Anthropic co-founder on cutting access to Windsurf
116 comments
·June 6, 2025princealiiiii
jsnell
The model providers are not in the low margin part of the business. The unit economies of paid-per-token APIs are clearly favorable, and scale amazingly well as long as you can procure enough compute.
I think it's the subscription-based models that are tricky to make work in the long term, since they suffer from adverse selection. Only the heaviest users will pay for a subscription, and those are the users that you either lose money on or make unhappy with strict usage limits. It's kind of the inverse of the gym membership model.
Honestly, I think the subscriptions are mainly used as a demand moderation method for advanced features.
raincole
> The model providers are not in the low margin part of the business.
Many people believe that model providers are running at negative margin.
(I don't know how true it is.)
jsnell
Yes, many people believe that, but it doesn't seem to be an evidence-based belief. I've written about this in some detail[0][1] before. But since just linking to one's own writing is a bit gauche and doesn't make for a good discussion, I'll summarize :)
1. There is no point in providing paid APIs at negative margins, since there's no platform power in having a larger paid API share (paid access can't be used for training data, no lock-in effects, no network effects, no customer loyalty, no pricing power on the supply side since Nvidia doesn't give preferential treatment to large customers). Even selling access at break-even makes no sense, since that is just compute you're not using for training, or not selling to other companies desperate for compute.
2. There are 3rd-party providers selling only the compute, not models, who have even less reason to sell at a loss. Their prices are comparable to 1st-party providers.
3. Deepseek published their inference cost structure for R1. According to that data their paid API traffic is very lucrative (their GPU rental costs for inference are under 20% of their standard pricing, i.e. >80% operating margins; and the rental costs would cover power, cooling, depreciation of the capital investment).
Insofar as frontier labs are unprofitable, I think it's primarily due to them giving out vast amounts of free access.
[0] https://www.snellman.net/blog/archive/2025-06-02-llms-are-ch...
apothegm
They probably have been running at negative margin, or at the very least started that way. But between hardware and software developments, their cost structures are undoubtedly improving over time —- otherwise we wouldn’t be seeing pricing drop with each new generation of models. In fact, I would bet that their margins are improving in spite of the price drops.
brookst
Citation needed.
Model providers spend a ton of money. It is unclear if they will ever have high margins. Today they are somewhere between zero and negative big numbers.
tonyhart7
subscription model is just there to serve B2C side of business which in turn them into B2B side
antrophic said themselves that enterprise is where the money at, but you cant just serve enterprise on the get go right
this is where the B2C indirect influence comes
mvdtnz
What evidence do you have that there's decent margin on the APIs?
hshdhdhj4444
Even to the extent that’s true, that doesn’t seem to be the issue here.
OpenAI is acquiring Windsurf which is its most direct competitor.
selcuka
True. Otherwise Anthropic would cut access to other code assistants too as they all compete with Claude Code.
SoftTalker
They might still. Why not?
Illustrates a risk of building a product with these AI coding tools. If your developers don't know how to build applications without using AI, then you're at the mercy of the AI companies. You might come to work one day and find that accidentally or deliberately or as the result of a merger or acquisition that the tools you use are suddenly gone.
brookst
I 100% agree with you except your framing makes it sound like the model providers are doing something wrong.
If I spend a ton of money money making the most amazing ceramic dinner plates ever and sell them to distributors for $10 each, and one distributor strikes gold in a market selling them at $100/plate, despite adding no value beyond distribution… hell yeah I’m cutting them off and selling direct.
I don’t really understand how it’s possible to see that in moral terms, let alone with the low-value partner somehow a victim.
pbh101
I don’t think it is at all clear that Windsurf adds zero value. Why do you think this is a helpful analogy?
osigurdson
The analogy is a bit like this. Imagine that there are 100 ceramic dinner plates for $6 each. Now someone comes in and buys them from you for $5 each - undercutting your margin. Then a 3rd company comes in and literally eats your lunch on your own ceramic dinner plates. The moral of the story is any story involving ceramic dinner plates is a good one, regardless of the utility of any analogy.
liuliu
Or AWS, and AWS managed services v.s. other managed services on top of AWS.
icelancer
This seems pretty reasonable to me. I don't really understand why Windsurf, owned by OpenAI (allegedly), should expect to have API access to their main competitor's API. People can still bring their own key and use whatever models they want in that way.
BoorishBears
It doesn't make sense and the co-founder is being an intentionally crappy communicator.
He's trying to paint it as removing access to something everyone gets by default, but in reality it's removing special treatment that they were previously giving Windsurf.
