Phase Space AI

Financial Model Notes

Snowflake [SNOW]

Snowflake [SNOW] — Financial Model Notes

As of 2026-07-29. Framework v1.5.1. Tier-2 memo, cluster member (data & dev infrastructure). Cluster analysis: reports/clusters/Data_Dev_Infrastructure_Cluster_2026-07-29.md

This memo issues no position verdict. The book decides.

1. Provenance — every figure in this memo traces here
Primary source SEC EDGAR — XBRL companyfacts and the primary 10-Q document, read directly
CIK 1640147
Latest 10-Q period ending 2026-04-30, filed 2026-05-29, accession 000164014726000030
Filing age at analysis date 61 days — recency asserted per valuation.md; no GOOGL-class staleness
Prices Alpaca /v2/stocks/snapshots, 2026-07-29 session close
Multiple history Alpaca monthly bars × quarterly weighted-average share count ÷ TTM revenue, with a 60-day filing lag applied so no figure is used before it was public
Reverse DCF ~/.claude/skills/investment-memo/assets/reverse_dcf.py
Options chain Alpaca /v2/options/contracts + /v1beta1/options/snapshots, Jan-2027 expiry, pulled 2026-07-29
Consensus NOT OBTAINED — Alpha Vantage 25/day quota shared and exhausted. Consensus Criteria INDETERMINATE; blocks nothing.

TTM revenue is built by summing the last four quarterly XBRL periods, never taken from the last fiscal year. The framework's MU (understated 136%) and SNDK (76%) failures both came from using last-FY revenue.

Where a fiscal Q4 is not separately tagged, it is derived as the annual figure minus the three tagged quarters. That derivation is applied to SNOW's Q4 and is arithmetically exact, not estimated.

2. The quarterly series this model is built on

All figures in $000s, from XBRL, verified against the 10-Q. Fiscal year ends 31 January.

Quarter ended Revenue YoY Gross margin GAAP op margin SBC / revenue
2025-04-30 1,042,074 +25.7% 66.5% -42.92% 36.4%
2025-07-31 1,144,969 +31.8% 67.5% -29.72% 35.3%
2025-10-31 1,212,909 +28.7% 67.8% -27.16% 34.0%
2026-01-31 1,283,994 +30.1% 66.8% -24.78% 31.4%
2026-04-30 1,390,951 +33.5% 66.6% -23.45% 28.9%

TTM revenue: $5,032.8m. TTM GAAP operating margin: -26.11%. TTM SBC: $1,622.6m = 32.2% of revenue.

Note the operating-margin column against the screen's figure. The screen computed operating margin on the last fiscal year, not the trailing twelve months, on every name in this cluster. That is one quarter stale on four of them and nine months stale on TEAM.

3. Balance sheet as used

@2026-04-30
Shares outstanding (period end, verified) 346.601m
Spot (2026-07-29) $283.04
Market capitalisation $98,102m
Net cash (financial-debt basis) $2,105.6m
Enterprise value $95,996m
EV / TTM Sales 19.07x

Net cash convention, applied identically across all five cluster names: cash and equivalents, plus all marketable securities whether current or non-current, plus restricted cash, less all financial debt including convertibles. Operating and finance lease liabilities are EXCLUDED as operating rather than financing obligations. Strategic non-marketable equity investments are excluded — they are neither cash nor marketable. Figures including leases as debt are given in the research document so either convention can be reconstructed.

Cross-check performed and passed: net income ÷ shares ≈ filed EPS. See §1 of SNOW_Research.md. This is the control against the 4.1x and 54% scale errors in the framework's record.

4. Model judgements, stated

The one line in this model that requires judgement: the $2,281,903k of convertible notes. They accrete ~$2.1m per quarter, they are not due before 2027–2029, and at $283.04 they are economically equity rather than debt. They are charged in full against net cash here — the conservative treatment, and the opposite of the screen, which omitted them entirely. A model that treated them as equity would show net cash of $4,387m and an EV/Sales of 19.1x → 18.6x. The difference is not decision-relevant at this multiple, which is why the conservative treatment is used without argument.

The second line that requires judgement: the prior-year comparison base is depressed by one-offs. Q1 FY26 carried a $106.5m asset impairment on office facility exits and a G&A spike to $209.6m (against $128.7m this year and ~$93m the year before). The −26.11% TTM operating margin and the +4.5pp four-quarter improvement are therefore both flattered by the base. The three-year trend (−40.2% → −30.6% → −26.1%) is the reliable read.

OtherLongTermInvestments of $353,531k is EXCLUDED from net cash. It is strategic equity in non-marketable securities — the note discloses $25.0m of it was in Observe, a company Snowflake also sold to and then acquired. It is not cash and it is not marketable. Whether any of the remaining $328m of holdings are also Snowflake customers is not disclosed.

5. What is NOT modelled, and why

Item Treatment Reason
AI revenue not modelled — zero credited No AI revenue, AI ARR or AI contribution figure is disclosed by any of the five cluster names. Per the brief, unquantified AI upside is not credited.
Consensus estimates absent Alpha Vantage quota exhausted. Consensus Criteria INDETERMINATE. Blocks nothing.
Transcript mention counts substituted Filing mention-frequency used instead (research document §6), from the primary 10-Q/10-K documents. Labelled as a substitution, not presented as transcript data.
Terminal margin as an input inverted The screen's universe-wide 10.7% is not used. Instead the reverse DCF solves for the required terminal margin and it is compared to the industry's observed ceiling. See SNOW_Valuation.md §2.
Interim free cash flows not built Terminal value is effectively 100% of EV at a negative GAAP operating margin, far above the 60% threshold. An elaborate interim build would add false precision without changing the answer, which is precisely the reverse DCF's design rationale.
Historical EV/Sales approximated as P/S Historical net cash is not reconstructible at monthly frequency across all five names. P/S is used for the percentile work and labelled as such. Net cash is 0.6%–23.6% of market cap depending on the name, so the approximation is tightest for TEAM and SNOW and loosest for GTLB — flagged in GTLB's valuation document.

6. Known limitations of this model

  1. The 12-month NTM revenue figure is a house estimate, built off the quarterly YoY trend, not a consensus number. It is labelled as such everywhere it appears. A consensus figure would be a better base and was not obtainable.
  2. The exit multiple is the highest-variance parameter and it is not observable. That is why the reverse DCF is the primary instrument and why the sensitivity is run over the exit multiple rather than over scenario probabilities. On two of the five cluster names the verdict changes inside the plausible range, and that is declared rather than resolved.
  3. The own-history multiple percentile spans a regime change. The 2021 bubble and 2022 de-rating are excluded by using post-2023 windows; the full-history figures are reported for completeness but not used as anchors. On GTLB and TEAM even the post-2024 window is a monotone de-rating, so the anchor is declared UNIDENTIFIED rather than defaulted to a median.
  4. Options quotes are a single snapshot taken on 2026-07-29 and will have moved. The open-interest and quoted-size figures are what matter for the fillability judgement and those move more slowly than the prices.