Visa V

Reference table

1. What the business actually is

The 10-Q filed 2026-07-29 states the identity plainly:

> "Visa is a global payments technology company that facilitates secure, reliable and efficient global commerce and money movement. We provide transaction processing services (primarily authorization, clearing and settlement) among consumers, issuing and acquiring financial institutions and sellers... Visa is not a financial institution. We do not issue cards, extend credit or set rates and fees for account holders of Visa products."

That last sentence is the whole risk profile in nineteen words. Visa carries no credit exposure, funds no receivables and earns no interest spread. Its balance sheet risk is settlement risk, not credit risk — and the 10-K notes that as of 2025-09-30 it held $9.2B of total available liquidity specifically to fund daily settlement if a client bank failed to settle.

Scale, from the 10-K: "During fiscal 2025, 329 billion payments and cash transactions with Visa's brand were processed by Visa or other networks, equating to an average of 901 million transactions per day. Of the 329 billion total transactions, 258 billion were processed by Visa."

Segments: there is exactly one. The 10-Q is explicit — "the Company's chief operating decision maker (CODM) is the Chief Executive Officer, who uses consolidated net income in assessing performance", and the filing refers to "the Company's one reportable segment." The vendor's seg_prod block is therefore not a segment table but a disaggregation of revenue by category, which is exactly what Visa reports. It is clean: the five lines sum to precisely $40.000B, the reported fiscal-2025 net revenue, to the dollar.

Revenue by category, fiscal years ending 30 September ($M):

The 10-Q supplies the driver for each line for the June 2026 quarter, and each one is independently growing:

• Service revenue — "increased... primarily due to growth in nominal payments volume of 11% and 10%, respectively, select pricing modifications and growth in card benefits." Note the mechanical lag the filing discloses: "Service revenue in a given quarter is primarily assessed based on nominal payments volume in the prior quarter."
• Data processing revenue — "growth in processed transactions of 10% and 9%... select pricing modifications, growth in value-added services and higher cross-border transaction mix."
• International transaction revenue — "growth in nominal cross-border volume of 14% and 15%, respectively, excluding transactions within Europe, partially offset by lower volatility of a broad range of currencies and business mix."
• Client incentives — "increased... primarily due to growth in payments volume."

Geography, and the finding hidden in it ($M):

United States net revenue grew 5.8% in fiscal 2025 while international grew 15.2%. Nearly three-fifths of Visa's revenue is now non-US, and essentially all of the growth is. This is the single most useful fact in the geography table and it cuts both ways: it is why Visa still compounds, and it is why the European litigation docket and the national-scheme risks in Brazil, India, Nigeria and the Gulf discussed in Section 8 matter more than a US-centric reading would suggest. The table is clean — the two lines sum exactly to $40.000B.

Value-added services is the growth engine and it is not in the vendor file at all. The 10-Q: "For the three months ended June 30, 2026 and 2025, revenue from value-added services was $3.8 billion and $2.8 billion, respectively. For the nine months ended June 30, 2026 and 2025, revenue from value-added services was $10.3 billion and $7.8 billion, respectively. Value-added services revenue increased 33% and 32%... primarily due to growth in Issuing Solutions, Acceptance Solutions and Advisory and Other Services." The 10-K sizes the fiscal-year series: $7.2B (FY2023), $8.8B (FY2024), $10.9B (FY2025, +24%), and it is now running above $15B annualised. The 10-K also states Visa offers "more than 200 products and services as of September 30, 2025." This line cuts across the reported revenue categories and appears nowhere in seg_prod; it exists only because we read the filing.

Two 2026 corporate actions the vendor's annual data cannot see: the acquisition of Prisma Medios de Pago S.A.U. and Newpay S.A.U. in Argentina for $1.5B in cash in February 2026, and the issuance in the same month of $3.0B of fixed-rate senior notes with 3-to-10-year maturities at 3.80%–4.70%. The vendor's newest cash flow year is FY2025 and shows acquisitions of $887M.

2. The share count nobody counts — the retrospective responsibility plan

This is the most important structurally distinctive fact about Visa, it is worth real money to class A holders, and no line in the vendor payload registers it.

Visa's 2008 listing left behind a machine for paying litigation bills. The 10-K describes it:

> "In accordance with the U.S. litigation escrow agreement, the Company maintains an escrow account, from which settlements of, or judgments in, the U.S. covered litigation are paid... Under the terms of the plan, when the Company funds the U.S. litigation escrow account, the value of the Company's class B-1 and B-2 common stock is subject to dilution through a downward adjustment to the rate at which shares of class B-1 and B-2 common stock ultimately convert into shares of class A common stock. This has the same economic effect on earnings per share as repurchasing the Company's class A common stock because it reduces the class B conversion rate and consequently, reduces the as-converted class A common stock share count with each deposit amount."

Class B is held, per the 8-K filed 2026-06-26, "predominantly by U.S. financial institutions and their affiliates and successors" — the banks that used to own Visa. When Visa pays for interchange litigation, the banks pay, and the class A share count falls.

The 8-K filed 2026-06-26 shows the machine running, with exact numbers:

A $250 million escrow deposit retired roughly 750,000 as-converted shares — an effective price of about $333 a share, at no cash cost to class A holders, because the money was going to the plaintiffs either way and the dilution lands on class B. Over the first nine months of fiscal 2026 Visa deposited $875 million into the escrow, and in fiscal 2025 another $875 million.

The May 2026 exchange offer made this mechanism sharper, and it is the single least-understood event of the year. From the 10-Q:

> "In May 2026, Visa accepted 3 million shares of class B-1 common stock and 120 million shares of class B-2 common stock tendered in the exchange offer. In exchange, Visa issued 61 million shares of class B-3 common stock and 23 million shares of class C common stock. The class B-1 and B-2 common shares exchanged have been retired. Future conversion rate adjustments for the class B-3 common stock will have four times and two times the impact compared to conversion rate adjustments for the class B-1 and B-2 common stock, respectively."

Read that carefully. Banks accepted a class of stock on which future litigation costs bite four times harder, in exchange for immediate liquidity in class C. From here, every dollar Visa deposits into the escrow retires four times as many as-converted shares as it would have before May 2026 — which is exactly what the 8-K table above shows, with the B-3 line absorbing 98.7% of the June adjustment. That is an accelerating, self-funding buyback whose funding source is the litigation itself.

How much fuel is left? The 10-K gives the number and it is the one to watch:

> "The estimated interchange reimbursement fees at issue in unresolved claims for damages in the U.S. covered litigation was approximately $49.6 billion as of October 1, 2023 and was approximately $39.4 billion as of October 1, 2025."

The claim pool has shrunk by $10.2B, or 20.6%, in two years. The 10-K also states the trigger for the next exchange offer: Visa may conduct a successive exchange "if the estimated interchange reimbursement fees at issue... have been reduced by 50% or more since the consummation of the prior exchange offer." Against the remaining as-converted class B-3 base of 90.6 million shares — worth roughly $33.5B at $369.59 — the mechanism still has real capacity, though it is not unlimited relative to a $39.4B claim pool.

The Europe plan works the same way through preferred stock. The 10-Q: Visa recovers "VE territory covered losses through periodic adjustments to the class A common stock conversion rates applicable to the series B and C preferred stock." Those conversion rates fell over the nine months from 0.6690 to 0.5830 (series B) and 0.7640 to 0.7140 (series C) — a 12.9% and 6.5% reduction in the as-converted claim of the European bank holders, again absorbing litigation cost away from class A. There is no escrow here; the equity absorbs it directly.

