SHOP

We retain Avoid at $168.51. The current-date model replay gives fair value of $83.18 per share, 50.6% below the quote. The operating forecast and valuation methods are unchanged from the October 15 underwrite; there has been no new Shopify filing since the Q2 2026 10-Q and release. We re-dated the model to October 23, refreshed the market snapshot, reran the Street path through the model and completed an independent opposite-side review.

The frozen October 15 report records $83.03 in its review table; replaying its saved inputs through the current same-code model gives $83.050157. Both are $83 at the report's whole-dollar headline precision. The comparable October 23 replay is $83.181799, a $0.131642 change from the shorter discount period alone; the operating forecast, cash flows, and exit value are unchanged.

This revision is unblinded and cannot restore the original blind status. The October 15 report was already partly unblinded: before its blind commit, the Q2 2026 10-Q equity-investment note let us infer historical share prices. The blind value and that exposure history are retained from the original report.

Terms used below. NOPAT is operating profit after tax and stock pay. GMV is the value of merchants' sales. “Bear,” “central” and “bull” are coherent downside, base and upside operating cases. “Street” means analyst consensus, which is a comparison rather than company guidance.

October 23 revision findings

This revision keeps the original central forecast because the review found no arithmetic error and no dated evidence that supports replacing the forecast with a different central margin. It also does not treat the forecast as proven. The key unresolved judgment is whether operating leverage can raise after-stock-pay operating margin to 19.8% in 2031. The available evidence supports the direction of expense leverage; it does not establish that exact 2031 level. We disclose the uncertainty and show the model impact without forcing a historical margin reset.

Street comparison (flag 17). The current market data snapshot is pinned in audit/current-market-snapshot.json. At 9.74%, the $176.16 average target discounts one year to $160.52; $83.18 is 0.518 times that amount. A target is not a company forecast, so we ran the vendor's 2027–2030 revenue and EPS estimates through the report's own operating model. The cases below keep 2032 NOPAT growth at 13% and the exit multiple at 21.5x. Only the EPS-derived cases change the operating-profit path; 2031 is an explicit extension because market data supplies no estimate that far out.

Both main EPS extensions remain below the current quote and discounted target. The last row is an accounting-crosswalk sensitivity, not the preferred path.

Market data says its analyst EPS and net-income estimates are non-GAAP, but does not itemize analysts' adjustments. We therefore do not state that every estimate specifically excludes stock pay. The main crosswalk treats consensus EPS as pre-stock-pay and subtracts stock pay at the historical 2025 ratio of 3.885% of revenue; the alternative leaves stock pay in the EPS estimate. Under that alternative, the same two extensions value the business at $110.43 and $117.34. The crosswalk and all four scenario outputs are reproducible in audit/street-replay.py. No 2031 consensus is available. The 2031 growth extensions and stock-pay crosswalk are assumptions, not observations. The independent advocate's saved-October-15 replay also supported Avoid; it reached $98.34 using a separate revenue-to-GMV and cash flow crosswalk. The difference from our saved-date actual-model replay is small and comes from its simpler proportional-revenue treatment of segment mix and loan funding; the current-date scenario above uses the report's segment margins and model.py directly.

Terminal margin (C6). The central case moves from a 12.7% reported 2025 operating margin to 19.8% in 2031 by reducing operating expenses before transaction and loan losses from 26.0% of revenue in 2027 to 21.5% in 2031; losses stay at 3.8%. The 2025 margin is the highest reported since 2016, but the data do not establish a precise cyclical boundary: 2021 margin was 5.8%, and 2022–2023 were loss years. We do not label 2021 or 2025 as the required “pre-upcycle” peak without evidence for that cycle date.

The Q1 2026 CFO said Shopify had kept headcount slightly down for three years, did not expect that to change, and could continue lowering sales and marketing, research and development, and general and administrative expense as percentages of revenue. Q2 showed continued near-term leverage, and the Q3 guide was 33–34% total operating expense as a share of revenue. These are dated support for direction and near-term discipline, not a 2031 margin guide. Holding every other input constant and changing only the 2031 margin gives $58.96 if it is reset to the 2025 actual of 12.70%, and $35.58 at the 2021 actual of 5.83%. These are diagnostic sensitivities, not central estimates. C6 remains an open forecast risk; neither sensitivity is adopted as the central case. The existing Q3 and 2027 falsifiers remain the tests for whether the operating-leverage path is succeeding.