The default Anthropic rate limit tiers don't support something Windsurf sized, and right now getting increased rate limits outside of their predefined tiers is an uphill battle because of how badly they're compute constrained.
consumer451
> Earlier this week, Windsurf said that Anthropic cut its direct access to Claude 3.5 Sonnet and Claude 3.7 Sonnet, two of the more popular AI models for coding, forcing the startup to find third-party computing providers on relatively short notice. Windsurf said it was disappointed in Anthropic’s decision and that it might cause short-term instability for users trying to access Claude via Windsurf.
As a Windsurf user, maybe Sonnet 3.x models are slower right now, but they don't require BYOK like 4 does. So this is a bit of an exaggeration, isn't it? Anthropic did not cut them off, they seem to have continued honoring existing quota on previous agreements.
What did Windsurf think was going to happen with this particular exit? Also, how embarrassing is this for OpenAI that it's even a big deal?
BoorishBears
They could have moved to Bedrock or Vertex, which both resell Claude access and have autonomy in who they allocate higher rate limits to.
But they're also crunched on Claude 4 capacity, if not more so than Anthropic's first-party access
null
hiddencost
IDK, I pay someone money for a service, I would like the terms of our contract to protect me from getting cut off capriciously. The fact that Anthropic is selling a service without providing a contract that provides consumers with protections kinda sucks?
hiddencost
Also, like, anti competitive behavior is kinda ... Illegal?
rafram
Not all anticompetitive practices are illegal, but this isn’t even anticompetitive.
Anticompetitive practices are actions that reduce competitiveness in a market by entrenching your dominance over the (usually smaller) competition.
Not allowing your competitor to buy your product arguably increases competition? It pushes them to improve their own product to be as good as yours.
threeseed
No it's not. It's only illegal if you are found guilty of it.
And for that to happen you need to be (a) an effective monopoly, (b) have a negative direct or indirect impact on consumers, (c) large enough for regulators to care about and (d) be in a regulatory environment that priorities this enforcement.
tuyguntn
This seems odd to me, I don't expect bakery to reject selling me a bread this morning, because I started working in another bakery nearby.
charlie0
No, but if YOU opened up a bakery and were packing their bread and re-selling it, they might not like that. This happened before and it's not great.
https://www.businessinsider.com/restaurant-accused-reselling...
freehorse
This is more like about cutting you off reselling their bread, which would be reasonable.
tuyguntn
By this logic NVIDIA should cut off Claude, because Claude is reselling their GPU hours, when NVIDIA itself can do it.
IMHO, there are 3 types of products in current LLM space (excluding hardware):
* Model makers - OpenAI, Claude, Gemini
* Infra makers - Groq, together.ai and etc
* Product makers - Cursor, Windsurf and others
zmgsabst
Is it?
Grocery and department stores routinely have brands that compete with those they resell — but they’re not cut off for that. Eg, Kroger operates its own bakery and resells bread.
What makes technology unlike those?
wincy
Wow excited for when Anthropic buys Cursor in a couple months and I get locked out of using OpenAI models with it. This is depressing. I just want stuff to work.
paxys
If LLMs were a sustainable business then Anthropic would have no problem selling Claude to a competitor. Heck they'd brag about it ("see OpenAI uses our models to code instead of their own!"). You see this in the industry all the time. Large tech companies compete with each other and sue each other and have deep business relationships at the same time.
What this change really says is that Anthropic doesn't want to burn VC money to help a competitor. And that is the reality of "I just want stuff to work". It won't just work because there's no stable business underneath it. Until that business can be found things will keep changing and get shittier.
34679
Void is basically the same thing, but open source and better. It's easy to use with any provider API key, even LM Studio for local models. You can use it with free models available from OpenRouter to try it out, but the quality of output is just as dependent on the model as any other solution.
lightbulbish
Have you used it? Haven't heard about it but tbh I can see how it would eventually outperform cursor and/or windsurf. As LLMS get better, and more background tasks etc, will come, I don't see a sustainably moat around IDEs generally (except switching cost, but if it is mostly vs code... )
saw you did below. What is your experience so far? fast requests are great. Anything big lacking?
I was using roo code for a bit and it was cool to see how fast it was going compared to windsurf.
34679
>What is your experience so far?
I cancelled my Cursor subscription and haven't used it since. I experimented with Aider for a bit, it's also pretty great. Their architect mode seems to be the way of the future. It allows you to use a pricier model for its reasoning abilities, and it directs a cheaper model to make the code changes.
That said, I keep going back to Void because Aider's system prompts have a tendency to introduce problems. If Void had Aider's architect mode, it would be perfect.
zwarag
People like Windsurf and Cursor because they offer them a flatrate.
