dbt Cloud vs Dataform vs Databricks Workflows: Real Analytics Engineering Cost
Short answer
Dataform, now owned by Google, has zero licensing cost at all — you only pay for the underlying BigQuery compute it runs on — but it works exclusively with BigQuery, making it a non-option for any team on Snowflake, Redshift, or Databricks. dbt Cloud is confirmed at $100/developer/month for its Starter tier (up to 5 seats), with Enterprise pricing negotiated and reported at $200–400/seat/month, or $90,000–400,000/year for larger organizations. Databricks Workflows isn't priced separately at all — its cost is folded entirely into Databricks' general DBU consumption model, making a direct per-seat comparison against dbt Cloud or Dataform structurally impossible.
Pricing at a glance
| dbt Cloud | Dataform | Databricks Workflows | |
|---|---|---|---|
| Public rate card | Yes, for Developer and Starter tiers; Enterprise and Enterprise+ are custom-quoted | Yes — free, confirmed | No separate rate card — Workflows is a feature of the broader Databricks platform, billed through general DBU consumption |
| Free tier | Yes — Developer: 1 seat, 3,000 models/month, 14-day trial of Starter features | Yes — the entire product has zero licensing cost | Databricks' general Free Edition applies (non-commercial, limited compute), not specific to Workflows |
| Entry paid tier | Starter: $100/user/month, up to 5 seats, 15,000 successful model builds/month, 5,000 queried metrics/month, 1 project | Not applicable — always free; only the BigQuery compute it triggers is billed | Not applicable as a standalone item — billed as ordinary Databricks job/compute consumption |
| Enterprise tier | Custom-quoted; Vendr transaction data reports $200–400/seat/month negotiated, and a 20-seat Enterprise deployment reported at $36,000–84,000/year after negotiation. A separate source reports broader Enterprise contracts at $90,000–400,000/year for larger organizations | Not applicable | Not applicable as a distinct tier |
| Warehouse/cloud lock-in | Works across Snowflake, BigQuery, Redshift, and Databricks — genuinely warehouse-agnostic | BigQuery only — a hard structural limitation, not a pricing choice | Databricks only, by definition |
| Median reported contract value | $26,460/year across 143 tracked transactions (range $14,404–87,914), though this blends Starter- and Enterprise-tier buyers together | Not applicable — no licensing spend to measure | Not separately measurable from broader Databricks spend |
What headline pricing excludes
Neither dbt Cloud's nor Dataform's pricing includes the warehouse compute their transformations actually run on. Both tools push SQL execution into the customer's own Snowflake, BigQuery, Redshift, or Databricks environment — the dbt Cloud or Dataform bill (where one exists) is entirely separate from, and can be smaller than, the warehouse compute bill generated by running the transformations themselves. For active teams running many models across development, staging, and production, warehouse compute cost from these runs is reported to potentially exceed the dbt Cloud subscription cost itself.
A widely circulated "dbt Enterprise starts at $50,000/year" figure is explicitly not an official dbt number — dbt Labs does not publish any Enterprise price at all, and this figure appears to be a secondary estimate that has taken on a life of its own in comparison content. The genuinely reported Enterprise range, from independent transaction data, spans considerably wider ($90,000–400,000/year for larger organizations, or $36,000–84,000/year specifically for a 20-seat deployment after negotiation).
Dataform's "free" pricing genuinely means zero licensing cost, but it isn't free of trade-offs. Its complete lock-in to BigQuery is a hard structural limitation, not a configuration choice — a team on any other warehouse cannot use it regardless of budget, and a team currently on BigQuery that later needs to move to another warehouse would need to migrate its entire transformation layer away from Dataform as part of that move.
Databricks Workflows has no identifiable cost of its own to compare against dbt Cloud or Dataform — orchestrating jobs through Workflows consumes ordinary Databricks compute (billed in DBUs, at rates varying by compute type from roughly $0.07 to $0.70/DBU), meaning "the cost of Databricks Workflows" is really just "the cost of whatever Databricks compute your workflows happen to consume."
Hidden costs
- dbt Cloud charges for CI/CD job runs in pull-request workflows, separately from the base seat cost — a high-velocity team merging many pull requests per day will accumulate CI run costs that a simple per-seat budget doesn't anticipate.
- dbt Cloud's compute-overage charges are a real, reported pain point — one specific account reported hitting overages within two weeks of starting a $300/month plan, illustrating how quickly a build-volume-based plan can exceed its included allocation for an active team.
- Adding developer seats to dbt Cloud mid-contract typically triggers pro-rated charges at the full contracted per-seat rate, not a blended or discounted rate — a real cost consideration for a growing team adding headcount between renewal cycles.
- Databricks' Standard tier is being actively retired during 2026 (new Azure Standard workspace creation was blocked April 1, 2026, with full automatic upgrade to Premium by October 1, 2026) — a genuinely current change that pushes many Databricks Workflows users onto pricier Premium-tier DBU rates regardless of their own preference.
- The compute type selected for a Databricks Workflows job is the single largest cost lever in the entire Databricks platform, since the DBU rate for functionally similar work can differ by 3–4x across different cluster/compute types — a configuration decision, not a pricing-tier decision, drives most of the real cost variance.
