PayPal, and the Ice Cube That Stopped Melting

We went long PayPal common stock on 23 October 2026 for a three-to-five-year hold. The one-line case is that PayPal is a cash machine priced like a declining business, in the middle of a transformation that is quietly working, with a free option on being the payment rails for AI agents attached for nothing. At about ten times forward earnings with a free cash flow yield in the low teens, you are not paying for the turnaround or the option. You are paying for a melting ice cube, and we do not think it is melting.

The bear case on PayPal is well known and not stupid: branded checkout, the high-margin button that is the heart of the business, has been losing share to Apple Pay and to Shopify's Shop Pay, and if that continues the whole thing slowly erodes. The reason to be long now is that the erosion has stopped getting worse at exactly the moment the market stopped believing it ever would. Branded checkout grew 2% on a currency-neutral basis in the most recent quarter, the second straight quarter of stabilisation, Venmo is finally being monetised, and management raised full-year profit guidance. The business is doing better while the stock is down 16% over the past year. That gap is the trade.

We are buying common stock for the long term, and we are doing it four days before PayPal reports its next quarter on 27 October. We say more about that in the position section; the short version is that a single print does not make or break a multi-year thesis, and we would rather own the setup than trade the number.

PayPal moves money. In the most recent quarter total payment volume was about $486 billion, up roughly 9 to 10%, on revenue of $8.68 billion. The number that actually matters is not revenue but transaction margin dollars, the gross profit of the payments business, because PayPal's revenue includes low-margin volume it does not care about. Transaction margin dollars grew 1% overall and 3% excluding interest income, and management guided the full year to about $15.6 billion. The job of the current management, under CEO Alex Chriss, is to bend that line, and the pieces are starting to move.

Branded checkout is the crown jewel: the PayPal button at checkout, the highest-margin volume PayPal has. It grew 2% currency-neutral for a second straight quarter. That is not exciting in isolation. It is very important as a trend, because the entire bear thesis is that this number goes negative forever, and two quarters of stabilisation is the first evidence against it.

Venmo is the asset the market has never given PayPal credit for. Its volume grew 14% to about $94 billion, "Pay with Venmo" grew 44%, and Venmo debit-card monthly actives grew more than 50%. For a decade Venmo was a beloved product that made no money. It is now being turned into a payments and commerce business, and that monetisation is early enough to be mostly upside.

Buy now, pay later volume grew 26%, and it matters beyond its size because BNPL users spend more and PayPal takes share of checkout when it offers it. The transformation is not one big lever; it is branded checkout stabilising, Venmo monetising and BNPL growing, all at once, against a share count that is shrinking.

This is the part we are getting for free, and it is the reason the house finds PayPal interesting rather than merely cheap. The house question is where the structural bottleneck sits, and in the coming world of AI agents that shop and buy on your behalf, the bottleneck is trust and settlement: an agent in a chat window needs a way to pay that the merchant trusts, the buyer has authorised, and someone stands behind if it goes wrong. That is a payments-rails problem, and it is PayPal's core competence.

PayPal has partnered with OpenAI to power Instant Checkout and agentic commerce inside ChatGPT, bringing its wallet and merchant network into the chat interface. Set aside whether this specific partnership is large; the point is directional. When commerce moves from a website with a checkout page to an agent in a conversation, the checkout button does not disappear, it moves, and whoever owns the trusted-settlement layer in the new interface owns the same bottleneck they owned in the old one. PayPal has 400 million-plus consumer accounts and tens of millions of merchants already connected, which is the hard part to replicate.

We are explicit that this is an option, not a forecast. It carries no disclosed near-term revenue and it may amount to little. We are not paying for it, because the stock is cheap on the existing business alone, so if agentic commerce becomes real and PayPal is a set of rails for it, that is upside we did not underwrite. If it comes to nothing, the base case still stands on its own.

The most reliable part of this thesis is arithmetic. PayPal generated about $6.6 billion of free cash flow over the trailing twelve months against a market capitalisation of roughly $49 billion. That is a free cash flow yield in the low-to-mid teens, which for a profitable, growing payments network is unusual and is the real reason to own it.

Management has spent the bulk of that cash buying back stock, and the effect compounds quietly: the share count has fallen from over a billion shares a few years ago to about 855 million today. When a company retires 5% or so of its shares a year, flat earnings become mid-single-digit earnings per share growth and modest earnings growth becomes double digits per share. Stack that on top of a business where transaction margin dollars are growing again, and you do not need a re-rating to make money. You need the buyback to continue and the operating line to not deteriorate, and the re-rating, if it comes, is extra.