Call. The revised evidence supports the same Avoid conclusion. Current Street revenue alone adds about $6 per share to fair value; the EPS-based path adds more, but depends on a non-GAAP-to-operating-profit crosswalk and an unobserved 2031 extension. The current quote still exceeds the modeled bear, central and bull values. This remains a valuation judgment, not a claim that the company or its recent operating trend is weak.

Carried-forward October 15 underwriting

The detailed company, segment, balance-sheet, forecast, case and original review analyses below come from the published October 15 report. They are retained because no new company filing changed those inputs. Any $178.08 price, Robinhood target or Street estimate below is the historical October 15 basis; the current market snapshot and review findings above govern this revision. The original report remains unchanged at its frozen published path.

Business economics: Shopify keeps about 3% of merchants' sales and needs almost no fixed capital

• Who pays. Merchants, from one-person shops to General Motors and Balmain (Q2 2026 call). Merchant solutions earn 2.33% of GMV (2025), up from 2.21% (2023): payment processing and currency fees (Shopify Payments penetration 61% in 2024, 65% in 2025, 68% in Q2 2026, in 40 countries), lending interest and fees ($175M in 1H26), partner revenue shares, shipping labels, POS hardware and ads. Subscription revenue grows with merchants and plan mix (monthly recurring revenue, MRR, $221M, +19%; Plus is 34% of MRR).
• Margins. Gross margin 48.1% (2025) because payments carry card-network costs: merchant solutions 37.7%, subscription 81.1%. Operating expenses before losses fell from 35.7% of revenue (2024) to 31.8% (2025) and 31.5% (1H26) on flat headcount. Transaction and loan losses rose from 2.6% to 3.6% and 3.8% of revenue as lending grew. Stock pay is inside GAAP operating expense (3.9% of revenue in 2025).
• Capital. The business needs almost no fixed capital (capex $26M in 2025). The capital it uses is the loan and cash-advance book: $2,184M net at 2026-06-30, up from $1,784M in six months, 5.7% of gross balances more than 180 days late (10-Q note 6). Shopify buys these loans from a bank partner; that book is what grows with GMV.
• Balance sheet. No debt since the $920M convertible matured in 2025. Cash and marketable securities $5,472M; equity and other investments $4,854M (Affirm $1,655M, Global-E $768M, Klaviyo $280M, private stakes $1,555M, convertible notes $587M) and the Flexport stake $559M (10-Q note 4). These swing GAAP net income by hundreds of millions a quarter; we exclude them from earnings and add them once in the bridge.
• What limits it. Fees are set against Stripe, Adyen and card networks, and large merchants negotiate; agentic shopping (OpenAI, Google, Meta's Muse) may move discovery and checkout off merchants' own sites, which management says carries the same economics today (Q2 2026 call); AI costs sit in subscription gross margin; the Canada Revenue Agency (CRA) has proposed a transfer-pricing assessment for 2020, disputed (10-Q note 12).

Shopify has beaten its own guide in every quarter since Q3 2024

Shopify guides growth in words ("low-thirties"). Taking word midpoints (low 2, mid 5, high 8 in the decade named), it beat every quarter from Q3 2024 to Q2 2026 (printed by model.py from the releases):

Q2 2026: revenue +34% against "high-twenties"; gross profit +31% against "mid-twenties"; opex 34.0% against 35-36%. Q3 2026 guide: revenue growth "low-thirties", gross profit "mid-to-high twenties", opex 33-34%, stock pay $150M, no significant FX. Our Q3 forecast is built bottom-up (GMV +31.5%, take rate +10 bp, subscription +21%): revenue $3,787M (+33.2%), about a quarter of the usual beat because the Q3 2025 comparison grew 32%.

The forecast and valuation: $83 central, from a 21.5x exit multiple

model.py, output in output.txt; assumptions and sources in assumptions.md.

