The DevTools Ceiling: great as Open Source, AND a Terrible Business

I had a very familiar pang of empathy when listening to the Stellate post mortem on the excellent OSS Startups podcast. Not because I once nearly joined Max Stoiber, who I greatly admire (with Tim Suchanek, now of Expand.ai). Stellate was a GraphQL CDN — working out caching nuances for production GraphQL users like Max was building Spectrum.chat. I was right for the wrong reasons: I was ultimately unsure about betting (more of) my career on GraphQL, but Max says Stellate petered out because the TAM was very small — the sweet spot for a GraphQL CDN is spiky, yet dynamic usecases, yet most spiky usecases like ecommerce and media sites are pretty darn static when you get down to it. Cool but also GraphQL lost appeal, let’s be honest. How’s Apollo doing?

No, I was sad because around about the halfway mark he starts going into the title topic: “Can DevTools Get to $1B ARR?” and (the host) Tim Chen agrees and says basically only Hashicorp, GitHub, and now Vercel “made it”, and everyone else has had disappointing, middling outcomes. (I might add GitLab, Sentry, JetBrains, Atlassian, presumably Linear, but yeah it’s a short list). Entire markets were savaged in that conversation - all API tooling — Kong ($2b), Postman (valued $6b, now $3b), kaput, B grade, thanks for playing.

It reminded me of an observation I once made about the Frontend Ceiling for individual developer careers - Frontend is important and valuable and difficult and noble and challenging and becoming more and more complex — all that is true, and yet, if you only do frontend dev, you will probably never be a CTO because that is not the kind of background that helps you work on valued systems for most businesses:

In other words, if you start off as a frontend person, you’re great, world needs you, thank you for your service, but if you want to be VP Eng/CTO, you’ll have to “grow up” by adding some backend/data to your mix. If you’re new to this line of thinking, read that thread for the debate, or watch the Cascadia JS talk I did that ended my frontend career, I don’t have space to re-litigate here.

In the pod, Max observes somewhat of a Ceiling for devtools businesses too. He observes that most devtools businesses eventually gravitate toward being infra businesses, even to the point of buying their own hardware and building datacenters. In other words, even if you have a great framework or open source developer tool, someday you’ll have to “grow up” by becoming an infra business to make yourself sticky or to serve more of the IT budget of any organization. This already hits some of the “winner” names I listed above - Sentry is (kinda) Open Source, but does well because they famously capture ~66% of users to Sentry’s infrastructure. Next.js is Open Source, but are very much funded by the excellent infrastructure of Vercel. Let’s just say Temporal and Airbyte are the same as well — great open source, but not in themselves valuable businesses until you built up Temporal Cloud and Airbyte Cloud to be competitive infrastructure solutions compared to self hosting. Hashicorp and other COSS companies have rather different/more traditional models - you can replace infra with good lawyers.

Devtools on their own? Great as open source — your users can contribute fixes, github stars go to the moon, it’s a nice virtuous cycle.

But if you want to get paid? Become an infra business ASAP, or run into the devtools ceiling.