---
ticker: "SNOW"
company: "Snowflake Inc."
date: "2026-10-23"
view: "Avoid"
fair_value: 95.22
price: 320.28
price_source: "last regular-session trade at 13:07:37 ET; previous official close $313.96 (2026-10-22)"
price_time: "2026-10-23T17:07:37Z"
price_vs_fair_value: "+236.4%"
bear: 20.33
bull: 199.16
exit_price: 163.17
p_price_justified: "under 20%"
restated: "2026-10-26"
method: "case-spread-2026-10-24"
original_fair_value: 95
original_view: "Avoid"
page: "https://alpha.dming.si/research/2026-10-23-snow"
---

# SNOW · Snowflake Inc. · Avoid

Research report as of 2026-10-23, partly unblinded. Page: https://alpha.dming.si/research/2026-10-23-snow

We value SNOW at $95.22 a share. The report's $320.28 price (Oct 23, 2026) was 236% above fair value: Avoid.

_Restated Oct 26, 2026 (method case-spread-2026-10-24); the report's original value was $95, Avoid._

| Lens | Range | Marks |
| --- | --- | --- |
| Our report | $20.33 to $199 | fair $95.22; original $95 |
| Our simulation | $28.62 to $225 | median $85.02 |
| Street targets (37) | $250 to $525 | average $418, median $436 |
| Price | | $320.28 |

Bands run from our bear to bull (the hollow mark is the original $95), our simulation's 10th to 90th percentile and the Street's low to high. Street targets forecast future prices, not value today. Market data, captured Oct 23, 2026.

| Figure | Value |
| --- | --- |
| Fair value | $95.22 |
| Price | $320.28 (market data, Oct 23, 10:07 AM PT) |
| Five-year return | -12.0% (a year, at the report's price) |
| Bear / bull | $20.33 / $199 |
| Exit price | $163 (on Jan 31, 2032) |
| Confidence | 30% |
| Price against our scenarios | The price is justified in under 20% of our scenarios (4,000 simulated outcomes, each drawing the uncertain drivers from ranges we chose: our judgement, not a measured chance) |

_This text argues the original call. The current figures are at the top._
**Original call, Oct 23, 2026 (before the Oct 26, 2026 restatement)**

_As of 2026-10-23 · partly unblinded_

- **Fair value today:** $95 per diluted share (cost of equity 10.75%)
- **Price:** $320.28, last regular-session trade, 2026-10-23 13:07:37 ET → downside 70%
- **Five-year exit price:** $163 on 2032-01-31 (long fade $152; at 10/20 years $155/$171; outside view $174; formula undefined, growth above WACC)
- **Five-year return:** -12.0% a year from today's price to the $163 exit (no dividends)
- **Bear / bull today:** $22 / $168 from coherent operating cases
- **Confidence:** 30%, meaning: our subjective chance that the value at the exit date lands within ±20% of the central exit price (not calibrated)
- **Moat:** Data gravity and governance: customers' data, sharing links and AI agents run inside one governed platform; open table formats and Databricks narrow it
- **Market expects:** an exit near $549 in January 2032: 42x our FY2032 operating income before stock pay, 80x after it
- **Price needs:** no single driver reaches the price within its range; the best, a 40% FY2032 margin before stock pay, gives $167
- **P(price justified):** under 20% in our scenarios (subjective, not calibrated)
- **Key driver:** FY2032 operating margin before stock pay: 21% to 32% moves fair value from $58 to $121 (central 27.5%, $95)
- **Street:** average target $418 (market data, range $250-525; 37 targets in the last quarter averaging $416, up from $326 over the last year); market data: 44 Buy, 7 Hold, 0 Sell, mean $440 (2026-10-14); 12-month targets

## The call in plain words

Snowflake is growing faster than at any time since FY2024: product revenue rose 37% in the
quarter to July 2026, and management raised the year's guide to 36%. The business is worth
about $95 a share on our forecast because most of today's reported profit is paid to
employees in stock. Charged as the cost it is, Snowflake still loses money on a GAAP basis
(a 17% operating loss in the July quarter). Our central case has growth slowing from 38% to
16% by FY2032 and the operating margin after stock pay reaching 14%, about where ServiceNow
is today. That earns a five-year exit of $163, worth $95 today at a 10.75% cost of equity.

