# Why data platform bills surprise teams, and how to keep lakehouse costs predictable

> Lakehouse bills surprise teams for the same few reasons: clusters left running, premium serverless rates, platform fees layered on top of cloud compute, unbounded ingestion and AI usage, and estimates that do not match real workloads. Predictable costs come from transparent unit pricing, auto-stop, per-team budgets and alerts, and right-sized compute.

Source: https://datanature.ai/blog/lakehouse-cost-surprises-and-how-to-avoid-them
Publisher: Dlytica Inc. (DataNature is a Dlytica company)
Updated: 2026-11-05

## What do practitioners complain about most?

In data engineering forums the same themes repeat: unpredictable monthly bills, calculators that underestimate real usage, idle compute, serverless premiums, and a platform fee on top of the underlying cloud compute.

## Where does the money actually go?

- **Idle clusters** that keep running after jobs finish
- **Oversized compute** chosen "to be safe"
- **Always-on SQL endpoints** for dashboards used a few hours a day
- **Ingestion** that re-reads whole tables instead of changes
- **AI and agent usage** with no budget or rate limit
- **Two copies of data** in a lake and a separate warehouse

## How do you keep costs predictable?

1. **Know your unit price.** Understand exactly what one unit of compute costs and what it includes.
2. **Auto-stop everything.** Interactive and SQL compute should stop when idle.
3. **Use change data capture** instead of full reloads.
4. **Budgets per team and per agent,** with alerts before limits, not after.
5. **Right-size with real metrics** after two weeks of production, not before.
6. **One copy of data** on open tables, served to every engine.

## How does DataNature price compute?

In DataNature Units (DNUs): vCPU divided by two plus RAM in GB divided by six, per hour of provisioned compute. You can estimate any cluster with the [pricing calculator](/pricing). Atlas shows usage and cost per workspace, team and agent, and the AI gateway enforces model budgets.

## Is a cheaper platform always better?

No. The cheapest platform is the one your team can run well. Count engineering time, migration effort and the cost of lock-in as well as compute.

## How do DataNature's compute types affect cost?

Each compute type has a different cost shape, and putting work on the right one is the biggest lever:

- **Spark (batch)**: the most expensive per hour and the most efficient per gigabyte. Use incremental dbt models so jobs process only new data.
- **Trino (interactive)**: pay for concurrency. Keep exploratory queries off Bronze and pre-aggregate common joins in Gold.
- **DN Warehouse (serving)**: materialize frequently used cubes so dashboards hit small views instead of large tables.
- **StreamHouse (hot path)**: runs continuously, so keep it to the tables and events where seconds matter.

DataNature bills DataNature Units on provisioned compute: DNU = vCPU ÷ 2 + RAM (GB) ÷ 6, per hour. Because the formula is published, you can size and forecast a workload before running it.

## How does DataNature Atlas keep this secure and compliant?

Governance is part of the architecture, not a separate tool bolted on later. [DataNature Atlas](/platform/atlas) is the platform's governance module:

- **Knows your data**: a catalog with automated PII detection and classification across Bronze, Silver and Gold.
- **Controls access**: role-based grants plus attribute-based policies, row filters and column masks (null, redact, partial, hash, year only).
- **Governs AI**: Voir agents are scoped principals; masks, row filters and PII guardrails apply to every tool call and answer.
- **Proves it**: lineage from source to metric to decision, and an audit log of who or what accessed which data.
- **Stays local**: Atlas runs inside your deployment, so policies, lineage and logs never leave your jurisdiction.

These controls support GDPR, PIPEDA, Quebec Law 25, CCPA, HIPAA and sector rules such as OSFI E-23. Read [DataNature Atlas in depth](/blog/datanature-atlas-governance-rbac-abac-pii-masking) or see [security and compliance](/security).
