A distribution model, rather than one financial product
Embedded finance places a financial service inside a nonfinancial workflow: accepting payment within commerce software, paying a supplier from an accounting platform, or arranging financing while buying a home improvement. The customer may never navigate to a conventional bank website. The regulated product and its economics nevertheless remain real. A software interface does not turn a deposit into software revenue or turn a loan into a payment service.
The expression covers several businesses with very different risks. Payments principally involve acceptance, settlement, disputes and fraud. Deposit products involve custody, recordkeeping and access to funds. Credit adds underwriting, repayment and funding. Insurance introduces a separate insurer and policy contract. Combining these categories into one market-size figure can count the same customer transaction several times. This article focuses on US payments, deposit and credit arrangements.
The parties behind a single screen
A merchant or software platform owns the customer workflow and distribution. A processor or infrastructure provider supplies APIs, payment connections and operational tools. A bank may hold deposits, issue cards or originate loans. A servicer handles billing, collections and disputes; a separate investor may ultimately own the receivable. One company can occupy several roles, but branding alone does not establish which roles it holds.
Current product disclosures illustrate the distinctions. Stripe Treasury for Platforms identifies Fifth Third Bank as its checking-account provider and says deposit coverage has conditions; Stripe itself is not an FDIC-insured bank. Its US Capital page separately identifies Celtic Bank or Lead Bank as loan issuers and YouLend as the merchant-cash-advance provider. These are product-specific relationships, not a universal bank partner list for everything Stripe sells. [2][3]
A loan creates a repayment obligation under a credit contract. A merchant cash advance is presented as a different financing form, commonly linked to business receipts. A lease-to-own arrangement concerns possession and eventual ownership of an asset. Their legal treatment depends on substance and applicable law; a shared application journey does not make the products interchangeable.
What established production looks like
Shopify provides a concrete payment example. Its 2025 annual report records $248.1 billion of gross merchandise volume facilitated through Shopify Payments, representing 65.6% of Shopify GMV, versus $181.0 billion and 61.9% in 2024. That is a volume-based penetration measure within Shopify, not the share of all merchants or the share of global embedded finance. The filing separately reports merchant adoption by region. A large merchant can lift payment-volume penetration without increasing the merchant-count adoption rate. [1]
Distribution can extend beyond checkout. ServiceTitan documents a TURNS second-look workflow that presents further financing possibilities following a first-provider decline. This is evidence of an implemented product flow; it does not establish how many customers funded a purchase or how much incremental profit the integration generated. [7]
Availability and adoption are separate. Stripe’s current Treasury for Platforms page presents upgraded features in preview, and identifies some capabilities as coming soon. A product page, named customer example and a generally available service are different levels of evidence. No reliable, consistently defined total for all US embedded-finance users is established by these sources. [2]
How the revenue pool is divided
The commercial appeal is distribution at the moment of need. The software provider already has a merchant relationship, transaction context and recurring engagement. Financial products may increase revenue per customer and make switching more costly. The economic benefit depends on whether added revenue exceeds partner charges, implementation, compliance, support and losses.
For payments, merchant pricing funds interchange, network charges, acquiring and processing costs, fraud and disputes, with the remainder divided under commercial agreements. For accounts, the value can include service fees and balance-linked compensation. For lending, interest or a fixed financing charge must cover funding, expected losses, servicing and distribution. Those are different revenue bases: payment volume is not revenue, deposits are not revenue, and originations are not profit.
Illustrative economics, not a company result: a platform processing $100 million at a net retained margin of 0.20% earns $200,000 before its own operating costs. If annual support and oversight cost $150,000 and unrecovered losses cost $75,000, the contribution is negative $25,000. The example explains why a headline volume milestone can coexist with unattractive economics. It does not estimate any provider’s actual margin.
Credit and funding survive the software layer
Transaction data may provide useful evidence of sales, seasonality and repayment capacity. Yet platform sales are not necessarily total business income, and strong recent sales may not persist. Borrowers can have obligations outside the platform. Repayment linked to revenue can move with business activity without eliminating the possibility of distress.
The bank that originates a loan and the investor that eventually funds it may be different parties. A , forward-flow buyer or securitization can expand capacity, but each introduces conditions, and renewal risk. Deposit-funded banks have their own capital and constraints. A platform can experience a sudden contraction in available financing even when its software continues to work.
A blended rate is difficult to interpret without its product, borrower population, aging definition and denominator. Rapid new originations can dilute the ratio before new accounts season. Portfolio returns also depend on fraud, recoveries, prepayment and customer acquisition costs. The cited product pages do not disclose comparable cohort-level credit outcomes or platform-specific risk-adjusted profitability.
Where convenience becomes a customer issue
Embedded services reduce repeated data entry and can bring financing closer to a purchase decision. The same proximity can make the financing provider less visible or encourage a focus on the monthly payment instead of total cost. A promotion can be subsidized through merchant fees, while a longer repayment term reduces the monthly amount without necessarily reducing total payments.
Deposit access is another distinct issue. FDIC insurance protects eligible insured deposits against a bank failure; it is not general insurance against a nonbank app’s failure or against a reconciliation dispute. The FDIC’s third-party-app explanation and the agencies’ July 2024 statement describe these boundaries and operational risks. The 2024 statement expressly says it creates no new supervisory expectations. [4][6]
A customer’s experience depends on what happens after onboarding: refund timing, unauthorized transfers, disputed work, account restrictions and continuity when a partner exits. These outcomes are poorly captured by registrations, announced partnerships or gross transaction volume.
The current policy setting
As of October 4, 2026, the agencies’ September 2026 third-party-risk document remains a proposal. The announcement says existing bank guidance would be replaced when the new guidance is finalized; it does not say replacement has already occurred. The proposal is described as principles-based, tailored and non-binding. [5]
Embedded finance has no single exemption from the laws applicable to the underlying activity. Responsibility follows the product, entity and conduct. Distribution contracts may allocate tasks and financial losses between companies, but a bank’s use of an intermediary does not itself erase its regulatory obligations. The economic question is therefore broader than whether software can add a financing button: it is whether the whole arrangement can deliver a durable, understandable service at an acceptable total cost.
Sources
- Shopify, 2025 Form 10-K, Payments penetration and merchant adoption; year ended December 31, 2025Filing / reportBack to text: ↑
- Stripe, Treasury for Platforms product availability and bank disclosures; checked October 4, 2026SourceBack to text: ↑1↑2
- Stripe, Capital product and provider disclosures; checked October 4, 2026SourceBack to text: ↑
- Banking agencies, deposit products distributed through third parties; July 25, 2024Official sourceBack to text: ↑
- Banking agencies and NCUA, proposed third-party risk management guidance; September 11, 2026, updated September 15Official releaseBack to text: ↑
- FDIC, Banking With Third-Party Apps; last updated May 31, 2024Official sourceBack to text: ↑
- ServiceTitan, second-look TURNS integration documentation; updated July 11, 2026SourceBack to text: ↑