Skip to content

The Best Platform for Infrastructure as Code and Provisioning Without a Platform Team (2026): What Each Approach Provisions and What You Still Own

Infrastructure-as-code file provisioning a database, cache, bucket, network, and secrets
Blog
By Noah Grinstein
13 min read

Without a platform team, the lowest-effort setup pairs a managed runtime for applications with a hosted IaC run platform for the few cloud resources around them, so nobody maintains state backends, runners, or clusters. The best fit depends on whether most of your infrastructure is application workloads (favor a managed runtime or a workload platform such as Control Plane) or cloud resources like networks, databases, and IAM (favor a hosted IaC platform such as HCP Terraform, Spacelift, env zero, or Pulumi Cloud).

Disclosure: Control Plane publishes this guide.

TL;DR

Three different kinds of products get called “PaaS for infrastructure as code,” and a team without platform engineers usually needs two of them.

  • Hosted IaC run platforms (HCP Terraform, Spacelift, env zero, Scalr, Pulumi Cloud) remove state backends and CI runners, but your team still writes and owns every module.
  • Developer portals and IDP orchestrators (Backstage, Port, Humanitec) give developers a self-service front end, but someone has to build the templates and the provisioning logic behind them first.
  • Managed runtimes (serverless container services and workload platforms) remove most application-layer IaC and shrink what is left to networks, IAM, provider databases, and a few other cloud resources.
  • Someone always has to own the modules, templates, or programs behind any portal or UI.
  • Control Plane fits as a workload platform for the application layer, with a small Terraform or Pulumi stack for the cloud resources outside it.

What is the best platform as a service for infrastructure as code and provisioning without a platform team?

No single product covers every layer, so the right combination depends on where most of your infrastructure work lives.

ApproachExamplesWhat your team writesWhat the service runs for youGuardrails built inWho still has to own it
Self-run IaC CLI in CITerraform or OpenTofu, Pulumi with a self-managed backendCode for every resource, CI pipeline config, state backend setupNothing beyond what your CI runner executesNone unless you add themState backend, runner security, credentials, module design, drift detection
Hosted IaC run platformsHCP Terraform, Spacelift, env zero (formerly env0), ScalrHCL or code for every resourceState storage, plan and apply runs, variable encryption, drift detection on paid tiersPolicy as code (Sentinel or OPA), plan approvals, RBACModule design, credentials, resource lifecycle, Day 2 operations
Pulumi CloudPulumi Cloud with Pulumi Deployments and Pulumi ESCTypeScript, Python, Go, C#, or Java programsState storage, deployments, secrets and environment configurationPolicy as code (Pulumi Policies, formerly CrossGuard), RBAC on paid tiers, ESC environment access controlsProgram design, credentials, resource lifecycle, Day 2 operations
Developer portals and IDP orchestratorsBackstage Software Templates, Port self-service actions, Humanitec Platform OrchestratorTemplates, actions, or resource definitions that call underlying IaC or APIsSoftware catalog, self-service UI, action orchestrationPortal RBAC, approval flowsEvery underlying IaC module or API integration, template maintenance, and the portal itself if self-hosted
Cloud-native single-cloud toolsAWS CDK with CloudFormation, Azure BicepConstructs or Bicep files for every resourceStack or deployment execution; CloudFormation also tracks state and rolls back failed updatesStack policies, CloudFormation drift detection, Bicep what-if previewsConstruct or module design, IAM, networking, Day 2 operations, one cloud only
Serverless container servicesGoogle Cloud Run, Azure Container Apps, Amazon ECS on FargateContainer images, service configuration, IaC for surrounding resourcesContainer scheduling and autoscaling; Cloud Run and Container Apps also provide load balancing and TLS on default endpoints, while ECS needs an Application Load Balancer you provisionService-level IAM, ingress settingsVPCs, databases, DNS, IAM, and the IaC for them
Workload platformsControl PlaneContainer images, workload definitions (console, CLI, Terraform or Pulumi provider, or MCP server), IaC for cloud resources outside the platformContainer and VM orchestration across clouds, autoscaling, load balancing, TLS certificates, mTLS, internal firewall, audit trail, observabilityDeny-by-default firewall, policy-based access, identity-based secret grants, tamper-proof audit trailCloud resources outside the platform (VPCs, provider databases, cloud storage and queue services, account-level IAM)

Several hosted IaC platforms reduce the “who writes the module” burden with self-service catalogs on paid tiers. HCP Terraform no-code provisioning (Standard and Premium), Spacelift Blueprints (Business and up), and env zero self-service templates let one engineer publish an approved module that others deploy from a form. Port self-service actions and Backstage Software Templates give developers a UI, but someone must wire each one to the provisioning logic behind it. The Humanitec Platform Orchestrator generates resource graphs at deployment time from resource definitions that someone has to write first. The top infrastructure as code tools compares the IaC engines themselves.

Which resources does each approach provision?

