06-reference/research

databricks q1 fy2027 disclosure timeline

2026-08-20·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)
databricksdisclosure-retreatic-memopre-s1gross-margin

There Is No Databricks Q1 FY2027 Release, and the Retreat Is Bigger Than a Two-Quarter Slide

The question

Verbatim: "What did Databricks' Q1 FY2027 release (reported ~May-June 2026) actually disclose — is that where the 'adjusted free cash flow' qualifier first appeared and NRR was dropped, making the disclosure retreat a two-quarter slide rather than a single-release event?"

Context: this is the single missing document behind the central claim in [[2026-08-14-databricks-pre-s1-ic-memo-thesis]], whose sub-agent could not locate the release. The backlog entry framed the stakes precisely: a two-quarter slide is a publishable pattern, a one-release change is a footnote.

Answer up front: the document does not exist. Databricks has never issued a Q1 press release in this series. But the underlying hypothesis survives in a stronger form than the one proposed, because the retreat runs four metrics deep and roughly twenty months long, and its intermediate step is a June 2026 analyst briefing rather than a press release.

What we already know (from the vault)

What the web says

Fiscal calendar, settled from a primary source. The Dec 17 2024 Series J release states the quarter-end dates in the company's own words: "Growing over 60% year-over-year in the third quarter ended October 31, 2024" and "be free cash flow positive in the fourth quarter ending January 31, 2025." Databricks' fiscal year ends January 31. Therefore Q1 FY2027 = February 1 to April 30, 2026, which on the company's own historic lag would have been announced late May or June 2026 (databricks.com, Dec 17 2024).

No Q1 release exists, verified three independent ways. (1) The complete press-release sitemap (press-releases-assets/sitemap/sitemap-0.xml, 598 URLs, 311 English) contains exactly four revenue-run-rate releases: Sept 8 2025, Dec 16 2025, Feb 9 2026, Aug 13 2026. Nothing between February and August 2026. (2) The only Databricks release in that window carrying a financial headline is the July 16 2026 term-sheet announcement, and it contains zero metrics: no run-rate, no growth rate, no cash-flow language, no retention (databricks.com, Jul 16 2026). (3) CNBC's June 2026 coverage compares the current figure directly to "$5.4 billion in the fiscal fourth quarter," skipping Q1 entirely.

The reason no Q1 release exists: Databricks discloses on a financing cadence, not a quarterly one. All four run-rate releases are attached to funding events (Series K close, Series L raise, Series L final close, Coatue round close). No financing was announced in Feb-Apr 2026, so no metrics were published.

The real intermediate disclosure event is June 16, 2026 — the Data + AI Summit analyst briefing. Databricks "told analysts at the conference that annualized revenue jumped over 80% from a year earlier and now sits at $6.9 billion." Critically, the artificial-intelligence-products line was still being disclosed: "His company now gets $1.7 billion in annual revenue from AI products, up from $1.4 billion in February." And on margin: "Ghodsi declined to provide Databricks' current gross margin, but he said it will go lower," attributing it to "the consumption-based business model, agentic AI coming... The agents are generating way more queries" (CNBC, Jordan Novet, Jun 16 2026). No net retention rate and no cash-flow language appear anywhere in that coverage.

A fourth dropped metric the parent brief missed: gross margin, dark since December 2024. Primary verification: Sept 14 2023 Series I — "Achieved record Non-GAAP subscription gross margins of 85%." Dec 17 2024 Series J — "Continuing to achieve non-GAAP subscription gross margins above 80%." Every release since is silent. A secondary source reaches the same conclusion independently: "That's the last time we can find any reporting from Databricks regarding its gross margins" (Cautious Optimism, undated, accessed 2026-08-20).

Verbatim profitability and retention language, full series (all fetched from databricks.com in this run):

Date Quarter Run-rate Cash-flow language (verbatim) Net retention (verbatim) AI products Gross margin
2023-09-14 Q2 FY2024 — — — — "record Non-GAAP subscription gross margins of 85%"
2024-12-17 Q3 FY2025 ~$3B expected "expected to achieve positive free cash flow" — — "above 80%"
2025-06-11 DAIS briefing $3.7B by July "close to being free cash flow positive" ">140%, unchanged from last year" (CNBC) — —
2025-09-08 Q2 FY2026 $4B "Achieving positive free cash flow over the last 12 months." "Net retention rate sustaining >140%." ">$1B" absent
2025-12-16 Q3 FY2026 $4.8B "Continuing to achieve positive free cash flow over the last 12 months." "Net retention rate sustaining >140%." ">$1B" absent
2026-02-09 Q4 FY2026 $5.4B "Delivering positive free cash flow over the last 12 months." "Sustaining net retention rate >140%." "$1.4B" absent
Q1 FY2027 Feb-Apr 2026 NO RELEASE — — — —
2026-06-16 mid-Q2 FY2027 $6.9B not reported not reported "$1.7 billion" declined; "it will go lower"
2026-07-16 — none none none none none
2026-08-13 Q2 FY2027 $7B "Continuing to deliver positive adjusted free cash flow over the last 12 months" absent absent absent

