V Deep Dive: A great network, and a price that already assumes it stays one

The question, and our answer

The rating is a research judgment about risk and reward at one price. It is not an instruction to buy or sell anything, it names no price target, and it would change with the price even if nothing about Visa changed.

Visa reported fiscal 2026 net revenue of $45.9 billion, up 14.8% from $40.0 billion the year before, at a 61.2% GAAP operating margin, and the shares closed at $425.51 on 11-02-2026, 18.4% above their close at the end of the fiscal year. The question for an investor is not whether Visa is a good business. It is plainly one of the best-run toll roads in commerce. The question is whether a price of 32.1 times fiscal 2026 non-GAAP earnings leaves room for the things that can go wrong: a growth rate whose pricing component cannot be separated from volume and value-added services, litigation provisions in each of the last five fiscal years with the last two the largest, a Department of Justice antitrust case whose status after its 06-23-2025 ruling is unknown to us, and several credible routes by which payments could move around the network.

Our answer is that it does not leave much room. Built from the company's own reported economics and a driver-based forecast, our reference case values an as-converted class A share at $365.04, which is 14.2% below the price. That number deserves a caveat that we put up front rather than in an appendix. The reference is an illustrative reference case: one fully specified set of assumptions that gives every other reading something to be measured against. It is not our estimate of what the shares are worth, and it is not a floor. Its operating path rests on four years of reported results, but its value also rests on two judgments that nothing we hold can settle, the cost of capital and an abrupt slowdown in growth after fiscal 2031, so we show what happens when each is moved. A lower cost of capital alone gives $399.88, 6.0% below the price. A five-year fade of growth alone gives $389.79, 8.4% below. Both together give $427.57, 0.5% above the price. Two further central readings, an exit-multiple check on the same forecast and a capitalization of fiscal 2026 owner earnings built from different inputs, give $406.27 and $332.72, which are 4.5% and 21.8% below the price. Our rating is 3 of 5, BALANCED, and we say plainly that the reference value itself does not reach the price and that the range around it is lopsided, with the downside case 44.2% below the price and the upside case only 8.7% above it. The rating rests on what the gap is made of. It comes mostly from the two judgments we cannot yet source and not from Visa's operating forecast, and we weight those judgments so that the price broadly reflects the opportunity and the risks. The closest competing reading is a 2 of 5, and we set it out in full.

A rating of 3 of 5 is not a forecast that Visa shares rise or fall, and it is not a verdict on the business, which is an exceptional network. It says that at $425.51 the price broadly reflects what we can value and the risks we can name, while the value we compute at our own settings sits below the price and the gap rests on inputs we have not been able to source. The rest of this report shows the work.

*Closes are the regular-session closing prices of the sessions shown. The sessions to 11-06-2026 come from market data retrieved 11-07-2026, and the November 9, 2026 regular-session closing price was observed after that session settled and retrieved 11-09-2026. The Treasury yields are the official par yields for the dates shown, retrieved 11-07-2026. The three case values are the accepted values of 11-02-2026 in both columns. Blockhaven calculations.*

What moved is the price, by less than one percent since the valuation date, and with it the distance to each case by less than a point: the reference case is 13.5% below the latest settled close against 14.2% below the valuation close, the downside 43.8% below against 44.2%, and the upside 9.5% above against 8.7%. The close sits where it sat at the valuation date, inside the range of the three cases, above the reference and below the upside. The ten-year Treasury par yield, the risk-free input to our cost of capital, was 4.91% on 11-02-2026, which is the rate the model uses, and 4.86% on 11-06-2026, the last date in the Treasury table we hold; we have no official yield for 11-09-2026 and we have not rebuilt the rate. For scale, the stored sensitivities give $405.38 for a cost of capital half a point lower and $331.97 for half a point higher, so five basis points is a tenth of that step and, other things equal, would lift the value by a small fraction of the distance between the reference case and the price.

The other readings that this report sets against the price keep their order as well. Each is a value of 11-02-2026, measured here against the later close.