Where the litigation actually stands, from the 10-Q filed 2026-07-29 — and it is improving:

• "On November 10, 2025, Visa and Mastercard entered into a superseding and amended settlement agreement to resolve the Injunctive Relief Class claims and the Injunctive Relief Class plaintiffs filed a motion for preliminary approval of the settlement, which was granted on June 9, 2026. On July 15, 2026, the Injunctive Relief Class plaintiffs filed a motion for final approval."
• "Visa has reached settlements with a number of merchants representing approximately 95% of the Visa-branded payment card sales volume of merchants who opted out of the Amended Settlement Agreement... all actions that were scheduled for trial beginning in April 2026 in the Southern District of New York have been resolved."
• The U.S. Securities Class Action was "granted without leave to amend on June 29, 2026"; the Debit Surcharge Class Action was dismissed "without further leave to amend"; MiCamp Solutions was dismissed.
• Against that: Europe is getting worse, not better. "Since July 2013, proceedings have been commenced by more than 1,200 Merchants... Visa has settled the claims asserted by over 950 Merchants, and there are over 100 Merchants with outstanding claims. In addition, merchants continue to threaten similar proceedings... The Company anticipates additional claims in the future." And new: "On April 20, 2026, a group of merchants from across Europe filed a claim in the UK High Court... In May and June 2026, additional merchants asserted claims... The plaintiffs' damages period goes back at least six years from filing."
• A new US class action was filed 2026-04-21 (Potayto-Potahto) against Visa and Mastercard together, on behalf of merchants accepting cards since 2019-01-25, and three merchants have moved for partial summary judgment seeking a declaration that the forward-looking release in the amended settlement is invalid and unenforceable. That motion is the single biggest live threat to the settlement architecture, because it would reopen a matter both networks have spent a decade closing.

The escrow balance itself tells the story of the year: $3.0B at 2025-09-30, $888M at 2026-06-30 after $875M of deposits — meaning roughly $3.0B was paid out to settling merchants in nine months. The accrual and the escrow are not the same thing and the 10-K is careful to say so: "The accrual related to the U.S. covered litigation could be either higher or lower than the U.S. litigation escrow account balance."

3. Growth, client incentives and the two earnings numbers

The acceleration is real

Quarterly, and the trend is up:

The fiscal-2025 margin collapse has one cause and it reconciles exactly. Operating margin fell from 65.7% to 60.0%, a 570-basis-point drop, on a year when revenue grew 11.3%. The 10-K states that during fiscal 2025 Visa "recorded additional accruals of $2.2 billion to address claims associated with the interchange multidistrict litigation." Add $2.2B back to operating income of $23.994B and the margin is 65.5% — within 20 basis points of fiscal 2024. The entire margin decline is the litigation provision. That is a clean reconciliation, and it is also the reason the next subsection matters.

The two earnings numbers — and every screen shows the wrong one

Trailing four-quarter GAAP diluted EPS is $11.76. The sum of the four reported non-GAAP figures is $12.78. The consensus forward estimates are on the non-GAAP basis. The proof is arithmetic and exact:

> Q1 FY26 $3.17 + Q2 FY26 $3.31 + Q3 FY26 $3.32 + Q4 FY26E $3.42 = $13.22, against the vendor's FY2026 epsAvg of $13.207 (23 analysts).

The reconciliation lands within a cent. The estimate block is unambiguously non-GAAP. The 10-Q lists what is excluded: "Litigation provision... Deferred tax benefit... Severance costs..." and, per the 10-K, "Gains and losses on equity investments." On the litigation provision the filing is candid about the reasoning — "Litigation provision associated with these matters can vary significantly... and do not correlate to the underlying performance of our business" — and that reasoning is defensible for any single quarter and indefensible across two decades. Visa accrued $2.2B in fiscal 2025 and $1.1B in the first nine months of fiscal 2026. A cost incurred every year is a cost.

The multiple ladder, stated honestly, at $369.59:

The honest reading: Visa trades at roughly 26x GAAP-equivalent fiscal-2027 earnings, not the 24.6x the estimate line implies and certainly not the low-twenties multiple a rolldown to fiscal 2028 suggests.

Client incentives — the cost of holding the network together

The incentive rate has risen 490 basis points in five years and has risen in every single year. Visa now returns 28.3 cents of every gross dollar to its bank and merchant clients to keep the volume on its rails. The 10-Q attributes the latest increase to "growth in payments volume" and warns that future incentives "will vary based on... amendments to existing contracts or the execution of new contracts." This is the quiet price of the duopoly our knowledge base admires: the network is not price-warring with Mastercard on the merchant side, it is competing for issuer relationships on the incentive side. If the incentive rate ever steps up by 200 basis points in a year rather than 60, roughly $1.1B of net revenue disappears.

Operating expense drivers for the June quarter, from the 10-Q: total operating expenses rose 19% in the quarter and 13% over nine months, "primarily driven by higher personnel expenses" — which the filing attributes to "higher severance costs resulting from actions taken to drive operational efficiencies and reinvest in high-growth opportunities, as well as a higher number of employees and compensation costs, including from acquisitions." Marketing rose on "the FIFA World Cup 2026 in each period, and by the Olympic and Paralympic Winter Games Milano Cortina 2026 in the nine-month period." Note that 19% expense growth against 14% revenue growth is why the June quarter's operating margin fell to 59.1%.

4. Balance sheet and capital returns

The debt figure is confirmed by the filing; the net-debt figure is not.

The vendor's netDebt of $5.017B is total debt less cash and cash equivalents only, silently discarding $1.833B of short-term investments. Correctly computed, FY2025 net debt is $3.184B — against $22.6B of trailing net income, which is to say, nothing. The enterpriseValueTTM of $701.541B carries the same defect: rebuilt from the vendor's own inputs ($690.042B + $25.171B − $21.987B) it is $693.226B, a +1.2% residual — immaterial, well within tolerance, and noted rather than smoothed.

The share-count problem, which is larger than the net-debt problem. Three different share counts appear in a single file:

The 10-Q is explicit that the correct count includes everything: "Diluted class A common stock earnings per share calculation includes the assumed conversion of all class B-1, B-2, B-3 and C common stock and participating securities on an as-converted basis." The vendor's market capitalisation appears to count class A alone. At the as-converted diluted count of 1.933B, Visa's equity value at $369.59 is $714.4B, not $690.0B — a $24.4B, 3.5% understatement.

Consequences, stated plainly: the price-to-earnings ratio is unaffected (price ÷ EPS needs no share count, and both are struck on the as-converted basis). But priceToSalesRatioTTM (15.51x), evToSalesTTM (15.77x), evToEBITDATTM (24.11x) and enterpriseValueMultipleTTM (24.11x) are all roughly 3.4% too low. We use the price-to-earnings ladder in Section 3 and Section 6, and we do not quote an EV multiple for this issuer.

Capital structure and returns:

• Debt: $25.171B carrying value at FY2025-end. February 2026 added $3.0B of new senior notes (3-to-10 year, 3.80%–4.70%); against that, €1.35B ($1.6B) and $4.0B of principal were repaid on maturity during the nine months — a net reduction of $2.6B in senior notes. Commercial paper stood at $1.5B at 2026-06-30 and $500M at 2026-07-28, at a weighted-average rate of 3.77%; the programme authorisation was raised from $3.0B to $7.0B in July 2026. A $7.0B revolving credit facility expiring May 2028 was undrawn. A $1.5B principal payment falls due April 2027.
• Buybacks: "In April 2026, our board of directors authorized a $20.0 billion share repurchase program... For the nine months ended June 30, 2026, we repurchased 50 million shares of our class A common stock in the open market for $16.5 billion. As of June 30, 2026, our share repurchase programs had remaining authorized funds of $28.4 billion." That is an average repurchase price of roughly $330 a share — 10.7% below today's close. In fiscal 2025 Visa repurchased 54 million shares for $18.2 billion per the 10-K.
• Dividend: raised to $0.670 per share quarterly on 2026-07-28, payable 2026-09-01. That is $2.68 annualised, a 0.73% yield. Dividends declared and paid were $1,273M in the June quarter and $3.9B over nine months, against $4.6B for all of fiscal 2025.
• Total shareholder yield ≈ 3.9% — roughly $22B annualised of buyback on a $690B vendor market cap (3.2%) plus 0.73% of dividend.