• Segments on their own paths. GMV growth fades from 31% (2026) by 3-7 points a year to 13% in 2031, from a company still accelerating (2023 +20%, 2024 +24%, 2025 +29%) on a base management puts above 14% of US e-commerce. The take rate gains 2-4 bp a year as payments penetration rises from 68%. Subscription revenue grows 18% falling to 11%.
• Margins. Merchant gross margin 37.5% and subscription 79% from 2027; operating expenses before losses fall 1.25 points a year to 21.5% of revenue (they fell 4 points a year in 2024-26); losses stay at 3.8%.
• Stock pay is a cost (inside GAAP opex); the share count is held at 1,294.5M (outstanding plus the 5.3M vested options; unvested options and RSUs are left out because their cost is in future stock pay), and buybacks absorb issuance. Counted once.
• Reinvestment. The loan book grows at 0.45% of GMV, charged against owner cash flow; capex equals depreciation.
• Bridge. Today: cash and securities $5,472M plus vested-option exercise cash $403M, less operating cash of 3% of the last twelve months' revenue ($398M), plus equity investments and Flexport ($5,413M) at June 30 carrying value: $8.4 a share at face. Each year's owner cash flow, less the growth in operating cash, is paid out at the year end (the $5B buyback authorization, 8-K 2026-06-02): $10.0 a share at 9.74%. The exit value is the operating business alone.
• Three methods, central. Judged exit multiple 21.5x 2032 NOPAT ($105), set at the case's own long fade (21.1x, rounded up); long fade, growth 13% fading to 3% over 15 years and return on new capital 60% fading to 40% ($103); outside view, 31.9x the peers' latest trailing GAAP P/E on 2031 NOPAT ($138). The value-driver formula at 6% perpetual growth, 40% return and 9.74% gives 22.7x, printed beside the judged multiple; it prices one mature return for ever. The blind 28x implied 6.8% growth for ever at a 40% return, above what the case's own fade pays for, and the reviews cut it (answer 4). The outside view sits 31% above the median because the peers' 32x is a market multiple of businesses whose prices we do not test; we believe the case's own multiple and fade.
• Cost of capital. Risk-free 5.079% (US 10-year, TradingView, 2026-10-15), Damodaran's implied premium 4.20% (2026-10-01), peer median Blume beta 1.11 (V, MA, PYPL, MELI, AMZN, BKNG, INTU, EBAY): cost of equity 9.74%. At Shopify's own Blume beta (2.02; raw 2.53) the cost of equity is 13.57%, above the 12% hurdle (review flag 11, below). At that rate the central fade is worth 12.5x 2032 NOPAT, so the exit multiple falls with it: exit $61, fair value $49, purchase price $52 (review flag 7, below). Bear $25, bull $74.
• Cases. Bear ($36): GMV growth 17% falling to 8% ($0.85T in 2031), take rate slipping to 2.29% under fee pressure, merchant gross margin 35.5%, losses 4.3%, opex stuck near 26%, 15.5x at exit (its fade 15.3x). Bull ($132): GMV 28% to 16% ($1.32T), take rate 2.68%, margin to 22%, 26.5x (its fade 26.2x). Weighted 25/50/25 (our weights): $84. The bear's outside view ($69 at exit) is 92% above its median because the peers' 32x does not fit a business growing 8%; we believe the bear's own multiple and fade ($36, $36).

Sensitivity of central fair value: NOPAT -20%/-10%/+10%/+20% gives $68, $75, $91, $98. Exit multiple 20x gives $82 (the fade then sets the median), 24x $91; from 28x up the median moves to the outside view and fair value stays near $103. Mature return on new capital from 25% to 100% moves the fade between $99 and $108 and fair value between $83 and $85.

Original October 15 price: 2.47 times our 2032 profit, or 53x it

From unblinded.py (output in unblinded-output.txt):

• After today's cash and stakes ($8) and the payouts to 2031 ($10), the 2031 exit must be $259 to earn the 9.74% cost of equity (our central $105); to earn 12%, $289. That needs 2032 NOPAT of $15.6B at our 21.5x (2.47 times our $6.3B), or 53x our 2032 NOPAT. Scaling every year's NOPAT by 2.24 (same multiples) reaches the price.
• The Street's revenue is close to ours: Street consensus (2026-10-15) has 2026 at $15.06B (38 analysts; ours +1.9%), 2027 $18.98B (+0.5%), 2028 $23.9B (-3.9%, 24 analysts), 2029 $28.6B (-5.8%, 22), 2030 $33.8B (-8.1%, 14). The Street's EPS excludes stock pay (its 2025 figure, $1.46, is about operating income plus stock pay plus interest after tax, not GAAP $0.94); on that basis, with 3.5% earned on today's surplus cash, our earnings are +2% against the Street in 2027 and -14% in 2030 (7-11 analysts). So the market is paying for a multiple far above any mature peer's, not for a faster business than the Street's.
• Alpha's view (alpha rank SHOP --raw --json, 2026-10-15): rank 294 of 713, score 58.9, Neutral; quality A (rank 43), growth B, momentum C, expectations C, valuation D (rank 636). Alpha agrees the business is excellent and the price is high; our Avoid rests on value, where Alpha is near the bottom.
• How the price got here (Robinhood daily closes). Q2 results on 2026-08-05: $123.30 (08-04 close) to $144.24, +17.0% on the day against QQQ -0.9% ($723.85 to $717.30); $152.61 on day 5 (2026-08-11) against QQQ $718.45. The stock then fell to $139.82 (2026-09-01) and rose 27% to $178.08 by today. Headlines in that stretch (TradingView, read 2026-10-15) include Meta's Muse agentic checkout using Shopify (2026-09-21) and a Wedbush Outperform reiteration at $176 (2026-09-23); no company release.

