openai's $3b acquisition of windsurf falls apart. after months of negotiations, they walk away.
That isn't accurate. Microsoft was an OpenAI investor and had rights and for MS reasons, exercised them. That's what killed the deal.
google announces they're paying $2.4b to hire windsurf's ceo and 41 researchers for deepmind. not to acquire windsurf. just the humans. the same day openai walks. what a coincidence!
That isn't accurate as well. Google also licensed the Windsurf IP.
My question is what happened to the $2.4B? Apparently very little of it made its way to the Windsurf employees, as #2 tweeted last week. It wasn't an acquisition although Cognition was. Cognition bought a company for $250M that just got a check for $2.4B. How exactly did this work?
The premise of this article seems to be that the researchers at Windsurf have been able to learn amazing things thanks to near unlimited VC money, increasing their value to astounding numbers.
That may be true (although I doubt the numbers add up).
But what is keeping those researchers from walking away, or underperform whenever they feel like it? Giving them a lot of money surely isn't going to motivate them to work harder.
It's probably giving them access to one of the world's largest clusters of compute that lures them in.
this reminds me when early on in Github Copilot journey when people were asking "what if it accidentally reproduces GPL code and I get sued" and Microsoft said they'll cover the legal costs for anyone using Github Copilot.
The big players (Google, Microsoft, Meta, Nvidia) don't want ai startups failing. In fact they are terrified of that. Can you imagine the market shock if windsurf just went under in 1 year. How fearful all investors would be? How the whole market is gonna react? We are told that AI is basically a money printer. If you release a product for $20/month and a couple of months later other companies (with much better margins by definition) release competitor for $250 then $300 then $400 something is clearly not adding up even among the higher numbers.
Putting aside the model and researchers windsurf have, is the windsurf business model sound? If not, does potential acquirer think the business model can be adjusted within reasonable time to create value?
If the answer is no for the whole ecosystem, it’s going to be much harder to see acquisition of companies like this.
Windsurf was just a recent rebrand of Codeium which was founded in 2021 and had plenty of enterprise customers using their self-hosted models well before they decided to make an IDE. I think this article is surprisingly thin on background.
This is a highly speculative post, with conjectures presented as facts. Some things that irked me:
- Cursor did not hire Anthropic's "researchers". It hired the guys who built Claude code (PM and dev). Who then promptly went back to Anthropic in 14 days. A researcher for Cursor need not come from Anthropic either. One high profile recruit for them was Jack Gallagher (Midjourney) who is probably one of the best at RL.
- Google's deal with Windsurf is structured that way because they likely could not directly acquire, or were not confident that it would have gotten past the antitrust. A signal for that is such deal increased in last few years after FTC refused to allow any deal over $100M or so. Microsoft has done such deals too. Meta would have acquired scale ai in older times. Not sure with Openai, but they arent as scrutinized as Google for such deals. To imply that this means Google did not care about ARR is not justified. and then google licensed Windsurf's IP too.
- Openai's agreements with Microsoft is more probable than they did not complete the acquisition because of negative gross margins.
- Plus, the old adage about how a growing startup is worth more because of a stellar team. You strip a team away and still get 2x multiple is sure enough valuing the current ARR highly.
I thought the userbase is valuable. A sale at this point made sense because they might not have been to get the money if they waited a year. Reasons laid out in the article are not why I think so.
When I used to play Transport Tycoon Deluxe the most annoying were new companies which emerged every 6 months after I bought the last one. I had to buy them quickly because otherwise they would build their tracks and grow and later it would be harder to buy them...
Hmmm... I'm just saying things that are loosely related
I like how there’s like 17 instances of using square brackets as parentheses, but also actual parentheses, & almost nothing is capitalized, even acronyms, except for the cases where they feel like acronyms actually should be capitalized. and the .log file is a screenshot of cat-ing a .txt file which has formatting in it
I honestly think all of these VSCode forks are moving in the wrong direction. They are horseless carriages, old-style tools with AI stapled to the side. I'm trying to figure out what agentic coding should look like, and I built Crystal as an agent management platform first, IDE second.
If it is true that these wrapper companies can't find positive sum economies for their product offerings, there will be a point in time where we will see them fail and it will cause a domino effect, rolling over to the big LLM providers and ultimately NVDA. This is highly speculative, but currently there are some realistic pointers to such a scenario.
In such a case, I think we will see a move to smaller, domain specific models. I think if the first cycle of the AI age is large, general models, the second will be smaller, domain specific ones that we can run on retail machines.
14 comments
[ 2.1 ms ] story [ 35.2 ms ] threadMy question is what happened to the $2.4B? Apparently very little of it made its way to the Windsurf employees, as #2 tweeted last week. It wasn't an acquisition although Cognition was. Cognition bought a company for $250M that just got a check for $2.4B. How exactly did this work?
That may be true (although I doubt the numbers add up). But what is keeping those researchers from walking away, or underperform whenever they feel like it? Giving them a lot of money surely isn't going to motivate them to work harder.
It's probably giving them access to one of the world's largest clusters of compute that lures them in.
The big players (Google, Microsoft, Meta, Nvidia) don't want ai startups failing. In fact they are terrified of that. Can you imagine the market shock if windsurf just went under in 1 year. How fearful all investors would be? How the whole market is gonna react? We are told that AI is basically a money printer. If you release a product for $20/month and a couple of months later other companies (with much better margins by definition) release competitor for $250 then $300 then $400 something is clearly not adding up even among the higher numbers.
If the answer is no for the whole ecosystem, it’s going to be much harder to see acquisition of companies like this.
- Cursor did not hire Anthropic's "researchers". It hired the guys who built Claude code (PM and dev). Who then promptly went back to Anthropic in 14 days. A researcher for Cursor need not come from Anthropic either. One high profile recruit for them was Jack Gallagher (Midjourney) who is probably one of the best at RL.
- Google's deal with Windsurf is structured that way because they likely could not directly acquire, or were not confident that it would have gotten past the antitrust. A signal for that is such deal increased in last few years after FTC refused to allow any deal over $100M or so. Microsoft has done such deals too. Meta would have acquired scale ai in older times. Not sure with Openai, but they arent as scrutinized as Google for such deals. To imply that this means Google did not care about ARR is not justified. and then google licensed Windsurf's IP too.
- Openai's agreements with Microsoft is more probable than they did not complete the acquisition because of negative gross margins.
- Plus, the old adage about how a growing startup is worth more because of a stellar team. You strip a team away and still get 2x multiple is sure enough valuing the current ARR highly.
I thought the userbase is valuable. A sale at this point made sense because they might not have been to get the money if they waited a year. Reasons laid out in the article are not why I think so.
https://youtu.be/r-GFmH0EK9Y?si=8kWjzP1YdWVp6oAp
Hmmm... I'm just saying things that are loosely related
It makes sense that the answer is no.
https://github.com/stravu/crystal/
In such a case, I think we will see a move to smaller, domain specific models. I think if the first cycle of the AI age is large, general models, the second will be smaller, domain specific ones that we can run on retail machines.