Quality of the balance sheet, honestly. Goodwill and intangibles are $47.525B, or 47.7% of total assets, so tangible book value per share is negative at −$6.89 and tangibleAssetValueTTM is −$13.179B. Working capital is −$465M and the current ratio is 0.985 — both artefacts of settlement payables and litigation accruals sitting in current liabilities, not distress signals for a business generating $23B of operating cash flow. Price to book is 20.1x and the vendor's rating model scores price-to-book at 1 out of 5 as a result; for a business with a 61.3% return on equity and negative tangible book, book value is not a meaningful valuation anchor and we do not use it.

5. Cash flow

Capital intensity is trivial and that is the whole business model: capital expenditure is 3.5% of revenue and 6.9% of operating cash flow, and free cash flow converts at 93.1% of operating cash flow. Free cash flow per share is $10.98 and the free cash flow yield is 3.05%.

FILING OVERRIDES VENDOR — the fiscal 2025 buyback line. The vendor's cash-flow block reports repurchases of $13.389B in fiscal 2025. The 10-K states: "During fiscal 2025, we repurchased 54 million shares of our class A common stock in the open market for $18.2 billion." Fifty-four million shares at $18.2B is an average of $337, which is consistent with the fiscal-2025 price range; the vendor's figure is not reconcilable with the share count. We use $18.2B. The same block is internally inconsistent in a second way: netChgCash of $5.224B does not reconcile with the $8.179B increase in cash and cash equivalents on the vendor's own FY2024→FY2025 balance sheets, a $2.955B gap that the restricted-cash treatment of the litigation escrow only partly explains. We use the operating-cash-flow, capex and free-cash-flow rows, which cross-check against the trailing per-share ratios, and we treat the financing and net-change rows as unreliable.

Operating cash flow is going the wrong way in the near term, and the filing says why: "Cash provided by operating activities decreased over the nine-month prior-year comparable period primarily due to higher litigation payments, higher incentive payments and timing of payments related to income taxes, partially offset by growth in our underlying business." Roughly $3.0B left the escrow account for settling merchants over those nine months. This is a genuine, temporary drag from the litigation resolving — cash going out the door as claims close. It is a good problem, and it is why trailing free cash flow understates the run-rate.

6. Valuation — priced in or room?

At $369.59 (vendor market cap $690.0B; as-converted equity value ≈$714.4B):

Fiscal-year labelling matters here and is easy to get wrong. Visa's fiscal year ends 30 September. The estimate rows dated 2026-09-30, 2027-09-30 and 2028-09-30 are fiscal 2026, 2027 and 2028 — so "FY2027E" covers October 2026 through September 2027 and is only about fourteen months forward from today, not twenty-six. The forward multiple is therefore nearer than a calendar-year reader would assume, which makes 24.6x less of a discount than it looks.

Peer context, and a warning about the peer set. The vendor peer list is ALLY ($13.7B), AXP ($234.1B), BAC ($446.4B), JPM ($958.0B), MA ($500.9B), PYPL ($50.1B), SEZL ($6.0B), SLM ($5.2B). Four of those — Ally, Bank of America, JPMorgan and SLM — are balance-sheet lenders whose earnings come from net interest income and whose risk is credit risk. They are not comparable to a network that takes no credit risk, and no inference is drawn from them. American Express operates a closed loop (issuer, network and acquirer in one) and carries card-member receivables; it is adjacent, not comparable. Mastercard is the only true comparable in the list, and it is dealt with directly in Section 11. No peer multiples are supplied in the file, so no peer-multiple table is drawn.

The vendor's rating model scores Visa B with an overall score of 3 — return-on-equity and return-on-assets sub-scores of 5, discounted-cash-flow 3, debt-to-equity 2, and price-to-earnings and price-to-book both 1. The two 1s are the model saying, in its own way, exactly what this section says: the business quality is exceptional and the price is not cheap.

6a. What today's price assumes (the inversion)

At $369.59 — 24.6x FY2027E consensus EPS of $15.005 on a non-GAAP basis, or roughly 26.4x on a GAAP-equivalent basis — the price embeds the following falsifiable claims:

• Net revenue compounds ≥10.4% annually through fiscal 2028, reaching $56.042B. (Consensus-derived: $45.800B FY2026E on 27 analysts, $50.761B FY2027E on 28, $56.042B FY2028E on 26 — coverage is deep and consistent across all three years, which is unusual and reassuring.) The nine-month actual of +15.3% currently runs ahead of this.
• EPS reaches ~$17.04 by fiscal 2028, a 13.6% annual compound from the FY2026 base. (Consensus-derived; 15 analysts on FY2028.) Note that this requires the buyback to keep running: consensus net income grows more slowly than consensus EPS.
• Client incentives stay near 28% of gross revenue and do not step up. (Our number, derived from the FY2020–FY2025 series in Section 3.) The rate has risen every year for six years; the price assumes the increments stay at 30-60 basis points a year rather than jumping.
• The market keeps paying ~25x forward non-GAAP earnings — i.e. no de-rating despite a 20-year litigation history, an accelerating European claim docket with over 100 outstanding merchants and new UK High Court filings in April, May and June 2026, and a named structural threat from alternative rails. (Our reading of the current multiple against the FY2027 consensus.)
• The litigation provision remains an "adjustment" rather than a cost. (Our number.) The market values Visa on non-GAAP earnings that exclude $2.2B (FY2025) and $1.1B (nine months FY2026) of interchange accrual. This is the most fragile assumption in the price. It is fragile precisely because it is not falsifiable in a single quarter — a bad quarter looks like an adjustment, and only a decade looks like a cost. The three merchants who moved on 2026-04-21 for a declaration that the forward-looking release in the amended settlement is invalid and unenforceable are the one event that could make it visible at once.

6b. The return bridge (why the multiple moves)

Expected return over 12–24 months decomposes as: EPS growth (+13.6% a year on consensus) + multiple drift (we assume a modest HOLD-to-slight-expansion, from 24.6x to ~26.5x FY2027E) + shareholder yield (~3.9%: 3.2% buyback plus 0.73% dividend).

Our base case assumes the multiple broadly holds, with a small expansion, and the honest reason is not conservatism — it is that the multiple has already de-rated. Twelve months ago the stock was near $339 (today's $369.59 less the +8.9% twelve-month return) on a trailing GAAP earnings base of roughly $9.90, or about 34x. Today it is 31.4x on trailing GAAP. Earnings compounded faster than price for a year; the multiple contracted by about three turns. That is the whole explanation for the strange pairing of "at the 52-week high" and "fifteen points behind the index."

Three forces argue against pushing the base multiple higher. First, the GAAP-equivalent multiple is already 26.4x for a 13.6% grower — a PEG near 2, and the vendor's own priceToEarningsGrowthRatioTTM of 2.12 agrees. Second, the incentive rate rises every year, which caps the operating-leverage story. Third, the structural questions in Section 8 have no resolution date, and multiples on businesses with unresolvable long-horizon questions do not expand for long.

Almost none of the expected return here is re-rating; the return is earnings growth plus a genuinely large shareholder yield. Base fair value of $398 is 26.5x FY2027E — about two turns of expansion off today, which is the return bridge's most contestable leg. Even without it, EPS growth of 13.6% plus a 3.9% yield is a respectable mid-teens gross return. If the multiple simply holds at 24.6x FY2027E, fair value is $369 — exactly today's price. That is the arithmetic behind the Hold.

The bull case at $450 requires 30x FY2027E, which is the street's own high target to the dollar, and roughly 32x on the GAAP-equivalent basis. That needs both the growth and a re-rating, and it is the fragile leg.

