Microsoft has clearly taken notice. They're already starting to lock down the upstream VSCode codebase, as seen with recent changes to the C/C++ extension [0]. It's not hard to imagine that future features like TypeScript 7.0 might be limited or even withheld from forks entirely. At the same time, Microsoft will likely replicate Windsurf and Cursor's features within a year. And deliver them with far greater stability and polish.
Both Windsurf and Cursor are riddled with bugs that don't exist upstream, _especially_ in their AI assistant features beyond the VSCode core. Context management which is supposed to be the core featured added is itself incredibly poorly implemented [1].
Ultimately, the future isn't about a smarter editor, it's about a smarter teammate. Tools like GitHub Copilot or future agents will handle entire engineering tickets: generating PRs with tests, taking feedback, and iterating like a real collaborator.
[0] https://www.theregister.com/2025/04/24/microsoft_vs_code_sub...
[1] https://www.reddit.com/r/cursor/comments/1kbt790/rules_in_49...
We've seen this before with Office.
We'll see it again.
Microsoft Build is this month [0] and it will tell where they are going next (other than price cuts).
I'm expecting disappointment for now, but also expecting GitHub Copilot to be upgraded. Then we'll see if they are ahead or so far behind.
If autonomous agents were just around the corner, then why wouldn't OpenAI bet on their own Codex product obviating (most) need for an IDE and save themselves the $3 billion?
Have you used Cursor on a daily basis? I have. Every day for six months now. I haven't encountered a single bug that prevent me to work.
Moreover, while Microsoft tries to catch up lately, it's still very far behind, especially on the "tab autocompletion" front.
I agree with the first part, I'm much less optimistic about the second part. I suspect they will create something that is worse, but cheaper if you already pay for Github/Office 365/whatever. Then many large enterprises will switch to save money whilst the engineers complain, just like with Teams.
I always felt that cursor and windsurf should be just extension to vscode instead of a fork. Was there some missing functionality is vscode that was missing? Is it still missing?
There are some extensions that work in this way and allow to use multiple implementations depending on task at hand without any long term commitment.
I feel like such fragmentation is by artificial just to lock users in single ecosystem.
Microsoft provides the editor base, foundation models provide the smarts, and Cursor provides some, in my experience, extremely buggy context management features. There is no moat.
[0] https://knowyourmeme.com/memes/unfinished-horse-drawing-flam...
Buying competition while everyone’s still fighting might straddle you with a lame horse
It's in a lot of ways the OpenAI story itself: Can they keep an edge? Or is there at least something that will keep people from just switching product?
Who knows. People have opinions, of course. OpenAIs opinion (which should reasonably count for something, them being the current AI-as-a-product leader) is worth $3B as of today.
Maybe it's fine if you only do local development in other languages (Javascript?), but I completely swore it off.
I actually find it a little reassuring that they can't seem to get out of their own way.
Meanwhile GitHub web integration is approaching seamless in vs code. To the point I often forget I’m in a browser instead of the app, until an extension I use doesn’t work.
I would also argue that the product could be built over two weekends with a small team. They offer some groundbreaking solutions, but since we know that they work and how, it's easy to replicate them. That also means they have significant talent there.
Hence, they are also buying the employees.
The code base itself is basically worth nothing, in my opinion.
https://code.visualstudio.com/docs/copilot/copilot-customiza...
Tale as old as time, song as old as rhyme: embrace, extend, extinguish.
Of course, back to reality. Today, at least in my workflow, I use / like Windsurf but it is a small part of what I am doing. For any code I want to keep I mostly write it by hand (using vim for a very bare-bones / cognitive mode experience). For me, the real flow state occurs in vim while ChatGPT and Windsurf are great for exploration.
The fact that they are is not the results of the Microsoft takeover.
I gave up on Cursor because my trial ran out, while VS Code with Copilot doesn't seem to charge me anything.
However, given that JetBrains also have their own AI offering[2], I'm not sure how long that will last too...
[0] https://plugins.jetbrains.com/plugin/20540-windsurf-plugin-f...
[1] https://marketplace.visualstudio.com/items?itemName=Codeium....
They should have restricted the Marketplace several years ago, however, they are doing it now.
