a totally different workflow. I think people who work in a domain where AI is not particularly strong would be better off with Zed, since Cursor’s way of reviewing edits is a little clumsy.
What about on the speed front? VS Code's biggest problem is with how slow it is. I'd already be done and on to the next (and maybe the next thing after that) by the time it finally gets around to things. I like the concept, but I only have so much time in a day.
Their main supplier (Anthropic) is also a direct competitor (Claude Code). I love Cursor but boy, what a tough place to be in. It's hard to see how it works out for them in the end.
Code isn't as strictly competitive as the IDE's are. Code even has solid VS Code integration. It's effectively a plugin, just one that isn't tied to any one IDE.
I use Roo Code. I love it. Are Cursor or Copilot or any of the other "front ends" so much better? I guess it's up to me to find out but wonder what others have found. [edit: grammar]
Since Codex web ui came out I stopped using Cursor and just direct pull requests on Codex web interface. I love it so much, I belive most people will move to this kind of development as models are getting stronger, and the whole agent+user workflow will switch to pull request based development.
It’s not like I’m not using Cursor at all, it just became the 10-20% of my workflow compared to almost 100% before.
Ok so OpenAI owns Windsurf now, so I would expect them to cut Cursor off from the really good stuff at some point. Anthropic has Claude Code, they could do the same. Google is a little farther behind but they do have an AI Studio thing that could be viewed as competitive.
Feels like Cursor has to make their own models to guarantee long-term survival? Especially if they’re not going for an acquisition (reports are they turned down OpenAI). Can they make a model that’s good enough for a world where OpenAI / Anthropic / Google all cut them off?
Good to see with that war chest they'll be competing in this space for some time yet - which we'll need more of to keep the pricing down (aka subsidized).
Never tried Cursor since I'm not prepared to leave my IDEs, but still got a full AI toolbox with augmentcode.com, GitHub Copilot, Claude Code and Gemini Code Assist enabled in both my primary VS Code and Rider IDEs.
$900 million?! That's an absolutely insane amount of money. How much of this will go into building the product itself, and how much into converting cash to marketshare? And they raised $100 million not even 5 months ago. Have they already burned through that pile of money?
I mean, their product is good – I'm using and paying for Cursor – but not fantastic. And there's a lot of competition. And the switching cost is relatively low.
they are approximately a service for transferring revenue from paying users to Microsoft or Anthropic, with a small software development project alongside - why would that seem like an insane amount of money?
I'm assuming a lot is coming from companies where there's fomo that if you don't use the latest and greatest tooling, your competitors will and you'll be left behind.
That's quite a war chest to raise in such a short period of time. The market has concentrated with the couple of recent acquisitions. Sounds like likely uses of that cash are
- Buy a few million more users with more generous free tier, as models get cheaper the cost to acquire the marginal customer goes down over time anyway
- Build your own foundation model for coding. tbh I'm skeptical that a company can do this better than the Big 3 AI cos.
- Go to war over enterprise. Do a deal with Deloitte/Accenture and get every single one of their consultants spending 8hrs a day in Cursor. Another flavor: compete head-on with Accenture by making your own service firm that undercuts them and delivers ahead of schedule for once.
> Build your own foundation model for coding. tbh I'm skeptical that a company can do this better than the Big 3 AI cos.
They already have their own "tab" model which might not be a very large one but definitely better than most open weight models on short code snippet completion. And for larger agentic LLMs, they can totally start from a pre-trained base model (e.g. deepseek R1) and only do post-train/RL/finetune, which is doable with a small team and their cash reserve. It's not hard to imagine a good base model (probably deepseek V4?) + cursor's user data leads to a model that surpass sonnet/gemini on coding tasks only.
> Deloitte/Accenture and get every single one of their consultants spending 8hrs a day in Cursor
Unless you change Cursor into a busy work assistant it is functionally useless for consultants.