6c. Variant perception (where we differ, what would surprise)

• We differ from every screen on the earnings basis, and it is worth about 8%. Screens show Visa on non-GAAP forward earnings against a trailing GAAP figure, mixing bases. Trailing GAAP diluted EPS is $11.76 (31.4x); trailing non-GAAP is $12.78 (28.9x); the difference is the litigation provision, which Visa has recorded in size every year for two decades. Our GAAP-equivalent FY2027 multiple is ~26.4x, not 24.6x. Watchable number: the litigation-provision line in each quarterly print, and the accrual rollforward in Note 16 of each 10-Q.
• We differ on the share count, in Visa's disfavour on multiples and in shareholders' favour on the mechanism. The vendor market cap implies 1.867B shares against the 1.933B as-converted diluted count Visa uses — every market-cap-derived multiple in the file is ~3.4% too low. But the same class structure is a genuine hidden buyback: the June 2026 escrow deposit of $250M retired ~750,000 as-converted shares at zero cash cost to class A, and post-exchange the class B-3 adjustments bite four times harder than the old class B-1. Watchable event: each escrow-deposit 8-K, which discloses the exact conversion-rate change and the exact share reduction.
• We differ from the bull lane on how much the "toll booth" framing settles. Our knowledge base is rich on the moat — "a toll booth on global commerce with a near-unassailable network-effect moat", "a rational duopoly that protects industry margins rather than price-warring" — and all of that is true and none of it is a valuation. A toll booth at 26x GAAP-equivalent forward earnings with a 13.6% growth rate is priced as a toll booth. We do not think the quality is in dispute; we think it is in the price.
• We do NOT have a variant perception on the growth. Net revenue +14% in the quarter, value-added services +33%, cross-border +14%, processed transactions +10% — these are excellent and consensus already models them. Where our view equals consensus we say so, and that is itself a verdict input: a no-edge name at a full price defaults toward Hold.
• Positive surprise that would force a re-rate: final approval of the Injunctive Relief Class settlement (moved 2026-07-15) combined with resolution of the Potayto-Potahto challenge to the forward-looking release, which together would let the market start treating the litigation provision as genuinely terminal rather than perpetual — and would collapse the GAAP/non-GAAP gap. Secondarily, a successive class B exchange offer, which the 10-K says can be triggered once the claim pool falls 50% from $49.6B (it is at $39.4B, i.e. 20.6% of the way). Or value-added services revenue passing 30% of net revenue while still growing above 30%. Watchable numbers: the escrow balance ($888M at 2026-06-30), the claim pool ($39.4B at 2025-10-01), value-added services revenue ($3.8B in the June quarter).
• Negative surprise that would break the thesis: the client-incentive rate stepping up 200 basis points in a year, which removes roughly $1.1B of net revenue and would signal that the duopoly's issuer-side competition has intensified. Or a US interchange cap — our knowledge base records that Visa fell 6% in a day on interchange-cap fears alone. Or the merchants' motion for a declaration that the forward-looking release is invalid and unenforceable succeeding, which would reopen the settlement architecture and put the $39.4B claim pool back in play. Or stablecoin settlement volume growing rapidly outside Visa's rails rather than on them. Watchable numbers: the incentive rate (28.3% in FY2025), the escrow deposits, Visa's disclosed stablecoin settlement run-rate ($2.5B annualised at 2025-09-30).

Synthos fair values

All three anchors are multiples of FY2027 consensus EPS of $15.005 (26 analysts, the deepest EPS coverage in the estimate block), cross-checked against FY2028E of $17.041 and against the GAAP-equivalent of roughly $14.00.

• Bear ~$300 — 20.0x FY2027E (21.4x GAAP-equivalent). Cross-check: 17.6x FY2028E; 25.5x trailing GAAP. The scenario: an adverse ruling reopens the interchange settlement architecture, or a US interchange cap arrives, or the incentive rate steps up sharply, and the market re-rates a business with a permanent legal cost toward a market-like multiple. Note $300 is 9.2% below the 200-day moving average of $330.51 — a genuine de-rating, not a crash. −18.8%.
• Base ~$398 — 26.5x FY2027E (28.4x GAAP-equivalent). Cross-check: 23.4x FY2028E; 30.2x trailing non-GAAP. The scenario: revenue compounds as consensus models, value-added services keeps outgrowing the network, litigation continues to resolve rather than expand, the buyback runs at ~$22B a year, and the multiple holds roughly where the last two years put it. +7.7%, plus a 0.73% dividend.
• Bull ~$450 — 30.0x FY2027E (32.1x GAAP-equivalent), which is the street's high target to the dollar. Cross-check: 26.4x FY2028E. The scenario: the injunctive-relief settlement receives final approval, the claim pool falls far enough to trigger a successive class B exchange offer, value-added services passes 30% of revenue, and the market pays a clean-earnings multiple for a business it no longer regards as legally encumbered. +21.8%.

Base is 7.7% above spot; asymmetry is roughly 1.16:1 to the upside (−18.8% down, +21.8% up). Against a street consensus of $416.88 (27.8x FY2027E), our base is 4.5% lower — we are marginally more cautious than a street with zero sell ratings on the name. Superb business. Fair price. Hold.

7. Knowledge base — 18 claims, and one important false-positive problem

The search pattern was the case-insensitive word \bVisa\b, which also matches the immigration sense of the word. Every hit was inspected individually.

The three immigration false positives, named so the discard is auditable:

1. 2025-09-27, all_in, tagged ents=['H-1B visa program'] — "New $100K one-time H-1B fee uses market scarcity to push visas toward genuinely high-skilled roles." This one was tagged entity-level by the pipeline and is the reason the honest count is 18, not 19.

2. 2025-09-22, andreas_steno, ents=['US dollar'] — "the visa fee is another dollar-negative", a macro claim about immigration fees and the dollar.

3. 2025-07-23, all_in, ents=['Microsoft'] — "Big tech laying off thousands while claiming labor shortages and applying for overseas visas."

Separately excluded: one 2023-11-29 claim ("Stablecoins settled ~$8T last year... transaction value now exceeds Visa and PayPal") is tagged to a quarantined-misattribution source and is excluded outright under the standing attribution policy, regardless of its relevance.

Also flagged as low-information but retained: a 2022-07-10 claim tagged to Visa ("Decentralized organizations that remove permission-seeking... adapt and innovate better") does not mention the company in its text at all; it is retained in the count for consistency with the pipeline but carries no weight in the reading.

Stance mix on the 18 true claims: 11 bullish, 3 neutral, 4 bearish, across 7 distinct sources (we_study_billionaires 5, all_in 3, invest_like_the_best 3, jordi_visser 2, real_vision 2, business_breakdowns 1, lyn_alden 1), spanning 2022-02-09 to 2026-02-27. This is a thin lane by our standards — a tenth the depth of the largest files in this batch — and the confidence rating reflects that.

The bull lane — the moat, articulately:

> 2026-02-12 · we_study_billionaires · bullish · conviction 78

> "Visa is a toll booth on global commerce with a near-unassailable network-effect moat; secular cash-to-digital shift is a durable tailwind."

> 2026-02-12 · we_study_billionaires · bullish · conviction 70

> "Visa/Mastercard are a rational duopoly that protects industry margins rather than price-warring, expanding the pie by converting remaining cash to digital."

> 2026-02-12 · we_study_billionaires · bullish · conviction 68

> "Visa's 80% gross / 60% operating margins and near-zero incremental transaction cost drive powerful operating leverage and high free-cash-flow conversion."

> 2026-02-12 · we_study_billionaires · bullish · conviction 60

> "Growth levers: higher-margin cross-border (10% of volume, >1/3 of revenue), value-added services (+25%), and new flows ($200T TAM, <1% penetrated)."

> 2022-02-09 · invest_like_the_best · bullish · conviction 90

> "Visa and Mastercard have extraordinary, durable margins and returns on capital — you could halve MA's margins twice and still beat the average business."

> 2023-05-20 · business_breakdowns · bullish · conviction 82

> "Card networks are an entrenched protocol layer between fragmented issuers and acquirers; a huge, defensible moat despite taking the smallest clip of economics."

Two of these check out against the filings to a striking degree, and one does not. The 80% gross / 60% operating margin claim is exactly right — FY2025 gross margin 80.4%, operating margin 60.0%. The cross-border claim ("10% of volume, >1/3 of revenue") is directionally confirmed by international transaction revenue of $14.166B on net revenue of $40.000B, which is 35.4%. But the value-added services figure of "+25%" is now stale and too low: the 10-Q filed 2026-07-29 puts it at +33% for the June quarter. The knowledge base is six months behind the filing on the single fastest-growing line in the business.

