Databricks Overtakes Snowflake in Scale: $6.9B vs $5.5B Run Rate

Databricks is on track to clear a $6.9 billion annualized revenue run rate in the first half of fiscal 2027, versus Jefferies' estimate of roughly $5.5 billion for Snowflake. That's a $1.4 billion gap at the top line, and it marks the first time Databricks is set to overtake Snowflake in scale. The more interesting number sits underneath: Databricks is compounding at 65% year-over-year in its core, 80% including LLM monetization, while Snowflake grows 32%. Two companies, one market, growth rates that differ by roughly 2x.

The Numbers

Start with the run rates, because scale determines negotiating use with hyperscalers, hardware allocation, and reference-customer gravity. As Proactive financial news reported, Jefferies pegs Databricks on track to exceed $6.9 billion in run rate versus Snowflake's approximate $5.5 billion, with Databricks' core growing 65% and LLM-inclusive growth around 80%. Snowflake grew 32% at the run-rate level and reported 34% product revenue growth in Q1 fiscal 2027, an acceleration of roughly four percentage points over the prior quarter.

The acceleration matters. Snowflake bulls have been arguing for six quarters that consumption growth would re-accelerate as AI workloads ramp on the platform. Two consecutive quarters of acceleration is the first hard evidence that thesis has legs. But the base rate comparison is unforgiving: even if Snowflake holds 34% product growth for a full year, Databricks growing at 65% closes the ratio faster than Snowflake can widen it.

The margin picture inverts the growth picture. Snowflake generates roughly 23% free cash flow margins. Databricks is described as near breakeven. On a rule-of-40 basis, Snowflake is running at approximately 55 to 57 (32% growth plus 23% FCF margin), which is elite for a company at $5.5B run rate. Databricks, if we assume near-zero FCF, is running at 65 to 80 on growth alone. Both are healthy, but they're healthy for different reasons and appeal to different capital pools.

One number the source does not disclose, which matters more than most people realize: the gross margin split between compute and storage for either company, and how much of Databricks' 80% growth is GPU-passthrough revenue at low take rates versus platform-level Delta Lake and SQL Warehouse consumption. Without that breakdown, comparing top-line growth apples-to-apples is directionally useful but not conclusive. If half of Databricks' incremental growth is GPU resale at hyperscaler margins, the profit trajectory looks very different from what the run rate implies.

Snowflake shares closed near $235, up about 7% year-to-date. That's underperformance relative to what a 34% growth reacceleration would normally command, which tells you the market is already pricing in the Databricks scale crossover.

What's Actually New

The narrative for the last three years has been "Snowflake owns the warehouse, Databricks owns the lakehouse, they'll meet in the middle." That narrative is now obsolete. Databricks' SQL Warehouse product has crossed a $1.5 billion annualized run rate, meaning Databricks is no longer a Spark shop that sells BI as an afterthought. It's a $1.5B warehouse business inside a $6.9B company, growing into Snowflake's core.

Jefferies notes Snowflake still runs a materially larger data warehousing business, which is true. But the delta is closing on both sides. Snowflake has narrowed the technical gap in AI capabilities significantly over the past year, with Snowflake shipping AI offerings branded CoCo and Snowflake CoWork. Databricks has narrowed the warehousing gap with SQL Warehouse. Feature parity is now a two-way street.

What's genuinely new is the second-order competition: business-user AI. Databricks' Genie platform integrates with Microsoft Teams, Slack, and Google Drive, letting non-technical employees query enterprise data through the tools they already open every morning. This is not the same fight as "which engine runs my nightly ETL faster". It's a distribution fight for who owns the analytical interface that a product manager or a finance analyst actually touches. Snowflake's CoCo and CoWork appear to be aimed at the same layer, though the source does not disclose their integration surface or adoption numbers, which is the metric that would actually settle the question.

The other thing that's new: Snowflake accelerated. For most of 2024 and 2025 the consensus was that Snowflake was structurally decelerating into the low 20s. Two quarters of reacceleration, capped by 34% product growth, invalidates the "melting ice cube" thesis. It doesn't invalidate the "getting outgrown" thesis, but those are different claims.

What's Priced In for Data Teams

For platform leads evaluating stacks in the second half of 2026, most of what's in this Jefferies note is already assumed. The engineering community priced in Databricks' AI leadership eighteen months ago. Nobody choosing between the two today is surprised that Databricks trains models faster or that its lakehouse story is cleaner for ML workloads. What's less priced in is the reverse flow: Snowflake being competitive on AI-driven consumption growth. If Snowflake's product revenue keeps accelerating from AI workloads on the existing warehouse footprint, that changes the calculus for teams who chose Snowflake for BI and were worried they'd need a second platform for ML.

What's also not priced in for most engineering teams: the SQL Warehouse traction. A $1.5B run rate on Databricks SQL means there is now a credible reference base for teams considering warehouse consolidation onto the lakehouse. Two years ago that migration was a risk-adjusted no for most CFOs. At $1.5B and growing, the reference calls exist. Expect more RFPs where Databricks is the incumbent lakehouse bidding to displace Snowflake as the warehouse, rather than the other way around.

The margin gap is priced in on the equity side but not on the procurement side. Snowflake's 23% FCF margin is what funds the aggressive credit terms, the enterprise discounting, and the AI investment. Databricks running near breakeven means every dollar of discounting comes out of growth capital. For a CTO negotiating a three-year commit in Q4, that's use worth remembering.

Contrarian View

The consensus read is: Databricks wins, Snowflake settles into a profitable but slower-growing niche. I'd push back on that. The 34% product growth reacceleration is not a rounding error. It's roughly four points of quarterly improvement, sustained for two quarters, at a $5.5B run rate. Reaccelerations at that scale are rare, and when they happen they usually run for four to six quarters before flattening. If Snowflake sustains this into fiscal 2028, the scale gap stops widening in ratio terms even if it keeps widening in absolute dollars.

The other contrarian angle: profitability optionality. Databricks near breakeven at $6.9B means the moment growth decelerates from 65% to 40%, the company faces a margin ramp that Snowflake has already completed. Snowflake did the painful part in 2023 and 2024. Databricks hasn't. Whether that shows up in an IPO S-1 or in private-round pricing, it's a compounding disadvantage the growth numbers don't capture. The unknown, and this matters, is what Databricks' gross margins look like ex-GPU-passthrough. If underlying platform gross margin is 75%+, the profit story is fine. If it's compressed by resale economics, the transition to Snowflake-like FCF margins will be painful.

Key Takeaways

• Scale crossover is here: Databricks' $6.9B run rate versus Snowflake's $5.5B marks the first time Databricks leads on top line, but Snowflake's 23% FCF margin versus Databricks' near-breakeven is the inverse story most coverage underweights.
• Snowflake reaccelerated: Four points of product revenue acceleration to 34% in Q1 fiscal 2027 breaks the "structurally decelerating" narrative. Two more quarters at this rate and the ratio gap stops widening.
• Warehouse competition is now two-way: Databricks SQL Warehouse at $1.5B run rate is a credible reference base for warehouse consolidation onto the lakehouse, which was not true 24 months ago.
• The business-user AI layer is the next fight: Genie's integration with Teams, Slack, and Drive versus Snowflake's CoCo and CoWork will decide who owns the analytical interface, though adoption metrics for both remain undisclosed.
• Testable prediction: If Snowflake sustains 32%+ run-rate growth through fiscal 2028, expect the equity to re-rate above the 7% YTD gain currently reflected at $235. If growth slips back below 28%, the scale gap widens fast enough that the reacceleration story dies within two quarters.