Worked scenarios
3 analytics engineers
| Estimated monthly cost | |
|---|---|
| dbt Cloud Starter | 3 × $100 = $300/month, comfortably within the 5-seat Starter cap, before warehouse compute |
| Dataform (BigQuery-native team only) | $0 in licensing, plus BigQuery compute cost for the transformations themselves |
| Databricks Workflows | No standalone figure — cost is whatever DBU consumption the team's specific jobs generate, which could be anywhere from very low (light, infrequent jobs on cheap compute types) to substantial (frequent, heavy transformations on expensive interactive compute) |
At this team size, a BigQuery-native team has a genuine, quantifiable reason to consider Dataform purely on cost — $300/month in avoided dbt Cloud licensing is a real saving for a team willing to accept BigQuery lock-in.
10-person team
| Estimated annual cost | |
|---|---|
| dbt Cloud (exceeds Starter's 5-seat cap, moves to negotiated Enterprise) | Using the reported $200–400/seat/month negotiated range: 10 × $200–400 × 12 = $24,000–48,000/year, before warehouse compute |
| Dataform (BigQuery-native team only) | $0 in licensing, regardless of team size, plus BigQuery compute |
| Databricks Workflows | No standalone figure; scales with actual job volume and compute-type selection rather than headcount |
This is the scale where dbt Cloud's per-seat Enterprise cost becomes a genuinely significant line item, while Dataform's zero-licensing model continues to cost nothing regardless of team growth — provided the team's warehouse choice permits it.
Enterprise team
| Estimated annual cost | |
|---|---|
| dbt Cloud (20-seat reported example) | Reported at $36,000–84,000/year after negotiation for a 20-seat Enterprise deployment |
| dbt Cloud (broader Enterprise range for larger organizations) | Reported at $90,000–400,000/year |
| Dataform | Still $0 in licensing at any team size, for a BigQuery-native organization |
| Databricks Workflows | No standalone figure; enterprise-scale Databricks compute spend is driven by total job volume and compute-type mix across the whole organization, not analytics-engineering headcount specifically |
Normalizing across billing units
dbt Cloud's per-seat pricing and Dataform's zero-licensing model are directly comparable — the entire question is whether a team can accept BigQuery-only lock-in in exchange for eliminating a real, per-seat cost. Databricks Workflows resists this comparison entirely, since it has no per-seat unit at all; its cost is inseparable from the broader Databricks DBU consumption a team would incur regardless of which specific orchestration approach it used within the platform.
Break-even and crossover
The clearest, most calculable finding in this article: a BigQuery-native team can eliminate its entire dbt Cloud licensing cost by using Dataform instead — $300/month for 3 engineers, $24,000–48,000/year for a 10-person team, up to $90,000–400,000/year for a large enterprise deployment. This is a genuine, quantifiable saving, but it is not a cost-free swap: BigQuery lock-in is the most obvious trade-off, though it isn't the only one. dbt Cloud brings a larger community, more mature CI and governance tooling, broader third-party integrations, and vendor support that Dataform's leaner feature set doesn't fully replicate — a team choosing Dataform is trading some of that depth for the licensing saving, not getting an identical product for free.
Who pays more, and when
- A BigQuery-only team should seriously evaluate Dataform before defaulting to dbt Cloud, given the complete elimination of per-seat licensing cost — but should weigh this against genuine differences in feature depth, CI and governance tooling, community size, and available support, not treat the two as functionally interchangeable products that differ only in warehouse compatibility.
- A multi-warehouse or warehouse-agnostic team, or one that anticipates possibly switching warehouses in the future, needs dbt Cloud (or self-hosted dbt Core) specifically because Dataform's BigQuery lock-in forecloses that flexibility entirely.
- A small team (under 5 seats) fits cleanly and affordably within dbt Cloud's Starter tier at $100/seat/month — the jump to negotiated Enterprise pricing only becomes relevant once a team exceeds this seat cap.
- A team already fully committed to Databricks for its broader data platform may find it makes the most sense to keep transformation orchestration inside Databricks Workflows rather than adding a separate dbt Cloud subscription on top — though this trades dbt's warehouse-agnostic tooling and large community for tighter platform integration.
- Any team negotiating dbt Cloud Enterprise pricing should reference Vendr's reported $200–400/seat/month range and the specific $36,000–84,000/year example for 20 seats as a benchmark, and should be skeptical of any secondary source's "$50,000/year Enterprise" figure, which does not originate from dbt Labs.
Limitations and uncertainty
dbt Cloud's Developer and Starter tier pricing is confirmed directly from dbt's own pricing page; Enterprise and Enterprise+ pricing is not published, and the figures shown are drawn from independent transaction-data benchmarks (Vendr) rather than an official rate card. Dataform's zero-licensing-cost model is confirmed, though the BigQuery compute cost it triggers is not separately estimated here, since it depends entirely on transformation volume and complexity. Available public sources do not identify a standalone price for Databricks Workflows — its cost is inseparable from general Databricks DBU consumption, so no artificial standalone figure is presented.
Official sources
- getdbt.com/pricing (Developer and Starter tiers confirmed; Enterprise custom-quoted)
- Dataform's zero-licensing-cost model is confirmed via Google's own product documentation
- databricks.com/product/pricing (general DBU rate structure; Workflows has no separate published rate)