The reverse valuation moves with the price by about as little. Solved with the same routine as the break-evens of the valuation date, and with every other reference assumption held, the reference matches the latest settled close at a cost of capital of 8.38% (8.34% at the valuation close), at a terminal growth of 4.78% (4.82%), or with 3.78 points more volume growth in every forecast year (3.98). These are conditional break-evens, not what investors believe, and the model's own cost of capital stays at 9.06%.

What did not move is everything else. No forecast, input or value has been changed, the share count is the one at 11-02-2026, and the cost of capital, the path of growth after fiscal 2031 and the legal position are treated exactly as before. The comparison therefore does not change our reading. The rating is still 3 of 5, BALANCED, assessed at the $425.51 close of 11-02-2026, and the relation behind it is the same at the later close: the reference value does not reach the price, only the upside case sits above it, and the two relaxations together, which are labeled sensitivities and not adopted, are above it. A lower price makes the reference case look slightly less far from the price, and a lower yield points the same way, and neither is close to the size of the judgments the rating rests on, which are described below. The annual report for fiscal 2026 arrived on 11-05-2026, and what it confirmed and what it left open is set out where it bears on the analysis and in the closing section.

The second is how growth ends. The reference has net revenue growing 8.4% in fiscal 2031 and then stepping straight down to 4.0% from fiscal 2032, a cliff for a network that grew 10.0% to 14.8% a year over the last three years. We tested fades that start from the fiscal 2031 growth rate, hold that year's margin, incentive, capital-expenditure and working-capital ratios, and fall in equal steps to 4.0%. A fade of three years gives $377.56, one of five years (7.5%, 6.6%, 5.8%, 4.9% and 4.0%) gives $389.79, and one of seven years gives $401.76. We have not put any of them in the reference, because nothing we hold says how long growth stays above 4.0% after fiscal 2031. The annual report states management's view of an addressable opportunity that is a multiple of Visa's current volume, which says the runway is large but not when it ends, and the pace at which growth slows inside our own forecast cannot choose a length: read one way it implies about six years, read another about thirty. A fade is also not free, because growth above the terminal rate has to be funded. If new capital earned only Visa's fiscal 2026 return on total capital of 56.5%, measured on our owner-basis after-tax operating income with its assumed tax rate and legal cost, so that growth carried reinvestment, the same three fades would give $356.98, $367.36 and $377.52, and the immediate step to 4.0% with that reinvestment gives $346.35, 18.6% below the price. A slower path of equal standing, five years at 3.0% as might follow a U.S. routing or interchange rule change, gives $352.20, 17.2% below. We read these as showing that the step to 4.0% is a placeholder and not a forecast, which is why we print the fades as sensitivities and keep the reference where it is.

Relaxing the cost of capital and adding the five-year fade together gives $427.57, 0.5% above the price. The mirror image, a beta of 1.00 and a premium of 5.00% with five years of 3.0% growth, gives $309.23, 27.3% below it. Neither is a new reference; they bracket the region where a reasonable value could sit, which is a wide region, and the price falls inside it.

The exit-multiple check and the owner-earnings reading do not rest on a relaxed rate or a fade. The independent reading is built from different objects. Starting from fiscal 2026 owner earnings of $24.9 billion, it grows them on one path that fades over ten years from 10% to the 4.0% terminal rate, reinvests according to the return on new capital instead of capital expenditure and depreciation, and discounts at the same 9.06%. It gives $332.72, which is 8.9% below the reference discounted cash flow, with starting growth of 8% and 12% giving $307.79 and $359.51. At the lower 8.62% rate it gives $365.37. This is a different object from the driver forecast, so it can catch an error in it, and it did not find one that favors the price; it is, if anything, more cautious. It remains a lens built on our own judgments, and we do not average it with anything.

The reverse question is how far the reference would have to move to meet the price, holding everything else fixed. The answer is a cost of capital of 8.34%, or terminal growth of 4.82% (just under the 4.91% risk-free rate), or 3.98 percentage points a year added to volume, transaction and cross-border growth in every forecast year. These are conditional break-even combinations. They are not statements about what investors believe or what the price requires. They do show that the price sits at the favorable edge of a plausible rate range, and that a change in the rate alone, with the operating forecast untouched, closes the gap, which is why we say the question between 2 of 5 and 3 of 5 is close.