With C++, they are part of MFC's, they are the legal owners, not like Google vs Oracle in case of Java.
Lastly, with AI Code IDEs I think yes, there is a case, the need for IDE might be very less. Like a steering on a self driving car.
This reminds me of "big companies moves slow.." line.
Microsoft software quality has gone downhill recently, and I'm not going to bet on them delivering something more polished than WS and Cursor here.
Side: all images on Microsoft websites are low resolution! it's like they don't even check their own website.
1) Cline (1.4mil downloads)
2) Roo Code (a fork of Cline, 450k downloads)
Still a drop in the bucket compared to Cursor in terms of # of users, but they're growing pretty fast.
Disclaimer: I maintain Kilo Code, which competes with 1) and 2) so I'm pretty familiar with this space/the growth patterns.
What groundbreaking solutions does Windsurf offer?
And I have long held that they are hungry, shipping like clockwork on or about the 20th of every month, showing up with actual improvements all the time https://about.gitlab.com/releases/ It seems this month brings 18.0 with it, for whatever that version bump happens to include
They also have a pretty good track record of "liberating" some premium features into the MIT side of things; I think it's luck of the draw, but it's not zero and it doesn't seem to be tied to any underhanded reason that I can spot
Or you chat and suddenly it wants to use the azure copilot instead because reasons.
Horrible experience.
Their “programmers” are more busy with making blogs and videos than functioning tests or technical documentation, and they start using JavaScript and Python for everything.
I’m not surprised their quality went to shit. There are some pearls left, C# in general is pretty good, and Aspire is becoming quite neat.
The latter I think mainly because David Fowler is just a great developer
The MS acquisition traded the developer community to briefly appeal to enterprises, then quickly let both down.
I think the GitHub brand is still stronger and people just don't "care" about gitlab.
We also have our own provider, which means no need to bring your own API keys (you can if you like, but it is batteries included by default) and we're not charging anything on top of the API pricing. Instead of monetizing on individual developers, we want it to be free for them and make money eventually off enterprise contracts [1]
[0]: https://blog.kilocode.ai/p/roo-or-cline-were-building-a-supe... [1]: https://kilocode.ai
What doesn't the current API allow plugins to do? I'm guessing custom UI stuff that lives outside a panel?
In fact, their own AI extension appears to be pluggable in and of itself. I think they see the value in being easy to adapt different AI solutions to rather than trying to only provide their own.
Probably.
> And deliver them with far greater stability and polish
That seems ... overly optimistic given MS's history.
Care to place a bet?
No, they should be LSPs so that they can be integrated into any editor, not just VSCode.
Because if you're referencing to a headline (without reading the article) that was on H a couple of days ago, it stated that 20-30% of the code in the repos was written by software. Software != AI
To quote wongarsu in the same post: "Considering that most of their software has been developed for decades and AI assistants have only started becoming useful in the last ~4 years it would be very surprising if 30% of their code is AI written. I doubt they even touched 30% of their code in the last 4 years. But what is perfectly plausible is that 30% of their code is written by code generators. Microsoft has a lot of interface code. All the windows DLLs that are just thin syscall interfaces, the COM and OLE interfaces in their office suite and everywhere else, whatever Office uses nowadays for interoperability to allow you to embed content of one product in another, whatever APIs their online products use, etc. In the leaked Windows XP source code it can be difficult to find the actual source code in between the boilerplate files containing repeated definitions, and in the decades since then the world has only leaned more into code generation."
Source: >>43841868
That’s why you won’t see a ton of work go into e.g. issues/projects on GitHub. Those features all already exist and are very robust in ADO, so if you need those kinds of things (and the reporting an enterprise would want to be able to run on that data), then you belong on ADO.
this is the question i am still asking...
I agree with the rest, they've all mostly lost market share or completely no longer exist due to VS Code, but not IntelliJ, that platform is going really strong.
Though no doubt, VS Code has pushed JetBrains to rethink some things, and be better in general.