Three terms used below. The **central case** is our most likely path; the **pessimistic
case** (bear) and **optimistic case** (bull) are coherent alternative worlds, each with its
own growth, margins and exit. The **long fade** extends each case 15 years past FY2032,
fading growth and returns to maturity; the **outside view** prices FY2032 operating income
at the median ten-year EV/EBIT of three mature software peers (Oracle, Microsoft, Teradata:
20.9x). The exit is the mean of the two.

## 1. Business economics: customers pay for compute they consume

- Who pays: about 14,550 enterprises buy capacity contracts (over 80% pay in advance, CFO at
  Investor Day 2026-06-02) and draw them down as they run queries, store data and now run AI
  functions and agents. 96% of revenue is this consumption ("product revenue"); the rest is
  professional services at a loss [F, Q2 FY2027 release].
- What limits it: Snowflake rents its compute from AWS, Azure and Google Cloud, so its gross
  margin (74% non-GAAP guide for FY2027) is below that of software sold by seat, and AI
  workloads, which buy model inference, carry "lower contribution margin today" [M, CFO, Q2
  FY2027 call]. It has committed $6.0B to AWS over five years to March 2031 [F, 10-Q note 11].
  Customers can optimize spend at any time; consumption slowed when they did in 2023.
- Who funds the customers: no customer is 10% of revenue [F, 10-Q note 3]; AI-native start-ups
  are "a small part" of revenue [M, CEO]. Customer prepayments fund the business: operating
  working capital is -52% of revenue at each January year end [F, our sum from the balance
  sheets].
- What "adjusted" leaves out: non-GAAP operating income ($489.7M in FY2026) excludes stock pay
  charges of $1,710.7M (36.5% of revenue). Adjusted free cash flow (23% FY2027 guide) is higher
  still because stock pay is not cash; the shares it issues are bought back with cash ($874M
  of buybacks in FY2026, market data cash flow). We charge stock pay as a cost and hold the share
  count flat.
- AI: CoCo (Cortex Code, a coding agent) is in 9,100 accounts and CoWork (the business-user
  agent) in 5,800; management says AI products gave about half of the recent acceleration
  [M, Q2 FY2027 call]. Evidence against: the figures that would show AI revenue sticking (AI
  revenue, its margin) are not disclosed.

## 2. Recent results against the bar Snowflake set

| Quarter | Product guide midpoint | Actual | Beat | Op. margin guide | Actual |
| --- | --- | --- | --- | --- | --- |
| Q3 FY2026 | $1,127.5M | $1,158.4M | +2.7% | 9% | 11% |
| Q4 FY2026 | $1,197.5M | $1,226.6M | +2.4% | 7% | 11% |
| Q1 FY2027 | $1,264.5M | $1,334.3M | +5.5% | 9% | 11.9% |
| Q2 FY2027 | $1,417.5M | $1,491.9M | +5.2% | 12.5% | 15.3% |

Over eight quarters product revenue beat the guide midpoint every time (median +4.6%, range
+2.4% to +5.6%) and the margin by a median 3.0 points (`output.txt`). The guide is set from
"observed consumption patterns" [M], which explains a steady beat; we use the measured beat,
not an inferred intent. Growth accelerated three quarters running (29% to 37%); net revenue
retention is 126%; RPO (contracted revenue not yet recognized) is $9.0B, +30% [F].

The September 2026 financing: $3.75B of 0% convertible notes due 2029 and 2031, about $3.70B
net, of which $383.5M bought capped calls and $548.3M repurchased part of the 2027 notes [F,
8-K 2026-10-02].

