How neobanks and fintechs offer stablecoin yield

Adding an earn product is mostly a decision about whose balance sheet the yield comes off, and that decision drives everything else.

A neobank offers stablecoin yield by holding customer balances as a stablecoin, routing some portion into a yield-generating position, and deciding how much of the return reaches the customer. Most run one of three models: earning on their own float and keeping it, passing a disclosed rate through to customers as a savings product, or offering a self-directed product where the customer picks the strategy. The choice determines the licensing, the disclosure burden, and how much of the spread stays in-house.

Getting that framing right first saves a lot of wasted scoping. Teams often start by evaluating yield sources and rates, then discover halfway through that the model they'd assumed requires a licence they don't hold, or that their regulator treats a headline APY as an investment promotion. The yield mechanics are the easier half of this problem. The structural questions below are the ones that decide whether a product ships.

Where does the yield actually come from?

Stablecoin yield originates from one of four places, and every product in the market is a wrapper around some combination of them. Knowing which one a partner is offering matters more than the number they quote, because the risk, the durability of the rate, and the regulatory treatment all follow from the source.

• Short-dated government paper. Tokenised Treasury bills and money market funds pay close to the policy rate. Predictable, well understood by regulators, and the rate falls when central banks cut.
• Overcollateralised lending. Depositing into a lending market and earning what borrowers pay. Rates float with utilisation, so they rise in active markets and compress in quiet ones.
• Multi-strategy DeFi. A curator running several positions at once rather than parking capital in one market: providing liquidity on decentralised exchanges and earning trading fees, carry trades that borrow a cheap asset on one lending market to supply a better-paying one, and looped positions that recycle borrowed capital back into the same market to amplify a spread. More moving parts than single-market lending, and the return depends on the curator's execution rather than on one published rate.
• Basis and funding strategies. Capturing the spread between spot and derivatives pricing. Historically strong returns, more complex to run, and dependent on funding conditions that can invert.
• Issuer subsidies and incentives. A chain or a token issuer paying to bootstrap balances. Real while it lasts, and a poor thing to build a product's unit economics on.

The distinction that matters commercially is between yield that comes from a rate someone else sets and yield that comes from a strategy someone actively runs. A neobank passing through a tokenised Treasury rate is reselling a rate. A neobank routing balances into a multi-strategy position is offering something a competitor cannot trivially match on price, which is the difference between a feature and a moat. Circle publishes plainly on how USDC reserves are held, and Tether's transparency reporting covers the same ground for USDT, both worth reading before assuming a stablecoin's own reserves generate anything for the holder.

The three models, and what each one costs to run

Model choice is the first architectural decision and the hardest to reverse. Each one implies a different licensing posture, a different disclosure obligation, and a different split of the economics.

Hybrid tiering is where most sophisticated products end up, because a single rate forces a compromise between the customer who wants safety and the customer who wants return. Running a conservative anchor product alongside one or two higher-variance options lets a neobank serve both without mispricing either, and it moves the risk decision to the customer where the disclosure is cleaner. Chainlink's write-up on the stablecoin neobank model is a decent overview of how these pieces fit together at the architecture level.

What does a fintech need before it can launch an earn product?

Six things gate a launch, and in practice the last two are what delay teams by a quarter. Working through them in order avoids building on an assumption that collapses at diligence.

1. A model decision from the table above, signed off by whoever owns regulatory risk rather than whoever owns the roadmap.
2. Custody. Where balances sit, and whether the custodian can hold the receipt token representing the yield position. Custodian whitelisting is a gating dependency, not a formality.
3. A yield source with published risk parameters, so the disclosure to customers can be specific rather than gestural.
4. Redemption mechanics that match the product promise. An app that shows a spendable balance needs redemptions that clear on the same timeline the interface implies.
5. Diligence material the compliance team can forward. Audits, risk framework, entity structure and third-party assessments, in a form that can be sent on without a call.
6. An answer on who is accountable when a strategy underperforms. Not a legal disclaimer, an operational answer about who monitors, who can pause, and what the customer sees.

