Snowflake Is ~$114.8B at ~20.8x, Both Disputed Figures Were Right, and the Premium Argument Survives With a Basis Correction
The question
Verbatim: "What is Snowflake's current market cap and EV/revenue multiple, and does the corrected figure hold up the Databricks IC memo's ~33x-vs-21x premium argument?"
Context: [[2026-08-20-databricks-q1-fy2027-disclosure-timeline]] flagged a live contradiction inside our own published work. CNBC put Snowflake at about $83B on 2026-06-16; [[2026-08-14-databricks-pre-s1-ic-memo-thesis]] carried about $116B eight weeks later and built its central valuation claim on the resulting ~21x multiple. Snowflake is the memo's only truly comparable public company, so the premium calculation moves on which figure is right.
Answer up front: both figures are correct at their own dates. Neither is an error. The $33B gap is a +38% price move, and the memo's premium argument holds — but the stated magnitude needs a basis correction, not a price correction.
What we already know (from the vault)
- The IC memo's exact assertion, quoted: "As of Aug 2026 it carries a ~$116 billion market cap at roughly 21.3x EV/revenue" (sourced to GuruFocus, not fetched directly), feeding claim 4: "$190B ÷ $7B exit run-rate = ~27x… Interpolating the disclosed run-rates gives estimated LTM revenue of roughly $5.8B… which puts Databricks near ~33x LTM revenue against Snowflake's ~21x." The bear-case paragraph converts that to "an estimated ~55% premium to the only truly comparable public company" ([[2026-08-14-databricks-pre-s1-ic-memo-thesis]]).
- The derivative brief's exact flag: "CNBC put Snowflake's market capitalisation at about $83 billion on June 16 2026; the parent brief carried roughly $116 billion for August 2026 via GuruFocus. A 40% move in eight weeks is possible but large enough that the EV/revenue comparison underpinning the '33x versus 21x' argument should be re-derived before publication" ([[2026-08-20-databricks-q1-fy2027-disclosure-timeline]]).
- The vault already did most of this arithmetic independently a month ago and got a compatible answer: EV/TTM revenue of ~21.1x at a $116.5B market cap on 2026-09-29, built from SEC share counts, cash plus investments minus convertible-note book value, and TTM total revenue ([[2026-09-29-snowflake-snow-valuation-strategy]]). That note also supplies the dated price path that resolves the contradiction: month-end closes of $255.55 (May), $254.50 (June), $293.28 (July), $331.43 (August).
- That note's own Databricks comparison used a different basis again - "about 27x run-rate versus SNOW's ~18.5x on the same annualized-quarter basis (~21x TTM)" - which is the first clue that the disagreement in our files is about denominators, not about prices.
What the web says
All figures below are computed by Ray from primary data. Price is the 2026-10-23 regular-session close (the last close before this brief; 2026-10-26 is a Monday pre-market). Balance-sheet and revenue inputs are from Snowflake's Q2 FY2027 10-Q (period ended 2026-07-31, filed 2026-09-04), pulled from the SEC XBRL companyconcept API, CIK 0001640147.