Original October 15 saved estimates: we agreed with the Street on revenue and disagreed on the multiple

The market's thesis: Shopify becomes the commerce layer under AI agents as well as stores, keeps GMV growing near 30% for years, and deserves a multiple no mature payments or commerce company now carries. Ours: the operating story is strong and mostly in our numbers (they track the Street to 2027), but a business that keeps 3% of GMV, pays card networks most of its payments revenue and lends to its merchants is worth a network-like multiple at most, and $178.08 needs either two and a half times our profit or two and a half times that multiple. What settles it: whether GMV growth holds above 25% into 2028 and operating expenses keep falling 3-4 points a year, as in 2024-26 (the bull's path and beyond).

We have no edge on the near-term forecast; the company beats its own guide every quarter and our Q3 numbers sit above the guide. Our edge, if any, is discipline on the exit multiple and on charging the loan book and stock pay as costs.

Why the other side could be right

• Growth is accelerating, not fading: GMV +20%, +24%, +29%, +33% (2023 to 1H26). If 2027-2031 growth stays near 25% instead of fading to 13%, 2031 GMV is about $1.5T and, with the bull's 22% margin, 2031 profit is about 1.45 times our central (our arithmetic: $1.5T at a 2.6% take plus $7B of subscriptions); with the margins in the next point it nears twice. At our 21.5x the price needs two and a half times; at the blind 28x, about twice.
• Opex fell 4 points of revenue a year in 2024-26 on flat headcount. If AI keeps headcount flat while revenue doubles, operating margin could reach 25-30% by 2031, not our 19.8%.
• Agentic commerce could raise Shopify's share of the economy's GMV, and the Shop Pay identity network could earn more per transaction than we allow.
• But the bull case already carries 16% growth in 2031, a 22% margin and 26.5x, and is worth $132. The price sits above our whole bear-to-bull range: the market is paying for growth beyond our bull and a multiple above every mature peer's, together. A multiple alone does not close the gap: from 28x up the median moves to the outside view and fair value stays near $103.
• The advocate's strongest set of changes it could defend (merchant margin held at 38.2%, faster take-rate gains, GMV halfway to the bull, the bridge fixes) gives $107 and a $97 purchase price on the blind 28x; with everything generous, $123. It could not make the case for Buy.

Catalysts and falsifiers

• 2026-11-03 (company events page): Q3 2026 results. Our forecasts: revenue $3,787M (+33.2%; guide low-thirties), GMV $121.0B (+31.5%), gross profit $1,805M (+29.8%; guide mid-to-high twenties), opex 32.7% of revenue (guide 33-34%), operating income $566M.
• February 2027 (not confirmed): Q4 2026 results and the first 2027 outlook.
• Falsifier for Avoid: 2027 GMV growth of 28% or more with opex before losses at or below 25% of revenue, or a 2027 operating margin of 20% or more; re-underwrite then.
• Toward the bear: take rate below 2.35% of GMV, losses above 4.5% of revenue for two quarters, or 2027 GMV growth below 18%.

Original October 15 review flags: six raised, each answered

The flags are numbered patterns in our review checklist (review.md).

