[hkarthik]

DevTools companies usually have to move to a model where they're starting to pull in Enterprise accounts with licensing priced in the $40K-$100K/year range.

The $30/month subscription is just to entice in startups. If the startups grow into larger companies, the strategy is to upgrade their pricing. Inevitability, this creates a weird situation where one day a company goes from paying a few hundred dollars per year to several thousands of dollars per year for the software they're already using and rely upon.

That's about the time that someone internally will say "we can just build this ourselves." That's where it can be exceedingly hard for devtools companies to convert if their initial target market was mostly startups. They have to up their sales game and tackle big companies with Enterprise features. That can be a real struggle and many never quite figure that out.

[ycomb9]

Actually the "we can just build this ourselves" is a hard thing to do. Most modern companies do not want to spend money on things like this which they can buy off the shelf. Maintaining a thing like your CI servers need people resources and such resources cost way more than what a service provider will charge. I am sure in some situations it will make sense but for the most part it will be ridiculously cheaper. Now other factors will influence - like missing pieces, unique situations, security concerns etc.

[yen223]

A number that stuck with me is that a 2-week sprint for a team of 5 engineers roughly costs the company somewhere around $50,000 - $100,000 in labour costs.

When framed like that, paying $100,000+ per year for a service contract is likely going to be a lot cheaper than hiring the equivalent team to do it in-house.

[cosmodisk]

Well, that model didn't stop Atlassian to become billion dollar corp. It may not work in all cases but often enough it does.