## 3. Forecast and valuation (model.py, output.txt)

| FY to January | 2027 | 2028 | 2029 | 2030 | 2031 | 2032 |
| --- | --- | --- | --- | --- | --- | --- |
| Product growth (central) | 38% | 30% | 25% | 21% | 18% | 16% |
| Revenue ($M) | 6,405 | 8,266 | 10,283 | 12,400 | 14,596 | 16,898 |
| Non-GAAP operating margin | 15.8%* | 19% | 21.5% | 24% | 26% | 27.5% |
| Stock pay charges, % revenue | 28.8%* | 22% | 18% | 15.5% | 13.5% | 12.5% |
| GAAP operating income ($M) | -1,033* | -413 | 206 | 905 | 1,649 | 2,332 |

\*FY2027 combines the reported first half with our second half.

- Where we differ: FY2027 product revenue $6,183M against the $6,070M guide (the beat
  history). The central case is about breakeven on GAAP operating income in late FY2028,
  consistent with management's Q4 FY2028 target [M, Investor Day]. The 2023 target of a 25%
  non-GAAP margin by FY2029 has not been reaffirmed; we reach 21.5% then.
- Taxes: $7.3B of federal tax losses shelter most of the forecast's profits through FY2031
  in the central case; the cash saved stays in the bridge and the rest is valued at the exit
  ($0.9B) [F, 10-K tax note; model].
- Bridge at 2032-01-31: $15.1B net cash (cash kept after repaying the new notes at par when
  they mature, $2.0B in October 2029 and $1.75B in October 2031, less operating cash, plus
  $0.37B strategic investments) plus $1.0B of tax losses; 370.8M diluted
  shares (352.8M outstanding, 9.3M options, 10.0M for the 2024 notes net of capped calls, less
  1.3M repurchased). Unvested RSUs are not counted because their cost is in forecast stock pay.
- Cost of equity 10.75% = 5.29% risk-free + 1.30 beta (median Blume beta of MDB, DDOG, ORCL,
  MSFT, NOW, CRM, TDC, ESTC) x 4.20% premium (Damodaran, 2026-10-01). Snowflake's own beta
  (1.38) gives $92.
- The exit: the long fade gives $152 (17.3x FY2032 EBIT) and the outside view $174 (20.9x);
  their mean is $163. In the pessimistic case the fade ($32) is 27% below the outside view
  ($44) because its 15% mature return and 7% growth leave little value for growth; we hold the
  method's mean.
- No change from the method's default.

| Sensitivity of $95 | Value |
| --- | --- |
| FY2032 non-GAAP margin 32% / 21% | $121 / $58 |
| FY2032 stock pay 10.5% / 16.5% | $107 / $72 |
| Product growth +3 / -5 points every year | $107 / $79 |
| Growth after FY2032 16% / 8% | $101 / $88 |
| Mature return 40% / 15% | $96 / $93 |
| Exit horizon 10 / 20 years (exit price) | $155 / $171 |

Case-weighted value at 25/50/25 (our weights): $95.

## 4. What the market expects

At $320.28 (market data, 2026-10-23 13:07 ET; previous close $313.96) the price needs an exit
near $549 on 2032-01-31, growing at our 10.75% cost of equity. On our central forecast that is
42x FY2032 operating income before stock pay (40x before stock pay and amortization), 80x
after both, or 11x revenue
(street_check.txt).

The Street does not forecast a better business than ours. Its average revenue (market data, fetched
in the market pass) is $6.32B for FY2027, $8.16B for FY2028 and $10.51B for FY2029 against our
$6.41B, $8.27B and $10.28B; its adjusted operating margin is 18.3% for FY2028 and 20.4% for
FY2029 against our 19% and 21.5%. Run through our model, the Street's path is worth $67 to
$102 a share depending on how its rows are read. Its EBIT and EPS rows disagree with each
other (FY2030's margin falls while its EPS rises 32%), and FY2030-31 rest on 4 to 14 analysts:

| Step from our $95 | Value | Change |
| --- | --- | --- |
| Street revenue FY2027-31 exactly (product growth 30%, 30%, 29%, 6%) | $93 | -$2 |
| plus the Street's adjusted EBIT margins FY2027-31 | $68 | -$25 |
| or instead the margins its EPS implies (14%, 16%, 18%, 19%, 23%) | $76 | -$17 |
| Street FY2027-30 revenue and FY2028-30 margins only, our path after | $102 | +$6 |

The value gap is therefore the price put on earnings, not the earnings. Valuing FY2032
operating income before stock pay at 40x, as if the $2.1B a year of stock pay cost nothing
(amortization still charged), gives $305 today, close to the price; at the peers' 20.9x it
gives $171. Stock pay is not the whole gap: removing it at the method's own exit adds about
$79 (the auditor's figure, `audit/numbers-codex.json`); the rest is the multiple.

