When IaC starts to outgrow a DSL
Hi friend,
Most people I talk to are skeptical about switching tools. Makes sense. Infrastructure is critical, and if you've got a working setup, why change?
The pattern I see is: things work fine until they don't. You've got 50 modules that are 95% identical. You're building wrappers and codegen around your IaC. Every "quick change" is a ticket, and you're the bottleneck.
At some point you realize you're not just managing infrastructure anymore. You're building a platform. The question is whether it's intentional or accidental.
Teams who made that shift intentionally:
- Starburst went from 2-week deploys to 3 hours
- Lemonade 's infrastructure team stopped being the bottleneck. Developers self-provision now.
- Supabase scaled from 2 infrastructure contributors to 40+
If you're already building a platform, you might as well build it on something designed for that. Pulumi is what teams like Supabase, Lemonade, and Starburst use. Components with guardrails. Self-service in code.
Cheers,
Adam
If you don't want to hear from me again, click here.
Pulumi Corporation, Two Union Square, 601 Union St, Pulumi Suite 1415 Seattle, WA.