The other assumptions move the value much less. The yield terms, the pricing and value-added services contribution, range from $347.52 on the downside path to $375.37 on the upside path. The reference's service and data processing yield terms sit inside the range of the yearly differences between each line's growth and its driver's growth in the seven growth years of the eight-year history, fiscal 2020 to 2026, but its service term of 1.0 point from fiscal 2029 is below the last three years, as the forecast section says. A client-incentive ratio held flat at its fiscal 2026 level gives $369.84. Holding the other assumptions fixed, the reference is therefore slightly less favorable than that alternative, by about 1% of value, because it lets the ratio drift up to 29.5% by fiscal 2031. Recurring legal cost of 0.5% or 2.0% of revenue gives $367.82 or $359.49. Holding the cost ratio at the fiscal 2026 level gives $360.06 and following the upside path $372.85. Depreciation held at its fiscal 2026 share of 2.6% of net revenue, the model's basis that leaves out acquired-intangible amortization, gives $361.36, and restructuring that recurred at 0.3% of revenue, $363.38. We looked for something omitted from the bridge and found only one candidate, other assets of $4.4 billion that may include investments whose gains the company excludes; if every dollar were a non-operating investment the reference would rise by $2.39 to $367.43. The split is not disclosed in what we hold, so it stays unknown and we do not add it.

*Effects as recorded by the independent challenge, in dollars per share against the earlier $352.00 reference. The fifth row's settings are points a year added to the yield terms. A cost of capital of 8.21% is the low setting and 9.76% the high one; the low setting adds value.*

The cost of capital is the largest of the eight, and it is one of the two assumptions, rows 1 and 3, on which the challenge judged its disagreement with our setting material. Our beta of 0.90 and premium of 4.75% are judgments with no captured source, and the challenge found outside readings below both, none registered as evidence. A rate of 8.21% adds $73.06 and a rate of 9.76% removes $43.56. We have kept 9.06% as a labeled judgment and show the effect of the alternatives, and the challenge's disagreement is accounted for by the rate grid and the break-even cost of capital of 8.34% rather than resolved, because the evidence that would resolve it has not been captured. Terminal growth was an agreement: 3.5% removes $26.91 and 4.5% adds $32.81, and nothing captured moves the center from 4.0%. The path from fiscal 2031 to the terminal rate was the second material disagreement: five years averaging 3.0% removes $12.28 and a path averaging 5.76% adds $23.67, and our response was to print three, five and seven year fades as sensitivities because nothing in the record chooses a length. Terminal reinvestment was an agreement with a caveat: charging growth at the fiscal 2026 return on total capital, 56.5% on our owner-basis after-tax operating income with its assumed tax rate and legal cost, removes $17.88, and holding reinvestment at zero adds $2.50. The yield terms, the service and data processing terms together, were an agreement: the downside path removes $16.86 and the upside adds $9.94.

The last three assumptions are smaller. The client incentive ratio in fiscal 2031 was judged disagree-immaterial, since continuing the measured trend to 30.46% removes $4.56 and holding the fiscal 2026 ratio adds $4.62, which is about 1% of value in each direction; the reference is therefore slightly less favorable than the flat-incentives alternative, other assumptions fixed. The recurring uncovered legal cost was an agreement, with 0.5% adding $2.68 and 2.0% removing $5.36 and the large-loss stress carried separately in the downside. The adjusted operating cost ratio was an agreement, with the upside path adding $7.51 and holding the fiscal 2026 ratio removing $4.79. Taken together the table says what the section on the rating says in words: five of the eight assumptions are settings where the evidence centers, two are the judgments that carry most of the distance between the reference and the price, and one is small.

