PayPal: a modest recovery worth buying at $56

The opportunity is in better transaction economics

Branded checkout, Venmo, merchant processing and credit share capabilities without earning identical economics. Familiarity may help the wallet win a consumer decision. A merchant may instead buy a more efficient guest checkout or processing service. The relevant unit is the payment work completed and the contribution left after costs. Volume passing through a recognizable brand leaves that contribution unresolved.

The Q2 reconciliation defines transaction-margin dollars before support, marketing, technology, administration and restructuring. It measures contribution before the platform’s remaining expenses. I first deduct processing and credit costs, then ask whether what remains is growing fast enough to fund the continuing system. Both steps must work for payment-volume growth to create operating leverage.

There are encouraging signs of deeper use. In Q2, currency-neutral branded online checkout volume grew 2%; Venmo volume grew 14%. Monthly active accounts reached 228 million, growing 1%, while transactions per active account excluding unbranded processing rose 7% on a trailing basis. Those figures suggest engagement can improve faster than the overall account count. Q2 investor presentation

Fastlane documentation describes recognition by email, saved details and password-less guest checkout. The consumer explanation says its profile exists outside a PayPal account. That creates a positive route beyond wallet-account growth: PayPal can improve a merchant’s checkout while the shopper uses a card. The capability leaves incremental conversion, revenue classification, and unit contribution unmeasured.

I would therefore keep two tests. Branded checkout and consumer engagement measure the wallet-recovery hypothesis. Merchant contribution after service and acquisition costs measures the infrastructure route. Success in the latter can matter without proving the former, and a volume gain can migrate toward a less valuable mix. Venmo activity and overlapping branded categories should not be summed into an invented new revenue base.

Revenue growth has outpaced contribution

The Q2 arithmetic is more informative than its headline growth rate. Revenue rose from $8.288 billion to $8.682 billion. Yet transaction-margin dollars increased only from $3.844 billion to $3.900 billion. Non-transaction expenses on the company’s adjusted basis rose faster, reducing adjusted operating income. Q2 earnings release

Only about 14 cents of the additional revenue dollar appeared as incremental transaction margin across these two quarters. The aggregate comparison combines changes in the existing base, mix, processing and losses. It cannot isolate the margin caused by a new customer or product. It shows why revenue cannot settle the recovery question. The expense budget then absorbed more than the contribution increase.

Excluding customer-interest effects gives a clearer view of payment work: contribution rose $93 million, offset by a $37 million fall in interest, leaving the $56 million headline increase. The interest is real income, but the rate environment can change it without a corresponding change in checkout usefulness. I would judge operating recovery on the ex-interest contribution and report the income on its own line in the complete financial account.

The company’s 2026 plan still involves investment. It guides to approximately $15.6 billion of transaction-margin dollars and 7–8% growth in adjusted non-transaction expenses. Applying that midpoint to the previous annual cost base implies about $5.83 billion of adjusted operating income, versus $6.378 billion in 2025. The announced $1.5 billion gross saving target over two to three years cannot be added directly to future earnings; some savings fund new spending. July IR newsletter

The recovery case is investable at the dated price if product improvements eventually produce more contribution than the added cost of delivering it. Faster guest checkout could help a merchant relationship; more frequent wallet use could help another cohort. Neither can be assumed to pay merely because it reduces friction. The full budget has to show where contribution improves and when the investment burden stops outrunning it.

The quarter also sets a near-term hurdle. Adjusted non-transaction costs rose $193 million, while transaction-margin dollars rose $56 million. Holding that expense change fixed, PayPal would have needed another $137 million of transaction margin merely to keep adjusted operating income level. Better volume or revenue is helpful only to the extent it closes that contribution gap.

I would use that bridge when evaluating the recovery: isolate customer-interest effects, identify how much contribution comes from payment economics and losses, then compare it with the cost of delivering the platform. The decomposition cannot identify one product’s marginal profitability from consolidated accounts. It can stop a favorable volume headline or a lower share count from answering a question the operating statement has left open.

