Spacelift CEO Pawel Hytry dives into the infrastructure automation challenges DevOps teams are facing following the picking up of an additional $51 million in financing.

Infrastructure-as‑code was supposed to make provisioning easy—copy some Terraform, hit “apply,” and move on. In practice, teams soon discover that version‑controlled snippets are only half the story. Once dozens of developers start spinning up resources, you still need collaboration, security reviews, cost oversight and a reliable audit trail. Hytry explains why his biggest customers hit a wall: what used to take a day to deploy now drags on for a week because manual approvals, rollbacks and duplicated code clog the pipeline. The tipping point usually arrives when an infrastructure lead realizes the team is hiring people just to babysit scripts.

Hytry argues that the answer isn’t to lock everything down; it’s to add “guard‑railed self‑service.” A central platform team defines reusable modules and policies, then lets application squads launch infrastructure inside those boundaries. That keeps velocity high while preventing surprise cloud bills or rogue ports. He frames the approach as a natural fit for the rise of platform engineering, where one group maintains the paved road so everyone else can drive fast without veering off a cliff.

AI enters as a helper, not a takeover artist. Hytry envisions agents that trace resource drift or predict cost spikes, with a human still approving changes—at least for now. The broader lesson, he says, is to replace brittle, ticket‑based processes with automated pipelines that scale as headcount and complexity grow. Teams that ignore the sprawl will keep paying in slow deploys, mounting cloud spend and tired engineers.