The rating, what would change it, and the strongest counterargument

Our Blockhaven Rating is 3 of 5, BALANCED, at the $425.51 close of 11-02-2026, and the facts that weigh against it come first. At our own settings no reading built on evidence we hold reaches the price. The readings that sentence covers are the three central ones: the reference value is $365.04, 14.2% below it, the exit-multiple check at 24 times fiscal 2031 earnings gives $406.27, 4.5% below, and the owner-earnings reading gives $332.72, 21.8% below. The upside case, at $462.33 and 8.7% above the price, and the 30-times exit check on its forecast, at $564.73, are above it, but they are conditional cases built to bound the range and not central readings. The terminal value is 80.9% of the reference enterprise value, so most of what we value lies after fiscal 2031, where our evidence is thinnest. The price sits at the favorable edge of the cost-of-capital range we think reasonable: across betas from 0.80 to 1.10 and premiums from 4.0% to 5.5%, only 3 of the 20 reference values with the immediate step to 4.0% growth are within 2% of the price or above it. The spread of cases is lopsided, with the downside 44.2% below the price and the upside 8.7% above it at our 9.06% rate. And the risks we name on the downside are real ones: an enduring impairment of the network's earning power, the Department of Justice debit case and the hypothetical $5 billion charge we carry for it, the interchange litigation, and the routes by which payments could move around the network, a risk we describe but cannot quantify.

What the rating rests on is the make-up of that gap. It is built mostly from two judgments that nothing we hold can settle, the cost of capital and the path of growth after fiscal 2031, and not from Visa's operating forecast, which is anchored on four years of reported results. With a beta of 0.85 and a premium of 4.50%, a cost of capital of 8.62%, the reference would be $399.88, 6.0% below the price. With the five-year fade alone it would be $389.79, 8.4% below, and with both together $427.57, 0.5% above. These are sensitivities that we print and have not adopted, and the reference stays at $365.04. Holding everything else fixed, the model meets the price at a cost of capital of 8.34% against our 9.06%. The level is therefore a judgment about how much weight those unsourced inputs deserve, and we weight them so that the price broadly reflects the fundamental opportunity and the risks. That is how the Blockhaven scale describes 3 of 5: the base-case economics sit near the relevant hurdle, or meaningful upside is substantially offset by uncertainty or downside. It is not a claim that the price equals the value, because the reference value itself does not reach the price.

The strongest counterargument is that the shares deserve 2 of 5, CAUTIOUS, and it should be read in full. Neither relaxation has evidence behind it in what we hold: we have captured no source for a lower beta or premium, and nothing we hold says how long growth stays above 4.0% after fiscal 2031. Adopting both to reach the price would mean choosing assumptions for where they put the value, which a valuation must not do. Every central reading at our own settings sits below the price, and the asymmetry survives the favorable settings, since the downside is 44.2% below against an upside 8.7% above at our rate, and 39.9% below against 20.3% above at 8.62%, where the upside case rises to $512.01 while the downside improves only to $255.80. On that reading the price asks for strong execution and the relaxations rest on judgment alone, and the case is reinforced by what our reference leaves out: any rule change on U.S. routing or interchange, and incentives drifting up as fast as they have recently. We stay at 3 of 5 because the same absence of evidence cuts both ways, and nothing we hold shows our own rate or the abrupt step to 4.0% to be the right settings either. A reader who weighs the central readings at our own settings more heavily than we do would reasonably land on 2 of 5.

Three things are kept apart in this report. The value we compute at the valuation date is what the model says Visa's cash flows are worth on stated assumptions, $365.04 per share in the reference case inside a range from $237.26 to $462.33. The rating is our judgment of the price against that value and the evidence behind it. How the shares trade over the next twelve months is a third question, and this research does not forecast it. A rating of 3 of 5 is not a forecast that Visa shares rise or fall, and it does not describe the business as anything less than the exceptional network the rest of this report shows.

What would change our view is mostly evidence, and one thing is price. The most useful single disclosure we could name was the pricing share of value-added services growth. The fiscal 2026 annual report, filed 11-05-2026, did not give it, so that question stands, and the next filing that quantifies it would tell us how much of the yield contribution repeats. Visa's fiscal 2027 outlook is known to us only as the qualitative guidance in the results deck of 10-27-2026, which is stated on a different basis from our model and so neither confirms nor contradicts the reference. The detail the company gave only on its earnings call, the expected tax rate and the growth of client incentives, is in no primary document we hold, and it would bear on two of our assumptions. The next steps in the Department of Justice debit case and any new routing or interchange rule would each move assumptions we now hold as judgments. Capturing dated sources for beta and the equity premium would resolve our largest uncertainty directly, and could move the reference by tens of dollars in either direction, and evidence on how long volume and yield growth stay above 4.0% after fiscal 2031 would settle the second judgment. A materially different share price would change the arithmetic of the rating even with no new facts. Catalysts with dates are the dividend record date of 11-10-2026 and payment on 12-01-2026, and the expected completion of the BioCatch acquisition by the end of fiscal 2027's second quarter. The annual report that was due has been filed.