Inputs, each labeled by source and as-of date:
| Input | Value | Source | As of |
|---|---|---|---|
| Share price (close) | $325.43 | Yahoo Finance chart API, pulled 2026-10-26 01:20 ET | 2026-10-23 |
| Shares outstanding | 352,800,000 | 10-Q cover page, dei:EntityCommonStockSharesOutstanding |
2026-08-21 |
| Cash and equivalents | $1,707.187M | us-gaap:CashAndCashEquivalentsAtCarryingValue |
2026-07-31 |
| Short-term investments | $637.508M | us-gaap:AvailableForSaleSecuritiesDebtSecuritiesCurrent |
2026-07-31 |
| Long-term investments | $1,984.482M | us-gaap:AvailableForSaleSecuritiesDebtSecuritiesNoncurrent |
2026-07-31 |
| Convertible notes (carrying) | $2,283.985M | us-gaap:ConvertibleDebtNoncurrent |
2026-07-31 |
| TTM total revenue | $5,434.647M | FY26 10-K $4,683.946M − H1 FY26 $2,187.043M + H1 FY27 $2,937.744M | 2025-08-01 to 2026-07-31 |
| TTM product revenue | $5,211.198M | FY26 $4,472.317M − H1 FY26 $2,087.309M + H1 FY27 $2,826.190M | 2025-08-01 to 2026-07-31 |
| FY27 product revenue guide | $6,070M (+36%) | Q2 FY27 8-K Ex. 99.1, 2026-09-02 | FY ending 2027-01-31 |
The arithmetic:
- Market cap = 352.8M × $325.43 = $114,811.7M ≈ $114.81B
- Enterprise value = $114,811.7M + $2,283.985M (converts) − $1,707.187M (cash) − $637.508M (ST inv) − $1,984.482M (LT inv) = $112,766.5M ≈ $112.77B
- EV / TTM total revenue = 112,766.5 ÷ 5,434.647 = 20.75x
- EV / TTM product revenue = 112,766.5 ÷ 5,211.198 = 21.64x
- EV / FY27 guided product revenue = 112,766.5 ÷ 6,070 = 18.58x
- EV / annualized Q2 FY27 product revenue ($1,491.861M × 4 = $5,967.4M) = 18.90x
Independent cross-check. stockanalysis.com's SNOW statistics page, fetched this run, reports price $325.43 (Oct 23 2026 close), market cap $114.70B, enterprise value $113.13B, shares 352.46M, revenue TTM $5.43B, EV/Sales 20.82. Every line agrees with the computation above to within 0.3%. The small EV gap is a convention difference (they likely carry operating-lease liabilities or a marginally different cash definition).
Databricks side, unchanged since the memo. A WebSearch for any post-August valuation or revenue event returned nothing later than the 2026-08-13 release: $5B closed at a $190B post-money, $7B exit run-rate, >80% YoY. No new round, no S-1, no updated run-rate as of 2026-10-26. The memo's Databricks inputs are still the current ones.
Convergences and contradictions
- The contradiction dissolves: both figures are right, and the gap is price. At the 2026-06-16 close of $238.32, against the then-latest primary share count of 346.6M (Q1 FY27 10-Q cover, as of 2026-05-15), market cap = $82.6B. CNBC's "
$83B" is accurate to within 0.5%. At the 2026-08-14 close of $328.92, market cap = $114.0B on 346.6M shares, or $116.0B on the 352.8M count later filed. The memo's "$116B" is accurate. The intervening move is +38.0% in eight weeks, consistent with the dated price path already in [[2026-09-29-snowflake-snow-valuation-strategy]] (Investor Day 2026-06-04, two +9.6% sessions in June, then a July-August climb). Neither figure needs correcting. Only their missing as-of dates do. - The memo's own 21.3x was mildly understated at its own date, which cuts against the memo. Recomputing EV/TTM total revenue as of 2026-08-14, using the then-latest Q1 FY27 balance sheet (2026-04-30) and TTM revenue of $5,032.8M, gives 22.64x, not 21.3x. A higher Snowflake multiple means a smaller premium. GuruFocus was reading low.
- Snowflake's multiple has compressed slightly since the memo, not expanded. 22.6x on 2026-08-14 → 20.75x today. Price is roughly flat ($328.92 → $325.43) while the Q2 FY27 print on 2026-09-02 lifted TTM revenue from $5.03B to $5.43B. Revenue outran price. The memo's "~21x" happens to be the right number today, for reasons that have nothing to do with why it was written.
- The real defect is a basis mismatch the memo never discloses, and it runs in one direction. The memo compares Databricks' equity value ($190B, no cash deducted) against Snowflake's enterprise value (~$2.05B of net cash deducted), and Databricks' all-in revenue against Snowflake's total revenue including ~4% professional services. Both asymmetries inflate the stated premium.