The cash I am willing to value

Reported free cash flow omits some cash the business asks shareholders to provide elsewhere. Employee equity carries an economic cost, and lending held for investment uses cash through investing activities. Charging both gives a stricter starting point for the proposed net share retirement. The cash-after-investment proxy deducts lending reinvestment conservatively. Loan principal remains an asset, and the following year’s lending requirement can differ.

The latest full trailing year is built from FY2025 plus the first half of 2026 minus the first half of 2025. It produced $34.128 billion of revenue and $4.902 billion of GAAP net income. I use that full-year basis to avoid extrapolating a strong quarter of cash collection. FY2025 filing, June 2026 filing

The compensation deductions value the economic burden of employee equity; the deductions do not describe immediate cash payments. Charging the current capitalized amount also recognizes compensation used to develop software. The loan deduction recognizes cash deployed through investing activities into loans held for investment. The principal remains an asset that can be repaid; funding it is not the same as recognizing a loan loss. But the cash committed to that asset cannot also be distributed to shareholders. The deduction measures what was left after making that commitment.

At the entry and the 855.461 million shares disclosed for July 22, this proxy yields approximately 10.02% on a $47.949 billion equity capitalization. I consider this the starting valuation because the cash measure reflects more of what the business asks shareholders to fund.

The larger cash headline needs reconciliation

The adjusted-FCF bridge removes the timing between held-for-sale originations and subsequent sales, producing $7.926 billion trailing adjusted FCF versus $6.586 billion reported. P6 P7 That can improve comparability when the sale cycle is intact. It cannot make sale pricing and access irrelevant. The Q2 filing measures some held-for-sale loans using expected sale prices with unobservable inputs. The timing adjustment improves comparability, with sale price and access still part of the credit risk.

The latest FY2026 adjusted-FCF guidance is $6 billion-plus. I do not make the larger trailing number permanent in the forecast. My stricter $4.804 billion anchor helps the long survive a less flattering cash interpretation.

A smaller loan book can also release cash while reducing future earning opportunity. That could release capital even while the recurring payment business remains unchanged. I keep the $653 million historical funding charge visible so the forecast can be reconciled as balances held for investment and sales change. No single cash headline can tell whether the improvement came from a better payment franchise or the release of an asset.

The balance sheet also has distinct cash blocks. Liquid corporate cash and investments, excluding strategic investments, total $13.530 billion against $13.400 billion of debt. Customer assets belong to customers. The $13.667 billion customer-cash component in the cash-flow reconciliation is not additional shareholder liquidity, and even corporate balances support operations and regulation.

The filing permits designated European customer balances to fund credit within its Luxembourg banking subsidiary. The approved amount was $2 billion at June 30, with another $500 million approved in July; exceptional circumstances could call on corporate liquidity. These obligations explain why a large displayed cash balance is not automatically an excess-cash bonus. I add no such bonus to the valuation. Liquidity and credit-funding disclosures

A modest recovery can fund meaningful net retirement

My base case grows consolidated revenue by 4% each year, from $34.128 billion to a $36.913 billion annual run rate near the horizon. Cash after compensation and reinvestment in loans held for investment reaches 14.5% of sales. The observed proxy is approximately 14.1%, so this assumes a modest improvement in conversion.

The result is $5.146 billion of annual cash in the first model year and $5.352 billion in the second. I budget another $300 million annually for transformation and liquidity. I also assume the current $0.14 quarterly dividend persists, subject to future declarations.

Assumed average repurchase prices of $65 and $75 support 5% net retirement per year, or 9.75% across both years. Compensation replacement is charged before allocating the net-purchase budget; I am not calling every dollar of gross repurchases value creation.

The $6 billion repurchase guide is not an independent operating result. The model’s more restrained budget funds compensation, dividends, transformation and liquidity before net retirement, leaving some cash uncredited at the horizon. If product development or credit capital requires more, shares must decline more slowly. Borrowing to preserve a repurchase target would change the risk without proving recovery.

