Comment by hedayet
7 months ago
I was so surprised (or shocked) to hear that Windsurf was getting acquired for 3 billion dollars, I made an HN post asking about the truth of that news - https://news.ycombinator.com/item?id=43933825. HN's system didn't like my tone I guess and removed it, lol.
But in any case, I just can't see how AI code editors like Windsurf or Cursor, without any proprietary model, can be valued at billions. What's the underlying IP that justifies these valuations?
Similar thing to what we’ve witnessed with crypto coins. It’s AI season and those with money invest in it, pump it and will exit post IPO. Difference here is, that besides value that those products “hold”, it’s possible also to provide AI as a service, making Google / Microsoft etc interested.
> I just can't see how AI code editors like Windsurf or Cursor, without any proprietary model, can be valued at billions.
Windsurf used to be GPU rental company before they decided to make a text editor. I imagine they own a lot of silicon they are running their coding assistant on.
https://www.afp.com/en/infos/windsurf-launches-first-gpu-clu...
It's not a rinky-dinky startup that just forked VS Code in a month.
you meant Exafunction? Nothing of Exafunction was discussed in OpenAI or Google deals.
Windsurf did start as a rinky-dinky starup, just as a plugin (Codeium). And in its current state, it still has no moat.
The only reason big tech leaders are throwing money at this is - they need to show they are doing something to stay ahead. And right now they are going on a hiring frenzy as a show for their investors.
I have been there when shopping businesses were being bought for billions of dollars, and then they all got bundled up into a mess swapped under the rug silently. I don't think this is going to be any different.
Codium was built on Exafunction infrastructure.
https://www.unite.ai/varun-mohan-co-founder-ceo-of-codeium-i...
It's really a simple story, they built infrastructure years ago, initially they rented it, so companies outsourced their computing tasks to them. But eventually came up with the idea what product could they put on top of their infrastructure to sell it for dollars rather than pennies. And they did.
They don't need a moat because they provide turnkey utility to a lot of very rich corporations. They are not in the business of selling the vs plug-in to you. They are in the business of selling developer experience powered by their own vast compute to a lot of very rich corporations.
They had millions in revenue before they hired their first sales rep. I don't think just a VS plug-in startup could do that.
https://www.saastr.com/cro-confidential-how-codium-built-a-b...
They sell stuff that actually works, and people who use it convince people who pay money to pay for it.
They both have proprietary models.
Anyone can build a proprietary model in under a week; but the point is - Cursor and Windsurf don't have proprietary models that remove dependency on external models, and/or have capabilities that competitors don't have.
Maybe Google sees something under the hood