A price anchor worth noting, because it is now historic:

> 2026-02-12 · we_study_billionaires · bullish · conviction 65

> "At ~$330 Visa looks attractive; recent 6% drop on interest-rate-cap fears was an irrational overreaction since Visa doesn't lend or earn interest."

> 2026-02-12 · we_study_billionaires · bullish · conviction 62

> "Chris Hone (TCI, 18%/$9.5B position, 18% CAGR 20+yrs) added to Visa in Q2/Q3 2025 near $330-350, signaling conviction at current prices."

Both of those calls were made at roughly $330. The stock is at $369.59 — 12% higher. They were right, and being right at $330 is not an argument for buying at $370. Note also that $330 is almost exactly where Visa's own buyback executed over the last nine months (~$330 average) and where the 200-day moving average sits ($330.51). Three independent processes converge on $330 as the reference price, and we are 12% above it.

The bear lane — smaller, more recent, and carried by the highest-skill voice in the file:

> 2026-02-27 · jordi_visser (skill 2.0) · bearish · conviction 80

> "Payment middlemen (fee-on-a-fee) get disrupted by stablecoins; large-headcount enterprises become defensive utilities, not growth businesses."

> 2026-02-21 · jordi_visser (skill 2.0) · bearish · conviction 75

> "Card networks are a deflationary victim; the K-shaped economy will move to stablecoins to save money, cutting out the middleman and stalling Mastercard growth."

> 2025-09-15 · all_in · bearish · conviction 65

> "Incumbents have regulatory muscle and scale but are slow to adopt tech and can't hire top talent, losing to nimble fintechs."

> 2022-09-03 · lyn_alden (skill 1.1) · bearish · conviction 65

> "A dollar stablecoin on Bitcoin/Lightning would be a big threat to Visa and Mastercard's debit networks — permissionless, cheaper, no ~3% fees."

The middle lane is the most interesting, because it argues both sides at once:

> 2025-09-18 · all_in · neutral · conviction 65

> "Visa/Mastercard are thin-margin (~10bps) tech companies that get stronger if they cut banks out via stablecoin transfers behind the scenes."

> 2025-08-27 · real_vision · bullish · conviction 70

> "Incumbent card networks are building on stablecoin rails, validating blockchain payments as the future settlement layer."

> 2025-11-22 · all_in · neutral · conviction 75

> "Ultra-high-margin stablecoin economics invite competition; Stripe, Visa and others will launch their own, so Tether won't keep the market to itself."

Three observations from this lane.

First, the entire bull case is essentially one conversation. Five of the eleven bullish entity-level claims share the date 2026-02-12 and a single source. That is depth of argument within one episode, not breadth of independent agreement. We treat it accordingly: the content is high quality and verifiable against the filings, but it is one voice, and the conviction rating is Moderate rather than High for exactly this reason.

Second, the bear case is narrower but sharper. The highest-skill source in the file (skill 2.0) makes the same structural argument twice within six days in February 2026, at convictions of 75 and 80, and it is not a valuation argument — it is a claim that the business model gets disintermediated. Nothing in the current numbers tests it. Volumes are growing, cross-border is growing 14%, the network is fine. That is precisely what makes it dangerous: it is unfalsifiable on a quarterly cadence, and so it is not in the price.

Third, Visa's own 10-K agrees with the bears about the mechanism and disagrees about the outcome. The filing names "cryptocurrency platforms (including stablecoins)" among alternative payments providers, notes that with the July 2025 GENIUS Act framework "stablecoins could potentially disrupt existing payment networks, including in cross-border and B2B transactions", and lists FedNow, the European Payments Initiative's Wero, a prospective digital euro, and on-shore processing mandates in Nigeria and South Africa. Visa's answer is co-option — four stablecoins, four blockchains, over $100B of crypto and stablecoin purchases facilitated since 2020, over $35B of spend through crypto-linked credentials, and a stablecoin settlement volume that "surpassed a $2.5 billion annualized run rate" as of 2025-09-30. Against roughly $16 trillion of annual payments volume, that run-rate is 0.016% of the business. It is a credible start and it is not yet an answer, and we score it as such.

Weak non-entity context, retained but unweighted: a 2026-07-18 claim that "A Stripe/Block/PayPal combo with 600-700M accounts plus stablecoin rails could vertically integrate into a powerful competitor"; a 2026-02-10 claim that blockchain will shift "trillions... from incumbents (Visa, Mastercard, custodians) to new rails"; a 2025-01-04 claim from gavin_baker that "Dollar stablecoins... will attack the Visa/MasterCard duopoly"; and, on the other side, a 2025-05-09 observation that "Hard data (Visa/Mastercard, earnings transcripts) shows consumer spending remarkably resilient" — which the June-quarter volume figures independently confirm.

Why net conviction is "positive" and confidence is "Moderate." Eleven bullish against four bearish is a positive lane. But five of the eleven are one conversation on one day; the four bearish include the highest-skill voice twice; the bull lane's price anchor was $330 and we are at $370; and the total is 18 claims from 7 sources, which is thin. A positive but shallow lane, with an unanswered structural objection, supports a Hold rather than a Buy.

8. Data integrity — what we rejected from the vendor file and why

Ten material items. Three are direct filing-versus-vendor contradictions, and in every one the filing wins.

1. FILING OVERRIDES VENDOR — the fiscal 2025 buyback. The vendor cash-flow block reports repurchases of $13.389B in FY2025. The 10-K filed 2025-11-06 states: "During fiscal 2025, we repurchased 54 million shares of our class A common stock in the open market for $18.2 billion." Fifty-four million shares at $13.389B implies an average price of $248, which is far below the fiscal-2025 trading range and is not credible; at $18.2B it implies $337, which is. We use $18.2B.

2. FILING OVERRIDES VENDOR — the value-added services line does not exist in the vendor file. The 10-Q discloses value-added services revenue of $3.8B in the June quarter (+33%) and $10.3B over nine months (+32%), and the 10-K gives $7.2B / $8.8B / $10.9B for FY2023–FY2025. This is roughly a quarter of net revenue, growing at twice the rate of the network, and it appears nowhere in seg_prod because it cuts across the reported revenue categories. Any growth analysis built on the vendor file alone misses the entire growth story.

3. FILING OVERRIDES VENDOR — the class B/C conversion mechanism and the true share count. The vendor payload contains no record of the May 2026 exchange offer (3M class B-1 and 120M class B-2 tendered; 61M class B-3 and 23M class C issued; future adjustments biting four times harder), of the 2026-06-24 escrow deposit of $250M which retired roughly 750,000 as-converted shares, or of the Europe plan's preferred-stock conversion-rate reductions (series B from 0.6690 to 0.5830; series C from 0.7640 to 0.7140 over nine months). The vendor's implied share count of 1.867B (market cap ÷ price) is 66 million shares — 3.4% — below the 1.933B as-converted diluted count Visa itself uses for EPS, and 46 million below the 1.913B implicit in the vendor's own per-share ratios. Three share counts, one file. Consequence: priceToSalesRatioTTM (15.51x), evToSalesTTM (15.77x), evToEBITDATTM (24.11x) and enterpriseValueMultipleTTM (24.11x) are all understated by roughly 3.4% and are not quoted in this dive. The price-to-earnings ladder is unaffected and is what we use.

4. netDebt ignores short-term investments — REJECTED and recomputed. Vendor netDebt of $5.017B is total debt ($25.171B) less cash and cash equivalents only ($20.154B), discarding $1.833B of short-term investments. Correctly computed, FY2025 net debt is $3.184B. The vendor's total debt figure of $25.171B is, by contrast, independently confirmed by the 10-K at a $25.2B carrying value — so the debt itself is sound and only the netting is wrong. Note the 10-K also discloses that the estimated fair value of that debt is $23.3B, 7.5% below carrying value, reflecting rates.