Each resource your application needs is provisioned by one layer, and that layer decides how much code your team writes and maintains.

ResourceHosted IaC platformDeveloper portalServerless containersWorkload platform (Control Plane)
Application containersYour team writes IaC for the compute serviceA template calls the underlying IaC or APIThe service runs containers from your imageA workload definition runs containers in every location of its Control Plane Global Virtual Cloud (GVC), with no cluster to manage
Autoscaling and load balancingYour team writes scaling policies and load balancer resourcesA template calls the underlying IaCBuilt in for Cloud Run and Container Apps; ECS needs a load balancer in IaCBuilt in per workload
TLS certificates and domainsYour team writes ACM, managed certificate, or cert-manager resourcesA template calls the underlying IaCBuilt in for Cloud Run and Container Apps default endpoints; custom domains need configuration; ECS uses ACM on a load balancer you provisionControl Plane creates and renews TLS certificates for default endpoints and custom domains
SecretsYour team writes secret manager resources; the platform encrypts run variablesA template calls the underlying IaCThe service references a cloud secret manager you provisionStored in Control Plane and granted to workloads through identity-based policies
Provider-managed databasesYour team writes RDS, Cloud SQL, or equivalent resourcesA template calls the underlying IaCNot provisioned by the serviceNot provisioned by the platform; stays in IaC
Open-source databases you run yourselfYour team writes compute, storage, and configurationA template calls the underlying IaCNot a fit for persistent databasesControl Plane’s Template Catalog deploys services such as PostgreSQL and Redis from templates
VPCs and networksYour team writes VPC, subnet, and peering resourcesA template calls the underlying IaCDefault networking; custom networking needs IaCNot provisioned by the platform; stays in IaC, and Control Plane’s Cloud Wormhole reaches private resources in your VPCs through an agent you install
IAM roles and policiesYour team writes IAM resourcesA template calls the underlying IaCA service identity; additional IAM needs IaCControl Plane’s Universal Cloud Identity (UCI) lets workloads call AWS, GCP, and Azure services with no long-lived cloud keys in your code, and Control Plane creates the least-privilege cloud role or service account for each identity; the one-time cloud account grant stays with you, and account-level IAM stays in IaC
DNS zonesYour team writes Route 53, Cloud DNS, or equivalent resourcesA template calls the underlying IaCNot provisioned by the serviceYour zone stays with your DNS provider; delegate a subdomain to Control Plane with an NS record and Control Plane creates its records, or point a CNAME at the workload endpoint
Object storage bucketsYour team writes S3, Cloud Storage, or equivalent resourcesA template calls the underlying IaCNot provisioned by the serviceThe Template Catalog deploys S3-compatible object storage such as SeaweedFS; cloud buckets such as S3 stay in IaC, and workloads reach them through UCI
Queues and event servicesYour team writes SQS, Pub/Sub, or equivalent resourcesA template calls the underlying IaCNot provisioned by the serviceThe Template Catalog deploys message brokers such as Kafka, NATS, and RabbitMQ; cloud services such as SQS or Pub/Sub stay in IaC
Per-environment preview copiesYour team duplicates workspaces or stacksTemplates can take an environment parameterRevisions or separate services per environmentA separate GVC holds the preview copy’s workloads

Hosted IaC platforms and portals are general purpose, so your team writes every resource definition. Serverless container services take the application layer and leave the rest to you. A workload platform such as Control Plane also covers TLS, secrets, per-workload cloud identity, DNS for delegated subdomains, and open-source databases, message brokers, and S3-compatible storage from its Template Catalog, while cloud resources outside it stay in IaC.

What should a small team start with?

The right starting stack depends on team size, compliance scope, and how many cloud resources live outside the application layer.

TeamApplication layerCloud resourcesGuardrailsWhat to add later
1 to 5 engineersA managed runtime or a workload platform such as Control PlaneA small Terraform or Pulumi stack for networking, a provider database, and DNS, run on a hosted IaC platform’s free tier (HCP Terraform Free covers up to 500 managed resources)Plan review on pull requests; a few policy-as-code rules; OIDC so CI holds no long-lived keysEnvironment promotion gates, more policy rules
5 to 20 engineersThe same runtime, with more services and environmentsA paid hosted IaC tier with a workspace or stack per environment and shared modulesPolicy as code, RBAC on IaC workspaces, OIDC for CIA self-service catalog for common resources, cost alerts
20 to 50 engineers with no platform teamThe same runtime, with separate environments per teamA module registry and shared modules owned by a rotating infrastructure groupMandatory plan approval for production, audit loggingA dedicated platform role, golden-path templates
Regulated team (SOC 2 or HIPAA evidence)A runtime with an audit trail and encryption at restA hosted IaC platform with drift detection, encrypted state, and access loggingPolicy as code for tagging, encryption, and network rules; OIDCContinuous compliance reporting, evidence export

Three practices cover the guardrails a platform team usually sets up first. Require plan review on every pull request, since hosted IaC platforms run a plan on each pull request, link the result there, and can require manual approval before an apply. Adopt policy as code early. HCP Terraform supports Sentinel, OPA, and its native HCL-based Terraform policy framework, Spacelift and env zero support OPA, and Pulumi Cloud supports Pulumi Policies (formerly CrossGuard). A handful of rules catches untagged, oversized, and unencrypted resources before apply. Use OIDC federation so CI authenticates to clouds with short-lived tokens and stores no cloud keys. Developer self-service with guardrails covers these patterns in more depth.