## 5. Our view against the market's, and what settles it

**Avoid.** The price is 3.4 times our central value and almost twice our optimistic case. In
our simulation (4,000 draws over the drivers' ranges and three weighted worlds,
odds.json) the chance of a value at or above the price is about 2%; no single
driver reaches the price within its range, and the best, a 40% FY2032 margin before stock pay,
gives $167.

Where we differ: we charge stock pay as the cost it is. Snowflake paid $1.71B of stock pay
charges in FY2026, 3.5 times its non-GAAP operating income; it spends cash on buybacks and
withholding taxes to stop the share count rising (FY2026: $874M of buybacks, market data). Our share
count is held flat on that basis. A market that values the company on earnings before stock
pay sees a business worth roughly the price; one that charges it sees our $95.

The market could be right if growth stays near 30% well past FY2032 while stock pay falls
toward 5% of revenue; our optimistic case (19% growth in FY2032, 32% margin, 10.5% stock pay)
reaches only $168. What settles it: the FY2028 guide (February 2027) and whether GAAP operating
profit arrives in Q4 FY2028 with stock pay near 20% of revenue.

The simulated median ($81) sits 15% below the central value because the bear world (25% of
draws, median $36) and the method's spread on the exit price skew the distribution down; the
optimistic world's median is $140, so the upside tail is thinner than the downside.

Our range (10th to 90th percentile $28 to $199) is 0.62 times as wide as the Street's target
range ($250 to $525) and the price lies outside it. We have not widened it toward the price:
the ranges come from Snowflake's own growth and stock pay history.

**Alpha's view (2026-10-23).** Rank 90 of 702, tier Strong (score 87.3): momentum A,
expectations A, growth B, quality C, valuation E (`alpha rank SNOW`, cache partial). Alpha's
126-day horizon rewards the acceleration and upward revisions we also see; its valuation grade
agrees with ours.

**Chart (timing, not value).** Grade C: advancing (close above a rising 30-week average,
$257.52) but the structure is damaged, below the 50-day ($331.91) after breaking the
2026-09-28 swing low; B above $340.16 with a higher low, D on a weekly close below $257.52
(`levels.py SNOW --group IGV`, bars to 2026-10-22).

## 6. Pessimistic case and why the other side could be right

The pessimistic case ($22) is a world where the AI acceleration is a migration bulge: growth
falls to 9% by FY2032, AI inference costs hold the margin before stock pay near 21%, and stock
pay stays above 16% to keep engineers. GAAP operating income is then only 3% of revenue in
FY2032. Snowflake's own history supports part of it: product growth fell from 70% to 29% in
three years after FY2023, and stock pay charges have never been below 30% of revenue in a reported half-year.

The optimistic case ($168) is the strongest argument against our central: if growth holds
near 19% in FY2032 and the margin reaches 32% with stock pay at 10.5%, the exit is $289. The
evidence for it: three quarters of acceleration, net new customers +32%, organic headcount
nearly flat while revenue grows 37%, and 828 customers above $1M (+27%). If Snowflake reaches
ServiceNow-like margins on a lower gross margin, our central is too low.

What would settle it: the operating margin after stock pay. A FY2028 in which stock pay
charges fall to about 22% of revenue while growth stays near 30% supports the central; stock
pay above 25% or growth below 25% moves toward the pessimistic case.