5. Enterprise value — a +1.2% residual, immaterial but stated. enterpriseValueTTM of $701.541B against a rebuild from the vendor's own inputs ($690.042B + $25.171B − $21.987B = $693.226B). Well within the 15% withholding threshold. But see item 3: on the correct as-converted share count the true enterprise value is nearer $717.5B, so the vendor figure is understated by about 2.2% in the direction that matters. No EV multiple is quoted in this dive.

6. The consensus estimate block is NON-GAAP and mixing it with trailing GAAP is the most common error on this name. The reconciliation is exact: Q1 FY26 $3.17 + Q2 $3.31 + Q3 $3.32 + Q4E $3.42 = $13.22 against epsAvg of $13.207. The earn_cal "actuals" ($3.32, $3.31, $3.17, $2.98) are non-GAAP; the inc_q diluted EPS figures ($2.97, $3.14, $3.03, $2.62) are GAAP. Both are correct; they are different measures, and the dive labels every multiple with its basis.

7. est net-income rows are internally inconsistent with est EPS rows — REJECTED. Dividing niAvg by epsAvg gives an implied diluted share count of 1.907B (FY2026E), 1.954B (FY2027E), 1.962B (FY2028E), 1.966B (FY2029E) — a share count that rises every year, at a company with $28.4B of buyback authorisation that has retired shares in every year on record (2.223B in FY2020 to 1.933B today). The implied net margins are equally implausible: 55.0%, 57.7%, 59.6% and 61.0% against a trailing 50.8%. All niAvg rows are rejected; forward valuation runs on epsAvg only.

8. The FY2030 estimate row is nonsensical — REJECTED outright. epsAvg falls from $19.597 (FY2029) to $19.160 (FY2030) and niAvg falls from $38.528B to $37.669B, while revAvg rises from $63.171B to $65.962B. Earnings declining on rising revenue at a business with a fixed cost base and a running buyback is not a forecast; it is thin coverage (6 EPS analysts against 9 for FY2029). No conclusion in this dive rests on FY2029 or FY2030.

9. est EBITDA and EBIT rows pass the sanity check but are not GAAP-comparable — used for nothing. Unlike roughly 70% of files, these rows are not obviously corrupt: FY2026E ebitdaAvg of $31.536B exceeds ebitAvg of $30.154B which exceeds niAvg of $25.189B, in the right order. But they are built on the same non-GAAP basis as the EPS rows and are therefore not comparable to the FY2025 actual EBITDA of $26.003B or operating income of $23.994B. No EBITDA multiple appears anywhere in this dive.

10. Internal vendor inconsistencies, flagged and resolved. (a) quote.yearHigh of $373.97 and yearLow of $293.89 are intraday extremes; tech.hi52 of $369.59 and tech.lo52 of $295.52 are closing extremes. We use the closing series throughout — which is what makes today's close exactly the 52-week high. (b) quote.sma50 of $340.4448 differs from tech.sma50 of $341.9126; we use the tech block, which is computed from the same close series as the rest of Section 9. (c) The vendor FY2025 cash-flow netChgCash of $5.224B does not reconcile with the $8.179B increase in cash and cash equivalents across the vendor's own FY2024 and FY2025 balance sheets — a $2.955B gap only partly explained by the restricted-cash treatment of the $3.0B litigation escrow. We use the operating-cash-flow, capex and free-cash-flow rows (which cross-check against the trailing per-share ratios) and reject the financing and net-change rows. (d) inc_q interest expense of $566M in the March 2026 quarter against $194M in each adjacent quarter is unexplained and is roughly 2.9x a rate consistent with $25B of debt at ~3.5%; the 10-Q's expense table did not survive text extraction, so it cannot be verified. Flagged, not used. (e) inc_q gross margin drops to 76.5% in the June 2026 quarter from 81.3% in March, a swing far outside the FY2020–FY2025 range of 79.3%–80.4%; this is a cost-of-revenue allocation artefact, not a business event — operating margin, which the filing corroborates, is used instead.

A note on what the filings could NOT provide. These SEC text extractions preserve narrative prose but drop numeric table cells. The volume tables, the components-of-net-revenue table, the balance sheet, the operating-expense table, the share-repurchase table and the earnings-per-share tables are all present as headings with their footnotes and their numbers absent. Every filing figure quoted in this dive comes from narrative sentences, not from tables — which is why, for example, we can state that payments volume grew 11% but not the dollar amount of payments volume. That is a real limitation and it is stated rather than papered over.

Not defects, correctly reported: dividendYieldTTM of 0.703% and dividendPerShareTTM of $2.60 are consistent with the pre-increase $0.65 quarterly rate and will step to $2.68 following the 2026-07-28 declaration. currentRatioTTM of 0.985 and workingCapitalTTM of −$465M look alarming and are correct — they reflect settlement payables and litigation accruals in current liabilities at a business with $23B of annual operating cash flow. priceToBookRatioTTM of 20.1x is arithmetically right and economically meaningless for a company with negative tangible book value; we do not use it.

Non-equity tripwire: checked and clear. V is class A common stock on the NYSE, beta 0.754, average volume 8.2M shares, a 52-week closing band of $295.52–$369.59 (a 25.1% range), and a growing rather than fixed dividend. Nothing resembling a preferred or baby-bond profile. (Note that Visa does have series A, B and C preferred stock outstanding under the retrospective responsibility plans — those are different, unlisted instruments discussed in Section 2, not the security this dive covers.)

9. Technicals — the cleanest chart in the batch, and that is the problem

• Price $369.59, up 1.07% on the day, on volume of 7.4M against an 8.2M average — an ordinary session that happened to print a record close.
• Exactly at the 52-week closing high. tech.hi52 is $369.59 and tech.pct_from_hi is 0.0. The stock is +25.1% above the 52-week closing low of $295.52.
• Maximum drawdown from peak over the trailing twelve months: −1.0%. That is the single most striking technical fact on this page. Over an entire year, the largest peak-to-trough decline on closing prices was one percent. There has been no correction to buy, at any point, for a year.
• Above both moving averages: +8.1% over the 50-day at $341.91 and +11.8% over the 200-day at $330.51, both rising.
• RSI 63.98 · MACD +7.09. Momentum is positive and extended but not overbought — RSI has roughly six points of headroom before the conventional 70 threshold.
• Relative performance, and the paradox:

• How a stock can sit at its 52-week high and still be fifteen points behind the index. Because it never fell. A year of low-volatility grinding produces a record close and a modest return simultaneously; the index produced a larger return with larger drawdowns along the way. The last three months are where the relative performance turned: +13.1% against SPY's +7.6%, meaning essentially all of the twelve-month gain arrived in the last quarter. Visa was flat-to-down relative to the market for nine months and then re-rated in three.
• Practical read for entry. There is no technical support to lean on within 8%. The 50-day at $341.91 is 7.5% below; the 200-day at $330.51 — which coincides both with the company's own average repurchase price over the last nine months (~$330) and with the price two knowledge-base sources called attractive in February — is 10.6% below. An investor initiating here is paying up for a chart with no recent risk, which is precisely when the risk is least visible. An investor who owns it has no technical reason to sell: nothing about a 1% maximum drawdown and an RSI of 64 says exit.
• Beta 0.754. This is a low-beta name; the three-month burst of +13.1% against SPY's +7.6% is unusual for it and is more likely mean-reverting than trend-confirming.

10. Insiders — routine option exercises, and one number worth pausing on

The dominant pattern is mechanical and should not be over-read. Every acquisition in this table is an option exercise at $109.82 — a strike set years ago — immediately followed in most cases by a sale at market. That is the standard exercise-and-sell of expiring in-the-money options and it carries almost no signal. Three separate officers exercised at the identical $109.82 strike within a month, which strongly suggests a common grant approaching expiry.

Two things are worth noting anyway.

First, one sale was not a matched exercise-and-sell. Kelly Mahon Tullier sold 19,991 shares at $364.81 on 2026-07-30 beyond the 37,281 shares that came from the option exercise, reducing her holding from 69,653 to 49,662 — a 28.7% cut to an already-reduced position, on the day after the earnings release. That is a discretionary disposal of long-held stock, not tax-driven mechanics. It is one officer, and it is a real sale near the high.