One trap worth recording. A web search for "Databricks Q1 FY2027" surfaces a "Q1 FY2027 Earnings Presentation, May 28, 2026" hosted on an investor-relations content-delivery network. It is SentinelOne (ticker S), not Databricks — verified by extracting the deck text, which names SentinelOne twelve times and defines its adjusted free cash flow as excluding Israeli Tax Authority assessment payments. Databricks is private and has no investor-relations site. Any future agent researching this question will hit the same false positive.

Convergences and contradictions

Synthesis for RDCO

The question asked whether the retreat was a two-quarter slide or a single-release event. The honest answer is that it is neither, and the third option is better than both. Databricks did not slide across quarters, because it has no quarterly reporting obligation and never files. What it did was withdraw four metrics, one at a time, across roughly twenty months, in an order that tracks almost exactly how damaging each number became. Gross margin went dark first, after December 2024, which is precisely when inference cost-of-goods-sold began compressing margins across consumption-priced software. Net retention held through February 2026 and then stopped appearing. The artificial-intelligence-products run-rate survived one event longer, last seen at $1.7 billion in June 2026. The free-cash-flow qualifier came last, in August 2026, which is what you would expect: you only need the word "adjusted" once the unadjusted number stops cooperating.

That reordering is the piece. The memo's original spine was "they went quiet on retention in a great quarter." The better spine is "the profitability claim was already hollow eighteen months before the retention line disappeared, and the sequence tells you which number turned first." A junior analyst reading the August 2026 release sees one qualifier and one missing bullet. An analyst who diffs seven primary documents and one conference briefing sees a company that has been serially retiring metrics since its margin story broke, and has replaced them with product-level run-rates that are individually impressive and collectively non-comparable to anything it published before.

The strongest single addition is not a missing document at all. It is Ali Ghodsi, on the record with CNBC on June 16 2026, declining to give a gross margin and volunteering that it will go lower, with the mechanism attached: consumption pricing plus agents generating far more queries. The parent brief pre-registered kill test number two as "confirms it if gross margin compresses as growth accelerates." The chief executive confirmed the direction verbally, in public, fourteen months before any S-1 could. That converts a speculative bear case into one with a named source and a date, and it does so without requiring the filing at all. It also connects cleanly to work the vault already holds on why usage-based pricing decays margin ([[2026-08-18-mostlymetrics-usage-based-pricing-one-way-door]]) and on where inference costs land on the profit-and-loss statement ([[2026-07-09-mostly-metrics-llm-costs-pl]]), which means the piece can explain the mechanism rather than just observing the silence.

Two pieces of arithmetic sharpen the close, both explicitly Ray's own and not reported figures. First, artificial-intelligence products were $1.4 billion against a $5.4 billion run-rate in February (25.9% of mix) and $1.7 billion against $6.9 billion in June (24.6%). AI revenue was growing more slowly than the company and shrinking as a share of it, which is a sufficient reason to stop printing it. Second, the run-rate moved from $6.9 billion on June 16 to $7 billion on August 13, about $100 million across eight weeks, with an identical ">80%" growth rate quoted both times. The August headline was substantially pre-announced in June. Neither observation requires a filing, and both are checkable by any reader against the same two public sources.

Why this is in the vault

It closes the one open gap in [[2026-08-14-databricks-pre-s1-ic-memo-thesis]] with a definitive negative finding plus a fourth retreat metric, which means the Sanity Check Databricks IC memo can now be drafted rather than blocked: the disclosure-retreat argument is upgraded from three metrics over six months to four metrics over twenty months, anchored by a dated CEO quote on gross margin, and the "two-quarter slide" framing that the whole downstream IC-memo series was waiting on can be retired instead of chased.

Open follow-ups

Related

Sources

Vault:

Web — primary, fetched and read verbatim in this run (company-stated run-rate figures, not audited):

Web — reporting:

Explicitly rejected as evidence:

Search paths exhausted without finding a Q1 FY2027 release: full press-release sitemap enumeration (598 URLs), Gatsby page-data endpoints for the 2026 archive, Wayback Machine snapshots of the newsroom index (May 16 and July 31 2026 captures, both client-rendered and uninformative), Wayback CDX enumeration, three web searches, and direct date-stamping of eleven candidate 2026 releases.

Ray's own arithmetic (estimates, not reported figures): AI-products share of run-rate 25.9% (Feb) to 24.6% (June); run-rate delta of ~$100M between June 16 and Aug 13 2026.