Databricks Doubles SQL Revenue: The Snowflake Squeeze Is Real

Databricks is reporting that revenue for its Snowflake-competing product roughly doubled, according to a Bloomberg headline dated June 16, 2026. That is the entire disclosed data point. The underlying article is gated behind a bot-detection wall (block reference cc0f881f-8988-11f1-9852-28f991937504 if you're curious), so what analysts have to work with right now is one number, one comparison, and a lot of context from the last three years of warehouse-versus-lakehouse positioning.

I'll be direct about the epistemic footing: a doubling is a 100 percent year-on-year growth rate, which is fast but not extraordinary for a product category that is still in land-grab mode. What matters is the base it doubled from, the margin profile of that revenue, and whether it is net-new workloads or migrations off Snowflake. None of that is in the source.

Key Details

The reportable facts are narrow. As Bloomberg.com framed it, Databricks sales doubled for the product that competes with Snowflake. The product in question is not named in the accessible headline, but by process of elimination in the current Databricks catalog, this maps to Databricks SQL, the serverless warehouse layer that sits on top of Delta Lake and directly targets Snowflake's core BI and analytics workloads. Databricks documents the architecture in the Databricks docs, but the exact SKU boundary Bloomberg is measuring is not disclosed.

Here is what the source does not tell us, and each unknown has a testable bound:

• The absolute revenue figure. If the base was 100 million dollars, doubling to 200 million is a meaningful but sub-tier line item against Snowflake's multi-billion product revenue reported in its most recent public filings. If the base was 500 million, the picture changes substantially. The source discloses neither.
• The comparison window. "Doubled" is presumably year-on-year, but it could be a trailing-twelve-month figure, a quarterly annualized run rate, or a bookings number. Each has different signal quality.
• Gross margin. Serverless SQL at scale is a compute-heavy business. Snowflake's published gross margins have been the benchmark. Databricks has historically run compute-heavier workloads on customer cloud accounts, which changes the margin math.
• Workload composition. Doubling on net-new AI-adjacent workloads is a very different story from doubling on migrated Snowflake dashboards.

The source also does not disclose customer counts, average contract value, or churn. That matters because a doubling driven by a handful of eight-figure enterprise expansions tells a different story than a doubling driven by broad mid-market adoption. Without that split, I would put the confidence interval on any strategic conclusion at roughly plus or minus one product cycle.

Why This Matters for Data Teams

The warehouse-versus-lakehouse debate has been running since roughly 2020. For most of that time, the honest engineering answer was: use Snowflake for BI and structured analytics, use Databricks for ML and unstructured pipelines, and glue them together with something like dbt. A doubling of Databricks SQL revenue is the market's way of saying that split is eroding.

The technical reason is straightforward. Once you have a governed table format (Delta, Iceberg, or Hudi), a decent query planner, and serverless autoscaling, the functional gap between a warehouse and a warehouse-on-a-lakehouse narrows to seconds of P95 latency and pennies of per-query cost. Snowflake's counter-move has been Iceberg tables and external-catalog interop, documented in the Snowflake docs. Databricks' counter-move has been Unity Catalog plus aggressive SQL performance work. Both vendors are converging on the same reference architecture from opposite sides.

For a platform lead choosing today, the decision is no longer "which engine". It's "which control plane do I want to be locked into for the next five years, and which one lets me keep the storage layer open". A doubling of the competing product is evidence, not proof, that the open-storage bet is paying off. It is not evidence that Snowflake is losing customers in absolute terms. Both can grow simultaneously in a market where analytics spend is still expanding.

There is also a second-order effect for teams running ClickHouse or self-hosted OLAP engines. If the two hyperscaler-adjacent vendors keep compressing on price and features, the "just run it yourself" calculus gets harder to justify for anything under a few petabytes. The unknown here, and it's a real one, is whether Databricks SQL pricing is actually competitive on steady-state dashboards or only on bursty workloads. The Bloomberg piece does not address unit economics.

Industry Impact

For the verticals RiverCore readers operate in, the implications split cleanly by workload shape.

In iGaming and ad-tech, where session-level event streams and sub-second aggregations dominate, the Databricks SQL story is interesting but not decisive. Those workloads still favor purpose-built OLAP engines or Snowflake's materialized views for the hot path. What changes is the cold-path economics: if you can keep raw events in Delta or Iceberg and query them ad-hoc without a separate warehouse ingest, you eliminate a whole pipeline layer. Teams using dbt as their transformation layer benefit either way, because dbt is engine-agnostic and the model code ports across.

In fintech, where governance and lineage are the binding constraint, the Unity Catalog versus Snowflake Horizon comparison matters more than raw query speed. A doubling in Databricks SQL revenue implies enterprise buyers are getting comfortable with Unity Catalog for regulated data, which was not obviously true eighteen months ago.

In crypto and DeFi analytics, where the workload is heavy on graph-like joins over historical chain data, neither vendor is a natural fit and most serious teams end up on custom infrastructure. The warehouse duopoly matters here mostly as a reporting layer downstream of the real compute.

The broader signal: analytics infrastructure is consolidating around two vendors with roughly symmetric capabilities and asymmetric go-to-market motions. That is good for buyers who negotiate hard, bad for buyers who don't benchmark, and neutral for anyone with an existing multi-year contract.

What to Watch

Three signals will tell us whether this doubling is a trend or a one-off.

First, Snowflake's next earnings print. If product revenue growth stays above roughly 25 percent year-on-year, the "Databricks is eating Snowflake" narrative is wrong and both are growing into an expanding market. If Snowflake growth decelerates below 20 percent while Databricks SQL keeps compounding, the substitution story gets real.

Second, pricing moves. Watch for Databricks SQL serverless list-price cuts or Snowflake credit-pricing changes in the next two quarters. Doubling revenue while cutting price would be the strongest possible signal of workload capture.

Third, the Iceberg question. If Databricks continues to treat Iceberg as a first-class citizen alongside Delta, and Snowflake continues to expand external-catalog support, the storage layer becomes genuinely portable and the competition moves entirely to compute and governance. That is the scenario where a doubling today becomes a tripling next year.

My prediction, testable within four quarters: if the doubling is real and sustained, Databricks SQL will be disclosed as a named revenue segment in the next Databricks S-1 or funding round, and the number will be above one billion dollars on a run-rate basis. If it stays hidden inside "platform revenue", the doubling was probably off a small base and the strategic conclusion should be scaled down accordingly.

Key Takeaways

• Databricks reports its Snowflake-competing product doubled in sales, per a June 2026 Bloomberg headline. The full article is not accessible, so the base number, margin, and workload mix are all unknown.
• A 100 percent growth rate is fast but not extraordinary in this category. The strategic weight of the number depends entirely on the base it doubled from, which the source does not disclose.
• The warehouse-versus-lakehouse functional gap is closing to seconds of latency and pennies per query. The real vendor lock-in question is now the catalog and governance layer, not the storage engine.
• For iGaming, ad-tech, and fintech teams, the practical impact is more use in vendor negotiations and a stronger case for keeping storage in open table formats.
• Watch Snowflake's next product-revenue growth rate. If it holds above 25 percent, both vendors are growing the pie. Below 20 percent while Databricks compounds, and the substitution thesis is confirmed.