Decision tests, written so that someone can check them later. The first, on the pricing share of fiscal 2026 growth in value-added services, was read against the annual report filed 11-05-2026 and resolved as no quantification, which is evidence neither way and leaves the reference's yield terms as they were; it carries forward to any later filing that gives the share. If pricing and value-added services are disclosed to be a minority of the growth in those lines, the reference's yield terms look cautious; if a court or regulator changes how debit or credit transactions may be routed, or the Department of Justice case moves against Visa, the downside case becomes more relevant; if repurchases are scaled back to match free cash flow, the share count falls more slowly than in the history, and our reference, which credits no repurchases after the 32 days between the balance sheet and the price, does not change.

Rating levels considered

The Blockhaven scale has five levels, from 5 of 5, COMPELLING, where the price is far below the value we can support, through 4 of 5, FAVORABLE, 3 of 5, BALANCED, and 2 of 5, CAUTIOUS, to 1 of 5, POOR, where the price is far above it. We considered the three in the middle and set aside the two ends. A 5 or a 4 would need a price meaningfully below the value, and no central reading we have produced shows that. Every central reading at the model's own settings, the reference discounted cash flow, the 24-times exit check and the owner-earnings reading, is below the price. The two relaxations of the reference together, the centered 8.62% rate with a five-year fade, neither of which has evidence behind it, give a value only 0.5% above the price, and the most favorable of the rate settings we tested with the immediate step to 4.0% growth, $460.91 at a beta of 0.80 and a premium of 4.0%, is 8.3% above it at the far edge of a band we cannot source; with the five-year fade in place of the immediate step, the same beta and premium give $493.77, 16.0% above. The upside case and the 30-times exit check are above the price, by 8.7% and more, but they are conditional cases that bound the range, not central readings, and a rating of 4 or 5 would have to rest on them. A 1 would need the price to sit far above value on the evidence we hold, and even the least favorable central reading, the owner-earnings check at the model's own rate, is 21.8% below the price and not far below it; the downside case, at 44.2% below, is a conditional stress and not a central reading. The real choice is between 3 of 5, BALANCED, and 2 of 5, CAUTIOUS, and the independent challenge's record is what frames it.

The challenge's overall verdict was mixed, and its two arguments are the two readings. Its counter-thesis to the 2 of 5 proposal that was then on the table was that the gap between the price and the reference was built mostly from two modeling choices and not from Visa's economics, a cost of capital whose inputs both sat above the outside observations and a one-year step from 8.4% to 4.0% growth for a network with a record of 10% to 15% growth, and that at a rate centered on those observations the evidence did not clearly separate a price above value from a price near value. Its opposing view was that the downside case and the lopsided range remained, and that neither relaxation had evidence behind it in the register. Its outside view found no approved consensus provider, so no Street estimate enters the work, and its own reading of outside rate evidence was unregistered and used only to test the inputs. The challenge also left one point to us that the later work resolved: the independent lens it asked for, built from different inputs, is the owner-earnings reading that gives $332.72, and it did not rescue the value.

We rate 3 of 5 because the gap between the reference and the price is made mostly of those two unsourced judgments, because the downside range around the reference is wide and a buyer at the price is paid for much of the opportunity, and because the two readings are close enough that the weight on the unsourced inputs decides between them. That is a judgment about weight and it could be made the other way. A reader who gives the central readings at our own settings more weight, all of which are below the price, and who treats the relaxations as choices of convenience would reasonably land on 2 of 5. What would move the level up is dated sources that place the rate or the growth path in the favorable region, and what would move it down is a ruling in the debit case, a routing or interchange rule change, or evidence that the yield terms of fiscal 2026 were mostly price and will not repeat. A different price would move it either way with no new facts.