I started with the exit, because it came with the cleaner numbers. On 7 January 2026 Goldman Sachs told the SEC that "the firm has entered into an agreement to transition the Apple Card program and associated accounts to a new issuer", and that "the transition is expected to take place in approximately 24 months."
The filing prices the goodbye. The deal was expected to add $0.46 to fourth-quarter diluted earnings per share through "a release of $2.48 billion of loan loss reserves reflected in provision for credit losses, partially offset by a reduction in net revenues of $2.26 billion related to markdowns on the outstanding credit card loan portfolio and contract termination obligations as well as $38 million of operating expenses." My arithmetic puts the net pre-tax effect at a positive $182 million. Almost all of it comes from released reserves, not from anyone paying for a going concern.
That is what leaving looked like for the bank, as far as I can read it. I wanted to see the other side, a platform that is building the same layer instead of handing it back.
Shopify's 2025 annual report splits revenue and cost of revenue into two lines, and that split is enough. Subscription solutions brought in $2,752 million, with $520 million of cost of revenue. Merchant solutions, the payments and financial layer, brought in $8,804 million with $5,481 million of cost.
Subtract and you get $2,232 million of gross profit from software at an 81.10% margin, and $3,323 million from the financial layer at 37.74%. The two add up to $5,555 million, which is the filed gross profit line exactly. I like it when a split reconciles to the dollar, since nothing can hide in between.
Here is the gap I think decides valuation. The financial layer supplies 76.19% of revenue but 59.82% of gross profit. By revenue it is 3.2 times the software business. By gross profit it is 1.5 times.
Shopify says the growth comes from payments. Its report notes that "merchant solutions revenues increased from $6.5 billion in the year ended December 31, 2024 to $8.8 billion in the year ended December 31, 2025, representing an increase of 35%", and that "the Shopify Payments penetration rate was 65.6%, resulting in GMV of $248.1 billion that was facilitated using Shopify Payments."
If you are buying or selling an embedded finance company, this is the part to use. Price each layer on its own. A single revenue multiple on a blended line hides a mix assumption, and two companies with the same revenue and the same growth can differ by a third in gross profit. For a founder, put the layer split in the data room early, because a buyer who has to derive 37.7% from cost of revenue will discount you for the trouble.
I prefer a per-unit view, because it travels between companies. Shopify processed $378.4 billion of gross merchandise volume in 2025. Divide each gross profit figure by that and the financial layer earns $8.78 per $1,000 of volume, software $5.90, together $14.68. Financial layer revenue runs at 2.327% of volume.
I have to flag a weakness here. Those per-unit numbers assume all merchant revenue follows volume, and Shopify only says that "the majority of our merchant solutions revenues are directionally correlated with the level of GMV that our merchants process through our platform." Shipping, hardware and advertising sit in the same line and follow other drivers. So treat $8.78 as an upper bound for the payments part alone.
A small aside first, not useful to anything. The Goldman filing also renames a segment, describing it as "activities related to issuing credit cards to and raising deposits from Apple Card customers, as well as results from activities related to Platform Solutions businesses that have been exited." I stopped on the phrase "have been exited", which is a very bank way to describe a retreat. Back to the Shopify numbers.
I held both layer margins where they are and changed only the mix. At today's 76.2% merchant share the blended gross margin is 48.1%. At 85% it falls to 44.2%, and at 90% to 42.1%. Neither product got worse in that exercise, yet the reported margin still slides.
I hear one claim from embedded lenders more than any other: their checkout data beats the credit bureau. The best evidence I know on that is "On the Rise of FinTechs: Credit Scoring Using Digital Footprints" by Tobias Berg, Valentin Burg, Ana Gombović and Manju Puri, in The Review of Financial Studies. They find that "even simple, easily accessible variables from a digital footprint match the information content of credit bureau scores."
The second finding matters more for price. The footprint "complements rather than substitutes for credit bureau information and affects access to credit and reduces default rates." Complement is the key word. An incumbent can buy the same bureau file, so I would not pay for checkout data as a moat on the strength of this paper.
The company explains the low margin itself: "gross profit margins on Shopify Payments, the biggest driver of merchant solutions revenue, are typically lower than on subscription solutions due to the associated third-party costs of providing this solution." A layer with a thin gross margin and little extra operating cost can still lift operating margin for the group. Gross profit dollars pay for engineers, not percentages.
So I am left choosing between errors. Value the payments layer on its 37.74% margin and you undervalue it. Value it on its 35% growth and you overvalue it. The layer earned $3,323 million of gross profit last year against $2,232 million from software, a 43-point margin gap going one way and a $1,091 million profit gap going the other, and I have not found a single figure in either filing that settles which error is smaller. It still bothers me that the most useful number here, operating cost by layer, is the one Shopify does not publish.