Second, the chief executive exercised and did NOT sell. Ryan McInerney's 2026-06-29 exercise of 20,970 options at $109.82 appears in the file with no corresponding disposal, taking his holding to 36,144 shares. He converted paper options into owned stock and kept it. Of the eight transactions in this table, that is the only one that increases economic exposure, and it belongs to the chief executive.

Set against the company's own behaviour, the aggregate reads neutral-to-positive. Visa repurchased $16.5B of its own stock at an average of about $330 over the nine months to June — 10.7% below today's close — and has $28.4B of authorisation left. That is a far larger and more informative signal than any officer's Form 4. A company still buying at these levels and a chief executive holding after exercise is a mildly constructive read; one discretionary sale near the high is noise. We do not weight this section heavily.

Governance note, from the 8-K filed 2026-07-15: the board amended the bylaws effective 2026-07-14 to designate the Delaware Court of Chancery as exclusive forum for certain actions against the company and the federal district courts for Securities Act claims. This is a common defensive amendment across large-cap issuers; it modestly narrows shareholder litigation options and is noted for completeness rather than as a signal.

11. Verdict, kill-criteria and flip conditions

Hold.

Visa is among the finest businesses in the world and today's price does not offer enough for a new position.

What is genuinely excellent. Net revenue accelerating from +11.3% (FY2025) to +14% in the June quarter and +15% over nine months. Value-added services growing 33% to $3.8B in a single quarter and running above $15B annualised. Cross-border volume +14%, processed transactions +10%, payments volume +11%. Gross margin 80.2%, operating margin 60.7%, net margin 50.8%, return on invested capital 34.2%. Capital expenditure of 3.5% of revenue and 93% free-cash-flow conversion. Net debt of $3.2B against $22.6B of net income and 23.2x interest coverage. $16.5B of buyback in nine months with $28.4B remaining, plus a dividend raised to $0.670 quarterly — a 3.9% total shareholder yield. And the retrospective responsibility plan, which quietly retires as-converted shares at bank holders' expense every time Visa pays a litigation bill, four times as efficiently since the May 2026 exchange offer. Not one of 62 covering analysts has a sell rating.

What holds it at Hold. The stock closed exactly at its 52-week closing high, with a maximum twelve-month drawdown of 1.0% — there is no discount available and no support within 8%. The honest multiple is ~26.4x GAAP-equivalent fiscal-2027 earnings, not the 24.6x the non-GAAP estimate line implies, because the consensus excludes a litigation provision Visa has recorded every year for two decades ($2.2B in FY2025, $1.1B in nine months of FY2026). Client incentives have risen from 23.4% to 28.3% of gross revenue in five years and rose again this year. United States net revenue grew only 5.8% last fiscal year. The European claim docket is expanding, with over 100 merchants outstanding, new UK High Court filings in April, May and June 2026 reaching back six years, and a live motion to declare the forward-looking release in the US settlement invalid and unenforceable. And the most skilled voice in our knowledge base argues, twice, that the entire fee-on-a-fee model is a stablecoin casualty — an objection nothing in the current numbers can test.

Base fair value of $398 is 7.7% above spot, against an 18.8% bear case and a street consensus of $416.88. That asymmetry is 1.16:1. It is not enough at a record high.

Pre-registered KILL criteria — what would move this to Avoid:

• The client-incentive rate rising above ~30% of gross revenue in any fiscal year (28.3% in FY2025, rising 60-90 basis points a year). A 200-basis-point step removes roughly $1.1B of net revenue and would signal that the duopoly's issuer-side competition has broken discipline.
• Success for the merchants' motion to declare the forward-looking release invalid and unenforceable (filed 2026-04-21 in MDL 1720). That reopens a settlement architecture a decade in the making and puts the $39.4B claim pool back in play.
• A US statutory interchange cap or routing mandate. Our knowledge base records a 6% single-day fall on cap fears alone; enactment is a different order of magnitude.
• Nominal cross-border volume growth falling below ~7% for two consecutive quarters (currently +14%). Cross-border is roughly a tenth of volume and more than a third of revenue; it is the highest-yield, most disintermediable piece.
• Stablecoin or account-to-account settlement volume visibly displacing card volume in any large corridor — the falsifier for the structural bear case. Visa's own stablecoin settlement run-rate ($2.5B annualised at 2025-09-30) growing while card cross-border decelerates would be the specific pattern.
• The escrow mechanism running out of headroom — as-converted class B-3 stood at 90.6 million shares (roughly $33.5B at $369.59) against a $39.4B claim pool. If deposits ever exhaust the class B value, class A starts absorbing the litigation directly.

Pre-registered FLIP TO BUY (upsize) criteria:

• A price at or below ~$335 — roughly 22x FY2027E, the 200-day moving average of $330.51, Visa's own nine-month average repurchase price of about $330, and the level two knowledge-base sources independently called attractive in February 2026. Four independent processes converge on that number; a retest is a buy.
• Final approval of the Injunctive Relief Class settlement (moved 2026-07-15) combined with dismissal of the release challenge — which would let the market start treating the litigation provision as terminal and collapse the 8% GAAP/non-GAAP gap.
• A successive class B exchange offer, which the 10-K says becomes available once the claim pool falls 50% from $49.6B. It is at $39.4B — 20.6% of the way. This would be direct evidence the legal tail is closing.
• Value-added services passing 30% of net revenue while still compounding above 30%, which would change the growth algorithm from "commerce plus price" to something faster.
• The incentive rate flattening or falling for two consecutive fiscal years — the single cleanest evidence that the duopoly's pricing discipline holds on both sides of the network.

Visa versus Mastercard — the honest comparison, and we prefer Mastercard. These are the two halves of the same duopoly and a reader deserves a direct answer. On the numbers as of 2026-08-04:

The two names trade at effectively the same forward multiple — 24.6x versus 24.8x — while Mastercard's consensus grows earnings roughly 240 basis points a year faster and its revenue roughly 180 basis points faster. Mastercard is also 4.7% below its own 52-week high rather than exactly at it, and its GAAP-to-non-GAAP gap is less than half of Visa's (3.2% against 8.0% on the trailing four quarters), meaning less of its reported profitability depends on excluding recurring legal costs. On price-for-growth, Mastercard is the better of the two here.

Where Visa is the safer of the two, and it genuinely is. Visa's balance sheet is real — $37.9B of equity against Mastercard's $7.7B, which is so depleted by buybacks that price-to-book (89.8x) carries no information at all. Visa's net debt is $3.2B against Mastercard's $8.1B. Visa's litigation is uniquely pre-funded: the retrospective responsibility plan means US bank holders absorb interchange costs through conversion-rate cuts, a protection Mastercard has no equivalent of. And Visa is the larger network with the deeper cross-border franchise. If the question is "which do I want to own through a bad decade," the answer is Visa. If the question is "which do I want to buy on 2026-08-04," the answer is Mastercard — and neither is a compelling buy at these prices.

Where V fits in the Synthos Framework Portfolio. The quality-compounder sleeve, as a held position rather than a new one. If already owned, a 2-3% weight is defensible on business quality alone and there is no reason to trim into strength. If not owned, this is a name to accumulate on weakness toward $330-345 — the 200-day moving average, the company's own repurchase level, and the knowledge base's stated attractive price all sit there — rather than to initiate at a record close. Note the house standing rule that mega-cap incumbents are guilty until proven innocent on a forward basis: Visa proves its innocence on growth quality (revenue accelerating to +14%, value-added services +33%) and on capital returns (3.9% shareholder yield), and fails it on entry price and on exponential potential. It is a 4 on exponential potential for a reason: margins are already at the ceiling.

Single biggest risk: structural disintermediation. Not litigation — litigation is expensive, chronic, partly pre-funded by the class B mechanism, and currently resolving. The real risk is the one Visa's own 10-K names in full: stablecoins under the GENIUS Act framework, account-to-account rails, FedNow, the European Payments Initiative's Wero, a prospective digital euro, national schemes in Brazil, India, Nigeria and South Africa, and closed-loop ecosystems. The highest-skill source in our knowledge base concludes on that basis that payment middlemen become "defensive utilities, not growth businesses." Visa's counter is that it will carry the new rails itself, and its stablecoin settlement run-rate of $2.5B annualised is a start against roughly $16 trillion of payments volume. We do not know which way this resolves, and neither does anyone quoting a 24.6x multiple.