Synthesis for RDCO
The headline answer: the premium argument holds, the magnitude is right to within a few points on the memo's own basis, and the number that should actually be published is a band, not a point. Recomputed with today's figures, Databricks sits at $190B ÷ ~$5.84B estimated LTM revenue = 32.5x, against Snowflake at 20.75x EV/TTM revenue. That is a +57% premium, against the memo's stated "~55%." On the memo's own chosen basis the claim survives essentially untouched, which is the cleanest possible outcome for a published number under audit. The two-month-old "~33x vs ~21x" line can stand as written provided it carries an as-of date.
But the memo should not publish a point estimate, because three defensible bases give three different answers and the spread is wide enough to matter. Basis one, the memo's own, equity value against enterprise value on LTM total revenue: 32.5x vs 20.75x, +57%. Basis two, correcting the asymmetry by deducting Databricks' net cash: the company has not disclosed its balance sheet, but a $5B raise closed in August on top of years of prior rounds makes material net cash near-certain. At an assumed and explicitly unverified $8B net cash, Databricks is $182B ÷ $5.84B = 31.2x, giving +50%. Basis three, consistent run-rate against run-rate: Databricks at $190B ÷ $7B = 27.1x against Snowflake's EV ÷ annualized Q2 FY27 product revenue of 18.90x, giving +44%. Every basis says the same qualitative thing - Databricks carries a large premium to the only true public comp - and no basis produces a number below +40%. The honest published form is "a 44% to 57% premium depending on basis, with the memo's preferred LTM-to-LTM construction at the top of that range." That is a stronger sentence than "~55%," because it survives the first analyst who recomputes it.
One correction the memo must carry, one it should resist. It must disclose the equity-value-versus-enterprise-value asymmetry, in a clause, in the bear-case paragraph. Writing "priced at an estimated ~55% premium" while silently comparing a cash-inclusive private mark to a cash-adjusted public one is exactly the kind of unforced error that a piece built on diffing other people's disclosures cannot afford. The credibility of the whole memo rests on being the person who reads the footnotes; the footnote has to apply to our own arithmetic too. What it should resist is retreating from the claim. The premium is real, it is large, and the direction of every correction found here is a few points down, never a reversal.
A dating discipline follows from this, and it generalizes past this memo. The entire eight-week "contradiction" existed only because two correct numbers were written down without their as-of dates. Snowflake moved 38% in eight weeks and 173% off its 2026-04-10 low; any market-cap or multiple figure in an RDCO artifact is a perishable good. Every such figure should be written as "$114.8B (2026-10-23 close)" rather than "$114.8B," and the IC memo needs an explicit refresh trigger: Snowflake's Q3 FY27 print, expected ~2026-12-02, which will lift TTM revenue to roughly $5.77B and - at an unchanged price - compress the multiple to ~19.5x, mechanically widening the published premium by about four points. Shipping the memo after that date with the August numbers would be a live error rather than a stale one.
Why this is in the vault
It resolves the open comp-figure question that [[2026-08-20-databricks-q1-fy2027-disclosure-timeline]] left blocking publication of the Q1-2027 Databricks IC memo in the Sanity Check lead-magnet series, and it supplies the exact amendment the memo needs: keep the premium claim, replace the point estimate with a 44-57% band, disclose the equity-value-versus-enterprise-value basis, date every figure, and set a 2026-12-02 refresh trigger.
Open follow-ups
- What is Databricks' actual net cash position? It is the single input that would turn the +50% enterprise-value-adjusted premium from an assumption into a number. No public disclosure exists; cumulative-raise estimates plus reported burn from secondary research providers are the only available path.
- Where do Databricks shares trade on secondary venues (Forge, EquityZen, Caplight) relative to the $190B primary mark as of October 2026? A persistent discount would give an independent read on whether the premium is market-clearing or negotiated, and it is the one Databricks valuation input that updates between funding rounds.
