Stop Guessing Your Databricks Bill.
The Billing Dashboard Is Here.

The Problem With Databricks Billing Visibility

Databricks bills in DBUs — Databricks Units — which map to dollars via per-SKU pricing. A single runaway job, an accidentally left-on SQL warehouse, or a pipeline running on All-Purpose compute instead of Jobs compute can easily add hundreds of dollars before anyone notices.

The official billing UI lives in the Databricks console under Account > Usage. It's fine for finance teams doing monthly reviews. It's not designed for engineers who want to check their spend while they're writing code.

That's the gap CatalystOps fills.

How It Works

CatalystOps queries the system.billing.usage Unity Catalog system table directly, joining against system.billing.list_prices to convert DBUs into dollar figures. The query runs through your existing SQL warehouse — no extra infrastructure, no external services.

SELECT
    CAST(u.usage_date AS STRING)           AS date,
    u.billing_origin_product              AS workloadType,
    u.identity_metadata.run_as            AS runAs,
    u.usage_metadata.job_id               AS jobId,
    u.sku_name                            AS skuName,
    SUM(u.usage_quantity)               AS dbus,
    SUM(u.usage_quantity * lp.pricing.default) AS dollars
FROM system.billing.usage u
LEFT JOIN system.billing.list_prices lp
    ON  u.sku_name = lp.sku_name
    AND u.usage_start_time >= lp.price_start_time
WHERE u.usage_date >= '${startDate}'
  AND u.usage_date <= '${endDate}'
  AND u.record_type = 'ORIGINAL'
GROUP BY date, workloadType, runAs, jobId, skuName
ORDER BY dollars DESC

Results are aggregated into a BillingSummary object and surfaced in the sidebar tree view, broken down three ways: by user, by workload type, and by job. Each view is sorted by dollars descending so the most expensive items surface immediately.

What You See in the Dashboard

Time periods

The dashboard supports three built-in periods — last 24 hours, last 7 days, and last 30 days — plus custom date ranges. Period selection is remembered between sessions via the 1-hour result cache, so you're not re-running the SQL every time the panel opens.

Warehouse auto-discovery

CatalystOps automatically selects a SQL warehouse to run the billing query, preferring serverless warehouses that are already running. If you want to pin a specific warehouse, set catalystops.billing.warehouseId in your VS Code settings.

Why This Matters for Engineers

Finance teams already have billing dashboards. This feature is for the engineer who just submitted a job and wants to know if it's eating $50 or $500 before it becomes a Slack message from their manager.

• Catch All-Purpose compute being used where Jobs compute should be (3x cost difference)
• Spot runaway jobs before they finish their full run
• Attribute spend to users so teams can self-govern their usage
• Correlate cost spikes with code changes — the data is right next to your editor

MCP Integration

The billing data is also exposed through the CatalystOps MCP server via the get_billing_summary and refresh_billing tools. This means you can ask Claude directly: "Which of my jobs cost the most last week?" and get a real answer backed by live data from your workspace.

Getting Started

1. Install CatalystOps from the VS Code Marketplace
2. Connect your Databricks workspace (CatalystOps: Configure Connection)
3. Open the CatalystOps sidebar panel
4. Click Fetch Billing Data — confirm the prompt
5. Switch between Last 24h / 7d / 30d to explore your spend

Unity Catalog with System Tables must be enabled on your workspace. If you see a "TABLE_OR_VIEW_NOT_FOUND" error, contact your Databricks workspace admin to enable system tables.