Work in consulting. The entire industry has flipped on its head as rates(interest and consulting) have collapsed. Most of the non-techie and somewhat-but-dangerous people(those of us left...) are now spending most of their time vibe coding demos and early prototypes.
> Work in consulting
What is it with western white collar (tech) workers and not using first person pronouns? This sentence took me a couple tries before I realized the “I” was missing.
I see it a lot on this site as well as my own company.
I don't think it's a tech thing, this is a pretty common form in english like "Went to the store. Got some butter."
I think it might feel more common in tech because it's more common in digital communications.
> Go to war over enterprise ... consultants spending 8hrs a day in Cursor.
And students. Sun's Java push, especially its proliferation as "object-oriented programming language" in CS courses world-wide, might offer a lesson or two.
Students aren’t a market though. Sun and Java was really just marketing to future users
Isn’t marketing basically spending money now that will result in revenue later?
Future market?
Java only became Java because of enterprise. Getting students to use it was just an exercise in soft power.
Why do you think enterprise loves it so much?
Inertia.
Wow only a 21.6x rev multiple. The others a month ago were 75x for windsurf acquisition by OpenAI and a 45x on cursors $200M raise at $9B val.
Separately, has anyone gotten through to cursor support? They sent me a welcome email asking for feedback but when I responded nobody answered back.
Edit: added old financing info.
I've gotten an AI generated email back when I asked for support, which did seem to understand my request at a surface level, but was not actually helpful. That's the extent of support I've received from cursor.
I assume this is our future for almost all support for every product and service.
Whether it's this or call centers, at the end of the day...
And let's not forget PLTR still hovering just over 100x revenue multiple.
They probably aren't wanting for cash so I wonder why they did this. Maybe a lot of it was secondary sales?
On the contrary, they are lighting piles of cash on fire. Their unlimited slow pool is bleeding them dry.
Re: support, no luck for me either. I even responded to a personal message I got from someone about a glitch with my signup and never heard back. I have no idea if I’m actually paying them.
“Please don't hesitate to reach out if there's anything we could do to improve Cursor for you. Making Cursor phenomenal for our pro users is our number one priority.
Best, The Cursor Devs ”
:(
I predict that in 6-12 months we'll all be back on VS Code. I would hate to have Microsoft as a direct competitor, especially in a space they care so much about (developers + AI).
This thesis has existed since Cursor first started, and the gap between them and VSCode has only widened since then. It’s worth spending some time thinking about why that may be before having such strong conviction about their demise.
> the gap between them and VSCode has only widened since then
What is in this gap? Do you know of any good resources that outline the features that Cursor provides over VSCode with Copilot?
You can't really name a list of features that cursor has that copilot doesn't. It's more like: Cursor appears to heavily dogfood their features, VSCode's copilot seems to check the feature boxes, but each one sucks to use. The autocomplete popups are jarring. The copilot agent doesn't seem to gather the correct context. They still haven't figured out tool calling. It's really something you have to try rather than look at a checklist of features.
I think your knowledge is a bit outdated? Cursor definetley still has an edge, but VSCode Github Copilot UI has come a long way and using the same underlying models for both the results are fairly similar and change only in ux niceties
stuff like background agents cursor is way ahead.
Zed Editor is a nice contender too
I tried copilot agent like 3 weeks ago. If that much has changed since then, props to Microsoft.
Zed is very nice, it’s just a totally different workflow. I think people who work in a domain where AI is not particularly strong would be better off with Zed, since Cursor’s way of reviewing edits is a little clumsy.
What about on the speed front? VS Code's biggest problem is with how slow it is. I'd already be done and on to the next (and maybe the next thing after that) by the time it finally gets around to things. I like the concept, but I only have so much time in a day.
If you find VS Code to be slow, you might give Zed a try. I have been using Zed with my Claude API key and it's really something.
have you tried using either of them?
You mean the gap in vscode compatibility?
