Cutting Power Digital's Infrastructure Automation Costs 10x …

Data visualization of client and resource growth climbing sharply

The Challenge

Every client in one state file: a 43,000-resource Terralith

Power Digital’s infrastructure had grown into a Terralith: one monolithic Terraform configuration, over 500 root-module files, holding every client deployment in a single state file. At 43,000+ resources, it was pushing Terraform Cloud to continually break:

Over 50% of runs failed from timeouts and out-of-memory errors, forcing engineers back to manual provisioning.
Plan and apply cycles took up to 25 minutes, costing an estimated 63 engineering hours every month in waiting and debugging.
Every change carried a high blast radius. Nothing could be isolated, so a small change could reach every client.
Scaling further meant an exponentially more expensive Terraform Cloud plan, with no guarantee of better performance.

Onboarding new clients is Power Digital’s revenue engine, and the platform meant to power it had become the bottleneck.

63 hours of engineering time lost each month waiting on plan and apply loops

The Migration Playbook

Rebuilding the platform without pausing the business

Client onboarding couldn’t stop while its platform was rebuilt, so nothing was migrated until a replacement was proven in production. The playbook:

Audit & roadmap

The engagement started with Masterpoint’s Infrastructure as Code audit: stakeholder interviews, a full review of the Terralith and Terraform Cloud deployment, and a migration strategy leadership could sign off on.

Build the new system in parallel

A modular platform on Spacelift and OpenTofu took shape alongside the Terralith, which kept serving every existing client untouched.

Onboard new clients to it first

New clients landed on the new system from day one, proving it under real production load before anything was migrated.

Decompose & migrate

With the new system proven, the Terralith was broken apart and existing clients moved into their own isolated stacks, client by client.

01 · Platform Migration

Terraform Cloud → Spacelift

After evaluating the options, Masterpoint recommended Spacelift, a leading TACOS (Terraform Automation and Collaboration Software) platform, and migrated all 43,000+ resources onto it without issue.

A pricing model that scales with actual usage, instead of the exponential jump Terraform Cloud required to keep growing.
Granular control over stacks and deployments, so large configurations decompose into manageable, connected pieces.
Guardrails as code. Operational and security policies are defined and enforced in the platform itself through Open Policy Agent, with no third-party bolt-ons.

Spacelift, the infrastructure orchestration platform Power Digital migrated to — Spacelift orchestrates Terraform, OpenTofu, and more with policy-as-code guardrails.

02 · Decomposing the Terralith

One monolith becomes isolated stacks per client

Breaking up the Terralith turned one fragile monolith into a fleet of small, fast, independent deployments.

Isolation: every client deployment now has its own resources and state, so changes to one client can never ripple across the rest.
Standardization: reusable modules for common infrastructure components replaced repetitive copy-paste configuration, with dependencies explicitly defined.

The payoff: client infrastructure that took 25 minutes just to plan now deploys end-to-end in under 3 minutes, sub-1 minute in some cases.

Plan and apply cycles reduced from 25 minutes to under 3 minutes

03 · Terraform → OpenTofu

Open source, zero vendor lock-in

In the same migration, every resource moved from Terraform to OpenTofu, the open-source fork of Terraform under the Linux Foundation. One migration window, two platform upgrades.

Licensing risk eliminated. Terraform’s license change no longer hangs over the platform’s future.
No commercial dependence. Power Digital is protected from vendor lock-in and pricing changes down the road.
A thriving community. OpenTofu’s momentum echoes the early Terraform community, and Masterpoint is a proud, active member of it.

OpenTofu, the open source infrastructure as code tool under the Linux Foundation

Handed off as a working platform, not a black box

Masterpoint closed the engagement with hands-on Spacelift and OpenTofu workshops, best-practices sessions, and troubleshooting runbooks. Power Digital’s platform team now ships its own improvements on infrastructure it fully understands and owns.

The Outcomes

Business Impact

ThenTerraform Cloud · one 43,000-resource Terralith

A required upgrade projected at ~$5,000/month, just to keep scaling.
25-minute plan and apply cycles, with 63 engineering hours lost every month.
Over 50% of runs failed, forcing manual provisioning workarounds.
Client onboarding, the revenue engine, bottlenecked by the platform.

NowSpacelift · OpenTofu · isolated client stacks

Superior automation for under $500/month, a 10x cost reduction.
End-to-end client deployments in under 3 minutes, sub-1 in some cases.
Failed runs eliminated, with provisioning fully automated and reliable.
100+ new clients onboarded without issue in the first 60 days.

Contained Blast Radius

Per-client isolation, standardized security practices, and improved access control mean a change to one client can no longer put every client at risk.

Nimble Disaster Recovery

Fast cycles, reliable automation, and isolated infrastructure let the team stand up replacement infrastructure quickly when trouble hits, before clients feel it.

Auditable by Default

Every change is version-controlled and peer-reviewed, with full audit trails across client deployments instead of untraceable console operations.

Takeaways

Three questions worth asking about your own platform

This project wasn’t a failure of Power Digital’s original design. It was an evolution to remove scaling limits — a good problem to have. It comes down to three questions:

Is it cost-effective?

Are you paying for capability, or paying a premium to work around your platform’s limits?

Does it meet today's needs?

Failed runs, long feedback loops, and manual workarounds are signals, not normal operating costs.

Can it handle tomorrow's?

The setup that got you here may not get you to 10x the clients, resources, and teams.

It’s okay if the answer to any of these is “no,” especially the last one. That’s not a verdict on your team — it’s a signal that the platform is ready to evolve.

👋 Curious if Masterpoint could help your team too? We partner with engineering organizations across the full spectrum — from unicorn startups to Fortune 500 enterprises — to deliver infrastructure systems and patterns that truly scale. Get in touch and we'll walk through your environment, your roadmap, and see where we can help.

Breaking Up a 43,000-Resource Terralith for a 10x Reduction in Infrastructure Automation Costs