• Flag 1 (medium; net move from the blind value): the value moved -$14.5 (-14.9%) from the blind $98 after the price was read; the largest change is the exit multiple (-$19.5), resting on the blind model's own fade, not on a management claim or a Street figure.
• Flag 3 (medium; value close to or far from the price): fair value 53% below the price (40% or more away).
• Flag 4 (low; outside view far from the median): the central outside view is 31% above its median and the bear's 92% (answered above and in answer 10).
• Flag 6 (medium; a lenient input chosen over a harsher sourced one; raised by the auditor, we had said it did not apply): the 28x was lenient against the case's own fade and is cut (answer 4); the beta is the rule's peer median, with the harsher own beta printed beside it (answer 9).
• Flag 7 (low; fair value moving a lot between own and peer beta): the purchase price is $76 at the peer beta and $52 at the own beta (-32%), because the own-beta cost of equity also lowers the exit multiple through the fade (answer 11).
• Flag 11 (low; cost of equity above the 12% hurdle): at the own beta only (13.57%).
• Other flags checked and not raised: 2 (view unchanged), 5 (central within 2-8% of the Street on revenue through 2030 and within 14% on earnings before stock pay), 8 (outstanding shares and vested options in the count; unvested awards are costed in stock pay; no convertible, no at-the-market program), 9 (10-K, two 10-Qs, eight releases, two calls read), 10 (today's batch of 15 records has 10 below and 5 above the price), 12 (the reviews moved the central value 14.9%), 13 (equity investments and Flexport carry filing lines in 10-Q note 4).

Original October 15 adversarial review: neither reviewer moved the view off Avoid

Two reviewers (model auditor, opposite-side advocate), one round. Neither moved the view off Avoid. Rows 1-11 are the auditor's findings and rows A1-A8 the advocate's. Effects are measured on the final model (.tmp scratch runs that patch one input of model.py), except where a row says "blind model". Running order of the accepted changes: blind $97.54 → bridge (1, 2, A2/A3) $102.15 → share count (A1) $102.56 → multiples (4) $83.03.

Flag 6 in full: we had written that flag 6 did not apply because the 28x was below the peers' 32x and the networks' 33-35x. The auditor was right that the harsher sourced alternative was our own fade, not the peers, and the multiple is cut. The beta is not a lenient choice under the rule: the method fixes the Blume peer median and prints the own beta beside it, here $49 against $83.

Sources and date basis

• Company financials: SEC EDGAR, CIK 1594805; FY2025 10-K filed 2026-02-11; Q1 and Q2 2026 10-Qs filed 2026-05-05 and 2026-08-05; Q2 2026 results and the Q3 guide filed 2026-08-05. The Q2 10-Q is available at SEC EDGAR, and the company's Q2 2026 results include the Q3 guide.
• Call evidence: Shopify Q1 2026 and Q2 2026 transcripts dated 2026-05-05 and 2026-08-05, retained under data/; the CFO statements are cited above with their dates.
• Original underwriting inputs: market data filings, statements, peer ratios, estimate capture and returns, plus company filings and calls, pinned by data-manifest.json; most vendor and company captures are dated 2026-10-15.
• Current market snapshot: market data quote, price-target-consensus, and annual analyst-estimates endpoints captured together at 2026-10-23T14:43:32.628612Z; the exact inputs are in audit/current-market-snapshot.json.
• Market data says Analyst Estimates EPS and net-income figures are non-GAAP; it does not itemize individual analysts' adjustments. See the Street's analyst-estimates FAQ.
• The current package was checked separately from this revision: package 20261023T130617793764Z-e02ccc2c8451, build 54e4a1eea4cdcd425aa94df0fb227db1732d7286; it remained published and verified and selected the October 15 report. This dated revision is not published.

Validation and review status

• uv run python model.py and uv run python unblinded.py were rerun in this folder; saved output is in output.txt and unblinded-output.txt.
• uv run python audit/street-replay.py and uv run python audit/margin-sensitivity.py reproduce the Street cases and margin diagnostics from the pinned inputs.
• uv run python scripts/research-events.py docs/research/2026-10-15-shop found no 8-K or 6-K after the 2026-08-05 Q2 filing; no company filing arrived between the source capture and this revision.
• uv run python scripts/research-check.py docs/research/2026-10-23-shop passed the record, source pins, model and unblinded replays, decision block, and required next-results preview. Current checker flags 1 (medium), 3 (medium), and 17 (high); flag 17 is answered above and in the audit. Required review depth was met by the independent model audit and opposite-side advocate. The auditor's final review after this revision is recorded in audit/review.md.
• The report's earlier review table remains the 2026-10-15 baseline record. Its accepted repairs still replay correctly; this revision does not reopen them. C6 remains an open forecast judgment as stated above, not a passed check.

Revision history

The report's source and figure basis is explicit: all current market values are from the pinned October 23 snapshot; the operating model remains grounded in the latest company filing available at the October 15 source cut. No trade, portfolio, model-setting or publishing action was taken.