34679
Yeah, for sure. That's why I tried Cursor for a month. But as soon as I ran out of fast requests it became unusable. It had me waiting several minutes for the first token. I didn't realize how bad the experience was, fast requests included, until I used Void. It makes Cursor fast requests seem slow, and I tend to need fewer of them. The differences being that Void requests go straight to the provider instead of their own server first, and their system prompts seem to keep the models more on task.
demosthanos
Does anyone here actually use OpenAI models in Cursor? I kind of forgot they were even there. I've just been alternating between Claude and Gemini, and the sense I've had in online discussions is that that's pretty normal.
brookst
Friendly advice: you will be happier if you defer reacting to things until they are actually real.
We can all imagine all sorts or terrible futures. Many of us do. But there is no upside in being prematurely outraged.
discohead
How Cursor is building the future of AI coding with Claude: https://www.youtube.com/watch?v=BGgsoIgbT_Y
OsrsNeedsf2P
This is why I use Aider. Open source or GTFO
TiredOfLife
And then Anthropic cuts api access to Aider
selcuka
Anthropic models are slightly better for coding tasks anyway. I believe Windsurf is in more trouble.
dboreham
Probably most here are not old enough to remember, but there was a time when Google had all sorts of data access APIs and encouraged developers to create applications that used said APIs. And then they disabled them all.
nickthegreek
Now your company needs to worry about building workflows on the back of these neat AI companies, they got bought out and cause a disruption that can reverberate through the org in unknown ways for unknown amounts of time. This and the openai no deletion court order should be a wake up call on how these technologies get implemented and the trust we cede to them.
sebmellen
It’s no different from the shell games that have been going on in enterprise software for the past 30 years.
You can choose independence if you’re willing to use a slightly worse open weight model.
widdakay
These are two different products. It's like SpaceX launching satellites for competitive satellite internet services. They didn't care that they were providing launch capabilities for a competitor and neither should Anthropic. What if Apple stopped allowing you to use an iPhone if you worked at Google?
tonyhart7
"They didn't care that they were providing launch capabilities for a competitor"
Yeah because another internet provider did not have SpaceX reusable rocket technology
its not really quite the same you know
demosthanos
> What if Apple stopped allowing you to use an iPhone if you worked at Google?
This wouldn't be remotely comparable. This is targeting of a competitor's employees, not targeting a competitor's subsidiaries.
If you want to go the Apple-Google route a better comparison would be that this is like Apple refusing to allow you to hook up an Air Tag on an Android phone. Which is something that they do, in fact, do.
null
danpalmer
Another way this isn't comparable is that the training data is critical for the services each provides. Seeing answers Claude gives (at scale, for free) would be a huge competitive advantage to OpenAI, whereas Apple Mail seeing the email you send via Gmail wouldn't convey any competitive advantage in email, for example.
thrdbndndn
I think the nature of your two examples, along with the one from the news, is too different from each other for the analogy to really hold. These situations can only be judged on a case-by-case basis.
paxys
If SpaceX was launching every rocket at a loss then they would absolutely care about a competitor taking advantage of it.
killerstorm
Do you understand that data can be used for training?
k__
When I read about Cursor and Windsurf, I'm quite happy that I only use Aider. An open source project, not associated with anyone else...
bravesoul2
Curser and Windserf
Yeah better to go with open technologies. Maybe use Groq for inference knowing you can switch over later if needed as you are using Llama or Deepseek.
pton_xd
I'm sure this all fits neatly into their EA "maximizing positive outcomes for humanity" world-view somehow.
CyberMacGyver
It could be argued from Antrhopics perspective that OpenAI and Worldcoin are not positive for humanity so this in fact necessary
dmix
I don't think EA people are pro going out of business so you don't have money.
bravesoul2
Can't save humanity from the heat death of the universe without a castle.
null
arnaudsm
We are about to enter the era of aggressive LLM monetisation and anti-competitivness. We got used to cheap subsidized models, in bet in 2 years we'll pay double for the same service.
jmward01
Decisions like this may shoot them in the foot later as opensource and cheaper compute continues to push into frontier model territory. I know I have no loyalty to the big providers and would take a minor quality/cost hit to jump off. Right now it isn't a minor quality / cost hit though. Knowing that they can cut you off if they think you are going to end up competing with them makes me tolerate an even bigger gap.
null
Any app built on top of these model providers could become a competitor to these providers. Since the model providers are currently in the lowest-margin part of the business, it is likely they will try to expand in to the app layer and start pulling the rug from under these businesses built on top of them.
Amazon had a similar tactic, where it would use other sellers on its marketplace to validate market demand for products, and then produce its own cheap copies of the successes.