Most fragile assumption in the price: that the litigation provision is an adjustment rather than a cost. Every forward multiple quoted on this company excludes it. Visa has recorded it in size every year for two decades — $2.2B in fiscal 2025, $1.1B in the first nine months of fiscal 2026 — and the European docket is expanding while the American one closes. Strip the adjustment and the same price is 26.4x fiscal-2027 earnings for a 13.6% grower. That is not expensive; it is simply not the bargain the screens describe.

Provenance & disclosures

• Traceability: 18 true entity-level knowledge-base claims name Visa Inc. — from 30 raw hits across all 51,928 records on the case-insensitive pattern \bVisa\b. Of the 19 hits the pipeline tagged entity-level, 1 was discarded as an immigration false positive (2025-09-27, all_in, tagged ents=['H-1B visa program'], about a $100K H-1B fee). Of the 11 non-entity mentions, 2 more were discarded as immigration false positives (2025-09-22 andreas_steno on the visa fee and the dollar; 2025-07-23 all_in on overseas visas in big-tech hiring), leaving 9 genuine company mentions retained as weak, unweighted context. One further claim (2023-11-29) is excluded outright under the quarantined-misattribution policy. Total discards: 4. Stance mix on the 18 retained claims: 11 bullish / 3 neutral / 4 bearish; breadth 7 distinct sources; range 2022-02-09 to 2026-02-27; net conviction positive. Five of the eleven bullish claims share a single date (2026-02-12) and a single source — that is one conversation, not a consensus, and the conviction rating is Moderate for that reason. No management-lane (skill 0.5) claims appear in this file; the house half-weighting rule was applied to management framing wherever company guidance is quoted from the filings. All quoted claims are verbatim from the stored claim text.
• Data as-of: fundamentals — income statement 2026-06-30 (fiscal Q3 2026, reported 2026-07-28, 10-Q filed 2026-07-29, six days before this dive); annual balance sheet and cash flow 2025-09-30 (fiscal 2025, 10-K filed 2025-11-06) with material items superseded by the 10-Q as documented in Section 8 · estimates 2026-08-04 · prices 2026-08-04 ($369.59; 50-DMA $341.91; 200-DMA $330.51; RSI 64.0; MACD +7.09) · knowledge-base claims 2026-08-04. Every figure comes from the Synthos vendor data file for V or from the SEC filings in filings/V/; no figure comes from memory, recall or external retrieval.
• Filings consulted (10 targeted searches across three documents): 10-K filed 2025-11-06 (fiscal year ended 2025-09-30) — searched for the debt footnote, the escrow and conversion mechanism, the share-repurchase disclosure, the interchange-claim pool, the competition and stablecoin risk factors, and the value-added-services series; 10-Q filed 2026-04-29; 10-Q filed 2026-07-29 (the primary source for this dive) — searched for volume and transaction growth, net-revenue components, operating-expense drivers, the debt note, stockholders' equity and the exchange offer, the escrow rollforward, legal matters, liquidity and risk factors; 8-K filed 2026-06-26 (the $250M escrow deposit and the exact conversion-rate adjustments); 8-K filed 2026-07-15 (bylaw amendment, exclusive forum); 8-K filed 2026-07-28 (fiscal Q3 results and the $0.670 dividend declaration).
• Filing-versus-vendor overrides (three, all resolved in favour of the filing): (1) fiscal 2025 buybacks — vendor cash flow $13.389B, 10-K "54 million shares... for $18.2 billion"; we use $18.2B; (2) value-added services revenue — absent from the vendor file entirely, disclosed by the 10-Q at $3.8B in the June quarter (+33%) and $10.3B over nine months (+32%), and by the 10-K at $7.2B/$8.8B/$10.9B for FY2023–FY2025; we use the filing; (3) share count and the class B/C conversion machinery — the vendor records none of the May 2026 exchange offer, the 2026-06-24 escrow deposit and its ~750,000-share retirement, or the Europe preferred conversion-rate reductions, and its implied share count of 1.867B is 3.4% below the 1.933B as-converted diluted count Visa uses for EPS; we use the filing's count and quote no market-cap-derived multiple. One vendor figure was positively confirmed by the filing: total debt of $25.171B against the 10-K's "carrying value... of debt was $25.2 billion."
• Other data rejections (detailed in Section 8): netDebt corrected from $5.017B to $3.184B (omits $1.833B of short-term investments); enterpriseValueTTM carries a +1.2% residual against a rebuild and a ~2.2% understatement against the correct share count, and no EV multiple is quoted; all est niAvg rows rejected as implying a rising share count (1.907B → 1.966B) and implausible net margins (55.0% → 61.0%); the FY2030 estimate row rejected outright for showing EPS falling to $19.160 from $19.597 while revenue rises; est EBITDA/EBIT rows pass the sanity check but are non-GAAP and are used for nothing; the FY2025 cash-flow netChgCash of $5.224B rejected as irreconcilable with the $8.179B change in cash on the vendor's own balance sheets; inc_q interest expense of $566M in the March 2026 quarter flagged as unexplained; inc_q gross margin of 76.5% in the June quarter treated as an allocation artefact and operating margin used instead; closing-price 52-week extremes (tech) preferred over intraday (quote); tech.sma50 preferred over quote.sma50; priceToBookRatioTTM of 20.1x noted as meaningless given negative tangible book value.
• Extraction limitation, stated plainly: the SEC text extractions preserve narrative prose but drop numeric table cells. The volume tables, net-revenue-component table, balance sheet, operating-expense table, share-repurchase table and earnings-per-share tables appear as headings and footnotes with the numbers absent. Every filing figure in this dive comes from narrative sentences, not tables — hence percentage growth rates for volume rather than dollar volumes.
• Estimate coverage: FY2026 23 analysts on EPS and 27 on revenue; FY2027 26 on EPS and 28 on revenue — the primary forward anchor and the deepest coverage in the block; FY2028 15 and 26; FY2029 9 and 13; FY2030 6 and 13. No conclusion in this dive rests on the FY2029 or FY2030 rows, and the FY2030 row is rejected outright.
• Peer note: the vendor peer set is ALLY, AXP, BAC, JPM, MA, PYPL, SEZL, SLM. Four (Ally, Bank of America, JPMorgan, SLM) are balance-sheet lenders whose earnings are net interest income and whose risk is credit risk — not comparable to a network that takes no credit risk, and no inference is drawn from them. American Express is a closed loop carrying card-member receivables and is adjacent rather than comparable. Mastercard is the only true comparable, and it is compared directly in Section 11. No peer multiples are supplied in the file, so no peer-multiple table is drawn.
• Fair-value caveat: the $300 / $398 / $450 anchors are multiples of FY2027 consensus EPS of $15.005 (20.0x / 26.5x / 30.0x), cross-checked against FY2028E of $17.041 and against a GAAP-equivalent FY2027 EPS of roughly $14.00. Stated arithmetic, not a discounted cash flow. The base case requires roughly two turns of multiple expansion off today's 24.6x; if the multiple simply holds, fair value is $369 — today's price. The bull case requires 30x and is flagged as the fragile leg.
• Timing: fiscal Q3 2026 was reported 2026-07-28 and the 10-Q filed 2026-07-29, six days before this dive, so the most recent quarter is fully reflected. The next print is 2026-10-27, 84 days away — no earnings banner applies, and this dive states that explicitly rather than leaving it inferred. Note that Visa's fiscal year ends 30 September, so every estimate row dated 2026-09-30 through 2030-09-30 is a September fiscal year and is labelled as such throughout.
• Accessibility note: all tables are labelled, all emphasis is carried by text rather than colour, and every figure is stated in full rather than encoded in a chart.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.