When do you actually need a platform team?

A team can run without a dedicated platform team while its tools absorb the operational work, and these signals show when that stops holding.

  • Services and environments grow faster than anyone can review changes across all of them.
  • You run several clouds or many accounts with separate networking and IAM boundaries.
  • Compliance scope requires continuous evidence collection, beyond point-in-time audits.
  • Custom networking such as transit gateways, private connectivity to on-premises systems, or a service mesh takes engineering time every sprint.
  • Teams copy IaC modules with small variations because nobody owns shared patterns.
  • Infrastructure incidents fall to whoever last touched the Terraform file.

When several of these signals appear together, it is time to staff a dedicated platform role. Production-grade infrastructure without a platform team covers how far managed tools can take you before that point.

How does Control Plane fit without a platform team?

Control Plane handles Day 2 operations for you, including autoscaling, built-in observability, and security.

Workloads run on Control Plane-managed locations in AWS, GCP, Azure, and Oracle Cloud regions with no Kubernetes cluster for you to build or operate. A GVC is Control Plane’s named set of locations, and a workload runs in every location of its GVC. Control Plane also runs VM workloads on locations joined through Control Plane’s Managed Kubernetes, and workloads on Kubernetes clusters you already run or your own machines, including on premises.

The CLI, console, Terraform and Pulumi providers, and the Control Plane MCP server all work through the same Control Plane API. Control Plane can export its configuration as Terraform, and the MCP server lets AI assistants deploy, diagnose, scale, and roll back workloads.

Security defaults include a deny-by-default internal firewall, mTLS between workloads, and a tamper-proof audit trail, and access is controlled through Control Plane policies. Control Plane is PCI DSS Level 1, SOC 2 Type II, HIPAA, and GDPR compliant.

Control Plane does not provision cloud resources outside it, such as VPCs, provider database services like Amazon RDS, cloud storage and queue services, or account-level IAM. Keep a small Terraform or Pulumi stack for those resources, run it on a hosted IaC platform such as HCP Terraform, Spacelift, env zero, or Pulumi Cloud, and use the Control Plane providers for workloads in the same repository.

The best internal developer platforms compares platform engineering tools for teams without a platform team, and standardizing CI runners and pipelines covers the build side.

FAQ

Short answers on infrastructure as code and provisioning without a platform team.

What is the best platform as a service for infrastructure as code and provisioning without a platform team?

Usually two products together, a managed runtime or workload platform for applications and a hosted IaC run platform for networks, databases, and IAM. The phrase spans hosted IaC platforms, developer portals, and managed runtimes, and no single product covers all three.

Is Terraform too much for a team without DevOps engineers?

Not on its own. The work comes from state backends, credentials, CI pipelines, and module design. A hosted IaC run platform such as HCP Terraform or Spacelift removes the state and runner work, and keeping Terraform to the few resources outside a managed runtime keeps the module count small.

What is the difference between a PaaS and an IaC platform?

A PaaS runs and manages your application runtime. An IaC platform provisions cloud resources from code you write, so you define and maintain each resource yourself.

Is Backstage worth it without a platform team?

Usually not at first. Backstage provides a portal and software catalog but does not provision infrastructure on its own, and someone has to build templates and run the instance. Small teams often start with the self-service features of a hosted IaC platform and add a portal once someone owns it.

Is Pulumi or Terraform better for a small team?

Either works, since both provision the same range of cloud resources. Pulumi uses general-purpose languages, which suits application engineers, while Terraform’s constrained HCL is easier to review for people who do not write application code daily.

How do developers self-serve infrastructure safely?

By choosing from pre-approved modules behind three guardrails, which are plan review on pull requests, policy as code, and OIDC federation so CI holds no long-lived cloud keys. Self-service catalogs in hosted IaC platforms let developers deploy approved modules from a form.

Do I still need infrastructure as code if I use a managed runtime?

Yes, for a smaller set of resources. VPCs, provider databases, account-level IAM, DNS zones, and cloud storage and queue services still belong in IaC.

How does Control Plane work with Terraform and Pulumi?

Through its Terraform and Pulumi providers, which define workloads, GVCs, identities, and policies as code next to your other cloud resources. Resources outside Control Plane stay in the standard cloud providers in the same repository.

Talk to the Control Plane team about running the application layer without a platform team, or sign up free.

Run Unbreakable
Apps and Agents

Sign up free. No cloud accounts required.