New $0 markup on your cloud compute — your data lives in your cloud account →
Pricing

Priced on capacity. Not per row, per seat, or per task.

Every plan runs the whole stack in your own cloud account. Plans differ by how much platform capacity they cover — never by which applications you're allowed to run.

Most common
Starter
One environment, one cloud
$1,000
per month
$12,000
$10,000
per year, billed once
Save $2,0002 months free
40 vCPU
sustained capacity
30,000 vCPU-hours / month
  • 1 environment
  • 10 users
  • One cloud — AWS, GCP, or Azure
  • Every application: Airflow, Airbyte, DataHub, Superset
  • SSO and SAML included
  • First response: 1 business day
Request early access
Most single-environment platforms we measure land here.
 
Team
Production plus staging, or a second team
$3,000
per month
$36,000
$30,000
per year, billed once
Save $6,0002 months free
80 vCPU
sustained capacity
60,000 vCPU-hours / month
  • 3 environments
  • 25 users
  • One cloud — AWS, GCP, or Azure
  • Every application: Airflow, Airbyte, DataHub, Superset
  • SSO and SAML included
  • First response: 8 business hours
Request early access
 
 
Business
Several environments, more than one cloud
$6,000
per month
$72,000
$60,000
per year, billed once
Save $12,0002 months free
200 vCPU
sustained capacity
150,000 vCPU-hours / month
  • 10 environments
  • Unlimited users
  • Multi-cloud — AWS, GCP, and Azure
  • Every application: Airflow, Airbyte, DataHub, Superset
  • SSO and SAML included
  • First response: 4 business hours
Request early access
 
 
Enterprise
Committed capacity and a signed SLA
Custom
priced to your commitment
Custom
annual or multi-year terms
Committed
sustained capacity
Capacity agreed up front
  • Unlimited environments
  • Unlimited users
  • Multi-cloud — AWS, GCP, and Azure
  • Every application: Airflow, Airbyte, DataHub, Superset
  • SSO and SAML included
  • First response: 1 hour for P1, 24x7
Talk to us
Pipeline and model development included.
Every plan starts with a 30-day free trial. Cloud infrastructure is billed to you by your cloud provider, at your own rates — we never mark it up.
How capacity works

Capacity is one pool of vCPU-hours per month, measured from the nodes actually running your platform. It exists so the price tracks the size of what we operate — not so it can be used against you.

One pool, not a quota per application
Your allowance covers every environment and every application on the plan. Nothing is metered per DAG, per row, per connector, or per seat.
Bursts cost nothing extra
Capacity is counted in vCPU-hours across the whole month, so a heavy backfill day is absorbed by the quieter days around it. One platform we measured swung from 32 to 101 vCPU inside a single month and stayed comfortably inside its allowance.
We tell you at 80%
You get a notification when the month reaches 80% of the allowance, and again at 100%. Nobody finds out from an invoice.
Nothing throttles, and there is no overage rate
Going over does not slow your platform down, queue your tasks, or bill you an increment. We talk to you about whether a different plan fits. A billing threshold should never be able to break a running data platform.
Which plan fits

These are measured shapes, taken from real production platforms rather than from a sizing spreadsheet.

One Airflow environment
Measured at 30–40 vCPU sustained on a live production platform. A single environment is smaller than most teams expect.
Starter
$1,000 / MO
Airflow and Airbyte together
Also 30–40 vCPU sustained. Adding a second application did not move the number — node-pool sizing sets platform size, not the application roster.
Starter
$1,000 / MO
Airflow, Airbyte, and DataHub
55–65 vCPU sustained, bursting past 100 during backfills. The bursts are absorbed by the monthly pool.
Team
$3,000 / MO
Production and staging side by side
A second environment doubles the floor before it does anything else, because each one carries its own control components.
Team
$3,000 / MO
Several environments, or more than one cloud
Multi-cloud, unlimited users, and a named engineer on escalation.
Business
$6,000 / MO
What the measurement showed

The number of applications you run does not predict the size of your platform. One platform we measured runs a single application at 36 vCPU; another runs two at 32. Size comes from how the node pools are sized — which is exactly why we never gate applications by tier.