What are your thoughts on why it might be?
* a small, focused team moves faster
* cursor has great taste and that's hard to replicate at MS scale
* Microsoft had allegiance to OpenAI early on which reduced their experimentation with other models
Their main supplier (Anthropic) is also a direct competitor (Claude Code). I love Cursor but boy, what a tough place to be in. It's hard to see how it works out for them in the end.
Cursor+Gemini MAX is pretty good these days. It seems like Claude Code and C+GM leapfrog each other every month or two.
Cursor has a lot of potential leverage owning the developer and the training data streams and commoditizing the underlying model.
Code isn't as strictly competitive as the IDE's are. Code even has solid VS Code integration. It's effectively a plugin, just one that isn't tied to any one IDE.
The new memory feature on Cursor is going to keep me locked in for the foreseeable future. It's _really_ good.
Agreed. There's no way Cursor can stay ahead when they really don't have much of a moat.
Don't get me wrong, I love Cursor but is seems Microsoft could just rip it all off and put it in base VS Code.
Which they have basically done and are closing in on them fast
You and I have not tried the same vscode
Have you used copilot recently? It is absolutely useless these days.
I use Roo Code. I love it. Are Cursor or Copilot or any of the other "front ends" so much better? I guess it's up to me to find out but wonder what others have found. [edit: grammar]
Since Codex web ui came out I stopped using Cursor and just direct pull requests on Codex web interface. I love it so much, I belive most people will move to this kind of development as models are getting stronger, and the whole agent+user workflow will switch to pull request based development.
It’s not like I’m not using Cursor at all, it just became the 10-20% of my workflow compared to almost 100% before.
Can you elaborate more on this? You’re using a web ide? And you love it?
Sure, chatgpt.com/codex
Ok so OpenAI owns Windsurf now, so I would expect them to cut Cursor off from the really good stuff at some point. Anthropic has Claude Code, they could do the same. Google is a little farther behind but they do have an AI Studio thing that could be viewed as competitive.
Feels like Cursor has to make their own models to guarantee long-term survival? Especially if they’re not going for an acquisition (reports are they turned down OpenAI). Can they make a model that’s good enough for a world where OpenAI / Anthropic / Google all cut them off?
I have a feeling we're getting into an "AI squeeze" soon. So many companies in this sphere are going to have to fight for their lives and or implode.
Good to see with that war chest they'll be competing in this space for some time yet - which we'll need more of to keep the pricing down (aka subsidized).
Never tried Cursor since I'm not prepared to leave my IDEs, but still got a full AI toolbox with augmentcode.com, GitHub Copilot, Claude Code and Gemini Code Assist enabled in both my primary VS Code and Rider IDEs.
$900 million?! That's an absolutely insane amount of money. How much of this will go into building the product itself, and how much into converting cash to marketshare? And they raised $100 million not even 5 months ago. Have they already burned through that pile of money?
I mean, their product is good – I'm using and paying for Cursor – but not fantastic. And there's a lot of competition. And the switching cost is relatively low.
I'm wondering if some was used to purchase shares of employees and founders.
they are approximately a service for transferring revenue from paying users to Microsoft or Anthropic, with a small software development project alongside - why would that seem like an insane amount of money?
Too busy rolling in money to actually write a substantial announcement?
so enough for them to build their own model
I don't understand funding whatsoever. How does a company go from Series B to C in 5 months?
Impressive. Didn't they just raise a bunch recently?
Their series B was in January: https://www.cursor.com/blog/series-b
Wow. That's an insane difference in only a few months.
I thought it was a pretty interesting choice to post the funding announcement as the team and not as the CEO, more companies should probably do that.
2 million paying users?
Idk, hard to believe.
Wow.
I'm assuming a lot is coming from companies where there's fomo that if you don't use the latest and greatest tooling, your competitors will and you'll be left behind.
It's very weird that they use their blog to post about this instead of all the feature they've been adding.