My $85 forecast and the valuation it requires

At the base case’s ending share count, $5.352 billion of annual cash becomes $6.93 per share. Twelve times that amount gives $83.19 in the model. I register a rounded $85 central forecast.

A 14% net-margin check produces about $6.69 of earnings per share. The exact cash valuation is approximately 12.4 times those earnings. This checks the reasonableness of the same business valuation; I do not add the two estimates.

The model’s 48.4% price gain and rounded forecast’s roughly 52% require both better cash per share and a higher multiple. Keeping the entry multiple gives approximately $69.19, or 23.5%. That is still prospective progress, but it is a different return. Sustained contribution and disciplined capital use must give future buyers a reason to move from roughly ten to twelve times cash.

The reverse hurdle quantifies the cushion. With base retirement and a twelve-times multiple, $4.363 billion of annual cash would support a 10% annualized price return from $56.05. That is roughly 9.2% below the observed cash proxy. Lower valuation multiples or slower share retirement would reduce that cushion.

Under base operating assumptions, nine times cash gives about $62.39 and fifteen times gives $103.99. The operating recovery is unchanged; the return depends on what future buyers pay for it.

The contrary case deserves numbers

The bear case represents deteriorating economics alongside weaker demand. Annual cash falls to about $3.442 billion, retirement slows and the multiple contracts. A rounded $35 outcome would mean substantial loss despite continuing repurchases and the assumed dividend.

The bull case needs better monetization and cost productivity to raise cash to $6.135 billion. Six-percent retirement remains cash-funded, and the multiple reaches fourteen. Its roughly $115 rounded outcome represents a successful recovery in this scenario.

The serious bear case is a company that processes more commerce while keeping less value and continually reinvesting to stay relevant. Guest checkout could become useful infrastructure yet earn too little after processing, service and product costs. Credit might require more capital at the same time. The positive countercase is that those capabilities reuse an established network and produce contribution at low incremental acquisition cost. Route-specific paid economics, rather than prominence of the wallet button, decide between them.

What would make me keep or change the view

The original Q2 2028 tests remain explicit: trailing ex-interest contribution above $16 billion, branded volume growth of at least 4% currency-neutral for two quarters, monthly active-account growth above 2% and strict cash above $5 billion. The account and branded measures test the consumer component of the pitch. Fastlane success would need its own contribution evidence; it cannot be inferred from those account counts or used to rename them.

Two quarters of negative branded growth, trailing ex-interest contribution below $14 billion or strict cash below $4 billion would materially weaken the dated long. A stronger merchant-only result could change the mix of the thesis, but it would have to be demonstrated in contribution and cash. Debt-funded capital return alongside deterioration would worsen the interpretation. A rising EPS needs stronger cash economics to rescue the case.

The catalysts are contribution and expenses moving in the right relationship, with lending held for investment and loan sales reconciled to the cash available for owners. The unchanged-dividend assumption adds $1.12 over eight quarters per original share; actual distributions will be evaluated separately from price. I would own a PayPal that earns more by making payments work better, whether visible to the shopper or embedded behind the merchant. The bargain is attractive when that usefulness leaves cash after the system is funded.

My long thesis and dated forecast

I favor PayPal at the dated entry because modest growth and existing cash can make a recovery valuable after the funding bill. I keep the $85 October 12, 2028 forecast against SPY. More ex-interest contribution and strict cash above $5 billion would earn the case. Product adoption must leave cash after the system is funded.

Assumptions

• USD millions and millions of shares unless stated; prices are USD per original share.
• TTM = annual result + current comparable interim − prior interim. Forward cash margins include economic compensation and company-specific recurring reinvestment.
• Net share reduction is cash-budget constrained. Unspent/protected cash has no terminal bonus; gross buybacks are not added to price returns. Cases are not probabilities.