## 7. Next results preview (Q3 FY2027, quarter to 2026-10-31)

The date is not yet announced; Q3 was reported on 2024-11-20 and 2025-12-03. The record uses
2026-12-02 as an assumed date. From preview.py:

| Metric | Ours | Range | Reason |
| --- | --- | --- | --- |
| Revenue ($B) | 1.709 | 1.672-1.747 | product guide midpoint +4%, services $55M |
| Product revenue ($B) | 1.654 | 1.617-1.692 | the beat range of eight quarters, widened $10M |
| Non-GAAP operating margin | 18.0% | 16.5-19.5% | 15.5% guide plus the median 3.0-point beat, less 0.5 |
| Non-GAAP diluted EPS ($) | 0.73 | 0.66-0.80 | margin, interest on $7.1B cash, Q2's 25% tax, 382M shares |
| Q4 product revenue guide midpoint ($B) | 1.628 | 1.60-1.69 | our Q4 less the median beat |
| FY2027 product revenue guide ($B) | 6.108 | 6.07-6.20 | H1 plus our Q3 plus that Q4 guide |
| Net revenue retention | 126% | 124-128% | flat for four quarters |
| RPO ($B) | 10.07 | 9.4-10.7 | the Q2-to-Q3 rise of the last two years |

Each $10M a quarter of product revenue above ours, carried as a higher level, is worth about
$0.28 a share; the margin path matters far more than one quarter.

## 8. Catalysts and falsifiers

- Q3 FY2027 results (assumed 2026-12-02): the preview above.
- Q4 FY2027 results and the first FY2028 guide (about late February 2027): a product guide
  below 25% would put the central path at risk (our 30% assumes the usual beat on a guide in
  the mid-20s).
- Stock pay: FY2027 at or below 27% of revenue (management plan) and FY2028 near 22%.
- GAAP operating profit in Q4 FY2028 (management target). Missing it falsifies the central
  margin path.
- The 2027 notes mature October 2027; Snowflake can settle in cash, shares or both.

## 9. Review round (2026-10-23): two fixes moved the value from $94.17 to $95.22

Three reviews graded the draft at commit 1c10c6c05, in one round, without seeing each other's
findings:

- research-eval, run by the analyst (Claude Opus 5.5): audit/research-eval.md. No blocking finding.
- The model auditor, OpenAI's Codex (`codex exec`, high reasoning effort, writing only to `audit/`): audit/auditor-codex.md.
- The opposite-side advocate, a fresh Claude Opus agent arguing Buy from its own sources: review-advocate/advocate.md.

On a copy of 1c10c6c05, both reviewers' scripts reproduced their saved outputs exactly:
the advocate's `findings.txt` and `combined.txt`, and the auditor's `numbers-codex.json`.
Our split of the change is in audit/answer_effects.txt.