- Does the premium hold against a broader consumption-priced comp set rather than Snowflake alone? Snowflake is the closest comp but it is a single observation, and a two-or-three-name set would test whether 44-57% is a Databricks premium or a private-markets premium.
Related
- [[2026-08-14-databricks-pre-s1-ic-memo-thesis]]
- [[2026-08-20-databricks-q1-fy2027-disclosure-timeline]]
- [[2026-09-29-snowflake-snow-valuation-strategy]]
- [[2026-06-12-2026-2027-ipo-pipeline-lead-magnet]]
- [[2026-06-21-mostly-metrics-ndr-benchmarks]]
- [[2026-09-12-pupius-four-axis-snowflake-data-ai-strategy]]
Sources
Vault (read this run):
~/rdco-vault/06-reference/research/2026-08-14-databricks-pre-s1-ic-memo-thesis.md- full read~/rdco-vault/04-finance/2026-09-29-snowflake-snow-valuation-strategy.md- full read~/rdco-vault/06-reference/research/2026-08-20-databricks-q1-fy2027-disclosure-timeline.md- targeted read (the two lines asserting the $83B/$116B conflict); full note not read this run
Primary - SEC EDGAR, pulled 2026-10-26 (CIK 0001640147):
- XBRL
companyconceptAPI, tagsdei:EntityCommonStockSharesOutstanding,us-gaap:RevenueFromContractWithCustomerExcludingAssessedTax,us-gaap:CashAndCashEquivalentsAtCarryingValue,us-gaap:AvailableForSaleSecuritiesDebtSecuritiesCurrent,us-gaap:AvailableForSaleSecuritiesDebtSecuritiesNoncurrent,us-gaap:ConvertibleDebtNoncurrent- https://data.sec.gov/api/xbrl/companyconcept/CIK0001640147/ - Q2 FY2027 earnings release (8-K Ex. 99.1, 2026-09-02) - https://www.sec.gov/Archives/edgar/data/1640147/000164014726000033/fy2027q2earnings.htm
- Q4 FY2026 earnings release (8-K Ex. 99.1, 2026-02-25) - https://www.sec.gov/Archives/edgar/data/1640147/000162828026011631/fy2026q4earnings.htm
Market data:
- Yahoo Finance chart API, SNOW daily closes, pulled 2026-10-26 01:20 ET. Closes used: 2026-06-16 $238.32, 2026-08-13 $337.38, 2026-08-14 $328.92, 2026-09-29 $330.31, 2026-10-23 $325.43.
- stockanalysis.com SNOW statistics, fetched this run (independent cross-check only) - https://stockanalysis.com/stocks/snow/statistics/
Databricks (no new primary data since the parent brief):
- Databricks newsroom, 2026-08-13 - https://www.databricks.com/company/newsroom/press-releases/databricks-grows-80-yoy-surpasses-7b-revenue-run-rate-scales
- WebSearch for any post-August 2026 Databricks valuation or run-rate event returned nothing later than 2026-08-13.
Not verified / flagged:
- The CNBC article itself was not fetched this run. Its ~$83B figure is corroborated arithmetically ($82.6B at the 2026-06-16 close on the then-filed share count), not by reading the source.
- Databricks' net cash is unverified. The $8B figure used in the basis-two sensitivity is an assumption, labeled as such, not a sourced number.
- Databricks' ~$5.84B LTM revenue remains Ray's interpolation from three disclosed exit run-rates, carried forward unchanged from [[2026-08-14-databricks-pre-s1-ic-memo-thesis]]. It is an estimate, not a reported figure.
- Share count of 352.8M is as of 2026-08-21 and predates roughly two months of stock-based-compensation vesting; true current shares are likely ~1% higher, which would raise market cap by ~$1.1B and the multiple by ~0.2x.