Support

Support ladders on response time, not on whether you're allowed to ask. The in-product messenger reaches us from every plan.

CommitmentStarterTeamBusinessEnterprise
In-product supportIncludedIncludedIncludedIncluded
First response1 business day8 business hours4 business hours1 hour for P1, 24x7
EscalationSupport queueSupport queueNamed engineerNamed engineer, quarterly review
Shared Slack channel——IncludedIncluded
Contractual SLA———Yes, with credits
Where the product ends and services begin

The line is labor, not capability. The subscription operates the platform. Building what runs on it is an engagement, priced separately — included at Enterprise, an add-on everywhere else.

Your subscription covers
  • Provisioning the platform into your cloud account, and tearing it down
  • Version upgrades, patches, and the compatibility work behind them
  • Health monitoring, alerting, and the analytics behind each application
  • Identity, SSO, per-application roles, and audit trails
  • The console, the CLI, and the API
  • Support at your plan's response time
Professional services covers
  • Writing and maintaining your DAGs and pipelines
  • Building custom connectors and source integrations
  • Data modelling, warehouse design, and dbt work
  • Migrating pipelines off an existing orchestrator
  • Embedded platform or analytics engineering
  • Training your team on the stack
If you want the pipelines built as well as the platform run, say so when you get in touch and we'll scope it as an engagement.
Common questions
What counts toward my capacity?
The vCPU of the nodes running your platform, sampled continuously and summed into vCPU-hours across the month. One vCPU held for one hour is one vCPU-hour, so a platform sitting on 10 vCPU for a full month is about 7,300 of them. It is measured, not estimated — the same node scan that reports your cluster's health reports its size.
What happens if I go over the allowance?
Nothing breaks. There is no overage rate, no throttling, and no automatic upgrade. You get a notification at 80% and again at 100%, and we have a conversation about whether a different plan fits. If your workload is consistently above the allowance we would rather move you to the right plan than surprise you with an invoice.
What does BYOC mean in practice?
Your data, compute, and storage all run in your own cloud account — GCP, AWS, or Azure. SkaleData operates a lean control plane for orchestration and auth. We don't store your data. It passes through the control plane only when you use a feature that needs it, such as the web terminal or the log viewer. For regulated industries, this is the difference between a compliant architecture and a non-starter.
What's an environment, exactly?
An environment is a managed deployment of the SkaleData stack — Airflow, Airbyte, DataHub, and Superset — running on Kubernetes in your cloud account. Most teams start with one for production. Larger teams run separate environments for production and staging, which is where the Team plan's 3-environment allowance starts to matter.
Do I pay for the cloud infrastructure separately?
Yes — and that's the point. Because it's BYOC, you pay your cloud provider directly for the nodes, storage, and networking, at your own negotiated rates and against your own committed-spend discounts and credits. We never mark up your infrastructure. What that costs depends on your cloud, your region, and your commitments, so price it with your provider's own calculator: the platforms we measured run 30–65 vCPU sustained, and that vCPU figure is the input. SkaleData's fee covers the managed platform on top of it.
Why isn't SSO paywalled?
Because paywalling SSO is a racket. Most vendors — Astronomer, Fivetran, Snowflake — charge tens of thousands a year to unlock it. It's table stakes for any real company. We include it from day one because it shouldn't be a deal-blocker.
Monthly or annual?
Either. Annual is billed once for 10 months of the year, so you get 2 months free against the monthly rate. Monthly has no commitment beyond the current month.
Can I change plans later?
Yes, from the organization settings in the console. Upgrades take effect immediately and are prorated; downgrades take effect at the end of the billing period. Your platform keeps running through either.

Start with 30 days free.
The whole platform, in your own cloud.

BYOC. SSO included. No per-row billing, ever.

Request early access →Talk to us about Enterprise