| # | Finding at 1c10c6c05 (reviewer) | Answer | Before | After | $/share |
| --- | --- | --- | --- | --- | --- |
| 1 | Amortization of acquired intangibles was charged as a cash cost, while acquisitions were also charged as cash (auditor F1, advocate F3) | **Fixed.** 81.3% of "amortization etc." is now added back to owner cash. That share is H1 FY2027's $82.8M of amortization out of $101.8M of non-stock-pay adjustments (Q2 FY2027 release). | $93.93 | $95.22 | +1.29 |
| 1a | The auditor also lowers net investment by the amortization (a further +$1.0 to +$1.7, `numbers-codex.json`) | **Disclosed, not made.** Counting the amortization of past deals as disinvestment makes the optimistic case's net reinvestment negative, and the method's return on new capital is then undefined. | | | +1.0 to +1.7 not taken |
| 2 | Cash earned interest on the new notes' principal until the exit (auditor F2) | **Fixed.** The principal is now repaid at maturity: $2.0B on 2029-10-15 and $1.75B on 2031-10-15. Starting the September proceeds' interest in July is worth -$0.02; disclosed, not modelled. | $94.17 | $93.93 | -0.24 |
| 3 | The Street run applied total revenue growth to product revenue, kept our own FY2027, and never used the Street's EPS (auditor F3, advocate F6) | **Fixed** in street_check.py. Section 4 has the new table. Other readings, from the reviewers' own scripts: the auditor's EPS reconstruction gives $63.42; the advocate's FY2028-30 EPS margins without interest give $88. The Street's rows disagree with each other, so we give a range rather than one figure. | $67.05 | $68.34 (EBIT); $76.41 (EPS) | headline 0 |
| 4 | The "before stock pay" figure also removed amortization (auditor F4) | **Fixed.** Removing stock pay alone gives 42.2x what the price needs, and $305 at 40x. | $317 at 40x | $305 at 40x | headline 0 |
| 5 | The 2024 notes are counted as 10.0M shares, although our value is below their conversion price of about $157 (auditor F5, advocate F4) | **Disclosed.** Cash settlement at maturity gives -$1.27 (auditor). Treating them as debt at the risk-free rate gives -$2.71 (advocate). Unchanged, because settlement depends on the price path and the capped calls, and today's price is above the conversion price. | | | -1.27 to -2.71 not taken |
| 6 | The 15-year fade holds the FY2032 GAAP margin of 13.8% flat, though it is still rising about 2.5 points a year (advocate F1) | **Rejected as a change; disclosed.** A GAAP margin that keeps rising after FY2032 needs a margin before stock pay above 27.5%, or stock pay below 12.5%. Our 12.5% is already below ServiceNow's 14.7% of 2025 (advocate's check), and that world is our optimistic case. Holding the margin flat is the method's default. | | | +7.8 (18%) to +11.5 (20%) not taken |
| 7 | Cash compounds at 2.8% after tax but is discounted at the 10.75% cost of equity (advocate F2) | **Disclosed.** This is the shared method's exit bridge, not this model's choice. It is proposed for the CIO in process.md. | | | +6.67 not taken |
| 8 | The 20.9x exit multiple, from Oracle, Microsoft and Teradata, is wrong for a company growing 16% (advocate F5; flags 16 and 17) | **Rejected as a change; disclosed.** 20.9x is the method's default anchor and was fixed before the blind commit. Outside view at 30x, 40x and 60x: +$16.8, +$35.1 and +$71.8. **This is the input that flips the call.** | | | see text |
| 9 | The odds stack the downside twice: skewed base ranges plus a 25% bear world (advocate F7) | **Disclosed.** The base world's median is $88, against $81 for the mixture. Fair value does not change. The world weights are the case weights set before the market pass. | | | 0 |

How each flag was answered:

- **Flag 3: cleared.** Both reviewers found no risk charged twice; the input the market reads differently is the exit multiple.
- **Flag 14: explained.** See section 5 and row 9.
- **Flag 16: answered by the advocate's price-right world.** That world combines:
  - our optimistic operating path, with growth 4 points above the central;
  - a GAAP margin that rises to 25% during the fade;
  - cash counted at face.

  It is worth $180 at 20.9x, $240 at 40x and $302 at 60x, and reaches the price at 66x FY2032 GAAP EBIT (`review-advocate/findings.txt`). ServiceNow was valued at 86x at its 2025 fiscal year end, with a similar growth and margin profile (advocate, `findings.txt` F5).
- **Flag 17: answered in section 4.** Run through our model, the Street's path is worth $68 to $102. The rest of the gap to its $377 discounted target is the multiple: about 176x FY2032 GAAP EBIT (advocate).
- **Flag 18: not widened.** The price sits outside our 10th-90th range of $28 to $199. The advocate notes that SNOW's lowest close since listing is $109 (2024-09-06), and that 32% of the last year's closes fall inside the range. We widen only from history, never toward the price.

Smells: the net change is +1.1% (P4 is clear). The two changes moved in opposite directions (P3 is clear).

## 10. Sources

All primary unless marked; filings masked before reading (assumptions.md lists the
exposures). 10-K FY2026 (filed 2026-03-20); 10-Qs Q1 and Q2 FY2027 (filed 2026-05-29,
2026-09-04); earnings releases Q2 FY2025 to Q2 FY2027 (8-K exhibit 99.1); 8-Ks 2025-10-27,
2026-03-31, 2026-06-30, 2026-07-16, 2026-10-02; transcripts of the Q1 FY2027 call
(2026-05-27), Investor Day (2026-06-02) and the Q2 FY2027 call (2026-09-02) via Quartr on
Market data (secondhand transcription of the company's own words); market data statements fetched
2026-10-23; Damodaran equity risk premium 2026-10-01.