But I agree with you about the first part, and I think it's awesome for me as a user that all this competition to build a matter mousetrap is happening right now! I'm not as certain as you are that Microsoft will end up building a better version. It's definitely one of the likely outcomes. But it's also totally plausible that Cursor or Windsurf can win the race, even if they need to replace every single one of the MS extensions and entirely diverge the core IDE from upstream. These products are well capitalized and it's just not that hard to build the core pieces of an IDE.
I don't like that GitLab lets you nest organizations and such, it makes it so painful to find things over time. I appreciate GitHub doesn't do this, I view it as a plus
I also disagree about GitLab CI, not that it wasn't smart for them to include alot sooner than GitHub, but Actions is really good and really easy to get up and moving with. I find they run faster, have better features - like they can annotate a PR with lint errors and test failures - with very little comparative configuration.
GitLab CI yaml is a mess by comparison. GitHub was smart to push things to the runner level once a certain complexity threshold is hit.
This has been my experience of course, and so much of it is really subjective admittedly, but I don't think GitLab is truly ahead at this point.
Only for azure devops, there are +6k problems listed on developer community website with 500 still not closed for the last 6 months. [1]
The complete integration in the ecosystem is what's flawless.
Any company with a better product has to fight that integration and they almost always lose (Sybase, Borland, WordPerfect, Lotus, Netscape...)
1 : https://developercommunity.visualstudio.com/AzureDevOps?ftyp...
Not sure how they'd do it considering you bring your own API keys. Can you link me to a resource?
I've tried just about every model on its own over the years, and yet there's something about the Cursor workflow that frequently still gives me chills when it shows me again that it had clearly anticipated what I would think next in a way I just don't experience with other tools.
Holistic seems like the right word?
If it's all smoke and mirrors as some folks imply, then it's Penn and Teller level smoke and mirrors. Beware those who tell you that they could duplicate anything of value in a weekend.
I think a few options for this already exist, but honestly they don't go far enough. I want something like an AI scrum master, for hyper agile teams, that can task out smaller tickets to AI sub agents.
I would integrate this thing in with something like an AI powered Jira.
Two arguments exists.
1. I need to take about 6 months off and start building this now, even if I don't know exactly how I'll get it done. Between a combination of vibe coding and maybe a bit of outsourced work ( looking at Eastern Europe), I could get this done with my personal funds.
2. To do this properly would probably require tens of millions of dollars. I'll probably burn myself out trying to do it solo without ultimately getting to a sellable product.
The biggest issue here is to actually scale I would need to either have users bring their own LLM keys or have tens of thousands to spend on LLM tokens.
But I'm glad OpenAI is getting into the tooling space in this way. I cant wait to use all the cool features they build after VSCode rips them off.
They don’t have access to copilot users in general, Microsoft and Google does. And perhaps they are realizing that Microsoft is hedging them over multiple LLM providers and maybe no longer feeding them juicy copilot data, with humans in a tight loop, correcting LLMs.
I am guessing you are talking about GitHub Copilot when you say VSCode. GitHub Copilot is far far inferior product when compared to Cursor, Windsurf or Augment Code. Most people who try almost any Copilot alternative for a reasonable amount of time end up canceling their Copilot subscription. I did, after two months of using both.
Mentally, I'm replacing claims like this with "it will do magic!" and I think I'm just about as likely to be correct.
And autocomplete is the least interesting thing an LLM can do. Cursor’s UX isn’t the end game but has lots great features.
The ideal UX is still being worked out. It’s good that different people are building tools to try different ideas.
I am all for everyone adopting LSP, but the reality is harsh.
To be honest I think both are quite limited by context length (in that they try to limit the context they send to the LLM and hence cost), and so I find myself using Gemini 2.5 in AI studio with the 1m context length, and asking it to generate instructions for Copilot (which seems to work pretty well)
Cursor and other forks have decided to circumvent this, some even going so far as to use proxies to bypass restrictions.
I'm not convinced Microsoft owes other billion dollar companies free access to a product they've built, curated, and supported for over a decade. Plug-in authors are not restricted from publishing their products on competing marketplaces.
it was crap compared to Borland's products 20 years ago
and today it's crap compared to JetBrains'
and christ knows how anyone could consider the Azure UI to be "great"
other than Teams I don't think I've used a worse piece of software
I disagree, but would love to be wrong. These tools exploded onto the scene and were massive productivity helpers, but since their initial integrations they’ve churned rather than improved in the last 2 years. They are even worse when you try to iterate rather than just get them to one shot the problem space.