That is also why David Einhorn's Greenlight built a new PayPal position in 2026 at an average around $43.53, arguing that 400 million-plus customers and the payment assets were undervalued at roughly eight times earnings. We did not open our position because Einhorn did, but a disciplined value investor reaching the same conclusion from the same arithmetic is worth noting.

PayPal trades at about $57, roughly ten times the $5.38 of full-year non-GAAP earnings management has guided to, with a market capitalisation near $49 billion. For context, the company raised that EPS guide this year, so the multiple is on rising, not falling, earnings. A payments network growing transaction margin dollars, with a low-teens free cash flow yield and a shrinking share count, does not usually trade at ten times earnings, and the gap is the opportunity.

We will be fair to the skeptics, because the house rule is to state the other side. The average sell-side price target sits around $56, essentially at the current price, which means Wall Street as a whole does not see much upside here. That is consistent with a stock the market has decided is cheap for a reason. Our disagreement is specific: the reason the market prices in is permanent branded-checkout decline, and the last two quarters are the first data that the decline has stopped. If that holds, a business this cash-generative does not stay at ten times earnings, and the combination of a modest re-rate, the buyback and mid-single-digit margin-dollar growth is a strong multi-year return without anything heroic.

We are long PayPal common stock, opened 23 October 2026, for a three-to-five-year horizon. Shares, not options. The disclosure block at the top of this note carries the terms.

The calendar fact, stated plainly as we always do: PayPal reports its next quarter on 27 October 2026, four days after we opened and three days after this note publishes. We are not positioned for that print, and we are telling you we are holding through it. If branded checkout re-accelerates or Venmo monetisation surprises, good; if the quarter is soft and the stock falls, the thesis is measured in years and a buyback getting cheaper stock is not the worst outcome for a long-term holder. We mention the date because a reader deserves to know a catalyst sits immediately ahead, not because it is why we are in the name. Sizing is Constructive, built to be added to on weakness.

The risks here are real and they are the reason the stock is cheap.

Branded checkout can resume losing share. Two quarters of 2% growth is stabilisation, not victory. Apple Pay is default on a billion iPhones and Shopify steers merchants toward Shop Pay. If branded checkout rolls back over, the highest-margin part of the business shrinks and the thesis is wrong. This is the one number that matters most, and we are watching it above all others.

Transaction margin compression. PayPal can grow volume while its take rate falls, so headline TPV growth can mask a shrinking gross-profit line. Transaction margin dollars grew only 1% overall last quarter; the ex-interest figure was better at 3%, but a reader should track the margin-dollar line, not the volume, and should know that falling interest rates pressure the interest-on-balances contribution.

The agentic-commerce story could cut the other way. We framed AI agents as an opportunity. They are also a threat: if an agent abstracts away the checkout choice entirely, or if Apple, Google or the model labs build their own trusted settlement, PayPal's button could be disintermediated in the new interface the same way it is pressured in the old one. We own this as an option precisely because its sign is uncertain.

Beneath those: the market's disinterest is itself a risk, because a cheap stock can stay cheap for years, and the next print lands in days. We think the cash flow and the buyback pay you to wait. We could be early.

The open questions we consider the next highest-value work on this name.

1. 01

   Branded checkout, every quarter, above everything. The whole thesis is whether the stabilisation holds. Track the currency-neutral growth rate and any disclosure on PayPal's share of US checkout.

2. 02

   Transaction margin dollars, ex-interest. Separate the gross-profit growth of the payments business from volume and from interest income, so a falling take rate cannot hide inside rising TPV.

3. 03

   Size Venmo's monetisation. Build an estimate of Venmo revenue and its trajectory. It is the most underappreciated asset and the clearest source of upside to the margin-dollar line.

4. 04

   Model the buyback. At the current price, quantify how much share-count reduction the free cash flow buys per year and what that alone does to EPS. This is the floor under the return.

5. 05

   Watch the agentic-commerce rails race. Track whether PayPal, the card networks, Apple, or the model labs end up owning settlement inside AI interfaces. This decides whether the option is worth anything.

• PayPal Q2 2026 results, reported 28 July 2026
• Branded checkout and Venmo metrics
• Transaction margin dollars and full-year guidance
• OpenAI Instant Checkout and agentic commerce in ChatGPT
• Greenlight Capital (Einhorn) PayPal position
• PayPal free cash flow and buybacks
• PYPL price and valuation
• PYPL short interest
• Buy now, pay later volume