The last cycle I remember of this IMO is iPython -> Jupyterhub/Jupyterlab. Of course, iPython has existed for a long time, though that change was made because data was too big to analyze locally and it turns out it was more convenient to centrally manage kernels/images/libraries for convenience.
MCP servers and Cursor/Windsurf changed that a bit, but it will end up centralized again at some point (or at least aggregated, if it's not already?). People are passing around lists of interesting MCP servers now, and that will be out of fashion in less than 12 months.
I’ll agree on Teams being crap though, mostly for how dumb it is that they’ve rewritten it multiple times and created a confusing slate of weird versions like “Teams (work or school)”
They can have all the money in the world and it doesn't mean much in this context.
For while Microsoft is going to invest heavily in a Cursor / Windsurf like product and likely do alot to ship it in their editors - likely with exclusions or lag times between updates on other platforms - there's zero reason for Google to do this for example, when they could sell through Gemini for Code as an extension across all editors.
I don't see JetBrains having issues because of AI tooling, for most of these companies, its a boon to be on the JetBrains platform. Especially because JetBrains has lots of enterprise customers who would naturally be very interested in buying AI tooling for their developers. Its a natural market
They are very different companies in structure and it certainly is a "pick your poison" but it's completely stupid to act like they're the same on this front. Apple is better on user privacy
...unless you care about state actors, which you should, in which case your data is the US government's either way.
But, if you tell aider to watch your files, you can drop a specially formatted comment into your file, and aider will see that and use it as a prompt.
So the integration is sort of “implicit”. Which sounds kinda like the cheap way to go, in comparison to the current brand name tools that have their own chat boxes, dropdowns with mode selectors (ask, edit, agent), and so on.
But look further into the future and an ambient interface is probably where we end up. Something where the Ai agent is just watching what you do, maybe even watching your eyes and seeing what you’re attending to, and then harmonizing its attention to what you are attending to.
But I dunno, i’m just guessing
Stable and polished are not words that ever came to my mind while using any Microsoft product.
But, anytime I am empowered to pick, I'm going to pick GitLab 100% of the time because it has every feature that I care about and "being popular" isn't a feature that I care about
For example, Github only autocompletes based on what file you have opened in the current editor's tab. Windsurf indexes your entire code base and seems able to autocomplete based on what other files you have in your project. Autocomplete also spans across multiple lines and open tabs.
Windsurf's agentic tool (Cascade) can run terminal commands and read the output without opening a terminal like copilot. It can undo the agent's actions easier than Copilot. Though I think Cursor is superior in that regard, it can undo multiple checkpoints.
Still evaluating Windsurf but it, Cursor, and Claude Code are all more sophisticated than Github copilot at the moment. I'm sure copilot will catchup but by that time the other tools may have already iterated ahead.
This doesn't mean that aider, claude code, etc. aren't very good tools, but it does make sense to distinguish between built-in tools vs external ones. A similar non-AI example is debugging or linting: IDE integration makes it much easier than using a separate tool.
Nah, I hate that. At my job we have a few different orgs, with terrible SSO boundaries (having to auth multiple times to GitHub because I work on repositories from different GitHub orgs). Allowing you to have a proper structure with nestedness, while still having good search, is great. You can also easily move projects and namespaces around, so if the structure doesn't work, it can evolve.
Why would you have the 50 library repositories you've had to fork as top level projects polluting your org? You also can't really do shared variable, environment, CI configs between repos of the same project/type.
vs-code is one of the few products coming of of microsoft that leads the pack by a big margin, and it is no surprise that all of these startups are forking it.
One exception in 50 years does not inspire confidence.
I didn’t mean to imply that MS wanted to migrate anyone, just that the different offerings serve different kinds of customers, so you can’t really just compare GitLab to GitHub and say MS is lacking in serving some group of them.
The official guidance from Microsoft since probably 2019 has been to encourage all greenfield projects to GitHub, as opposed to ADO.