How the infrastructure decision works

Build-versus-integrate resolves quickly once a team prices the build honestly. Vault contracts, NAV accounting, fee logic, withdrawal queues, risk controls and an audit programme are a multi-quarter engineering commitment before a single customer sees a rate, and the ongoing cost is a permanent risk-management function rather than a one-time project. Almost every fintech that has shipped an earn product in the last two years integrated rather than built.

Upshift is non-custodial vault infrastructure for onchain yield, used by fintechs, exchanges, wallets, neobanks, chains and asset managers to launch custom yield products. Vault contracts, risk controls, NAV accounting, fee logic and withdrawal processing are production-ready, and a partner configures a vault to its own parameters through Vault-as-a-Service rather than depositing into a pool someone else designed. As of July 2026 the platform has processed more than $550 million in deposits at peak across over 50 vaults on more than 30 chains, with contracts that have completed 10 smart contract audits across 6 independent firms.

The strategy itself is run by a curator, which is typically a hedge fund or asset manager that allocates within limits the vault enforces. Sentora runs vaults on the platform, and a partner can bring its own curator or use an existing one. Kraken Institutional took this route to give clients access to onchain yield from inside their existing custody arrangements, which is documented in the write-up of that integration. On the developer side, the SDK is live with Tria and app.monad.xyz.

The ERC-4626 standard underneath matters more than it sounds. A standardised receipt token can be held by a custodian, used as collateral elsewhere, or integrated by a third party without bespoke work, which is what turns a yield position into something a product team can build on rather than a closed integration.

Where does regulation actually bind?

Regulatory treatment turns on whether the product looks like a deposit, an investment, or a payment balance, and the answer differs by jurisdiction in ways that change the product rather than just the paperwork. In the United States the GENIUS Act and CLARITY Act reshaped how yield attaches to stablecoin balances, and the practical effect for a neobank is that the yield source shifts from issuer subsidies toward vault-generated returns. That distinction is covered in more depth in how the GENIUS Act and CLARITY Act reshape stablecoin yield.

Under European e-money rules, balances held for customers carry constraints on deployment, which is why several European fintechs run yield as an explicitly separate opt-in product rather than a rate on the main balance. Marketing rules bite here too: a headline APY on a homepage is a financial promotion in most regulated markets, with all the approval requirements that implies. Wharton's Stablecoin Toolkit is a vendor-neutral primer worth handing to a compliance team early.

The broader direction of travel is toward stablecoin balances being treated as ordinary financial infrastructure, which Bessemer traces in its work on stablecoins moving from DeFi primitive to global financial infrastructure. Products designed for that end state tend to age better than products designed around a current arbitrage.

The risks worth disclosing properly

Smart contract risk is the tail risk in any onchain yield product, which is why audit history and per-protocol exposure limits deserve more attention than a headline rate. Non-custodial architecture constrains the damage, because funds move only between the vault contract and whitelisted strategy contracts rather than to an external wallet, and neither the platform nor the curator can withdraw customer funds to an arbitrary address. Upshift's risk management framework documents the buffer, the NAV volatility caps and the whitelisting model.

Rate risk is the one customers feel. Variable yields fall, and a product that acquired customers on a headline number will face churn when that number compresses, so the messaging should set expectations that survive a rate cycle. Liquidity risk shows up when a redemption promise in the interface outruns the redemption mechanics underneath, and it's worth stress-testing against a coordinated withdrawal rather than an average day. Counterparty risk applies wherever a strategy touches a lending venue or a centralised borrower, and the honest treatment is exposure limits plus transparency on where capital sits. Where a strategy allocates into lending markets, both Morpho and Aave publish live utilisation and rate data, which is the right place to sanity-check a projected number before it reaches a customer-facing screen.

Always make sure to do your own research and be aware of the above and any other risks before depositing.

Frequently asked questions

How much of the yield do neobanks typically keep?

Pass-through savings products commonly retain 15 to 30 percent of the gross yield as revenue and pass the remainder to customers. Float monetisation models keep all of it and offer customers no rate, which is a defensible choice as long as the deployment is disclosed. The split is a commercial decision constrained by what the licence and the local promotion rules allow.

Does a fintech need a banking licence to offer stablecoin yield?

It depends on the model rather than on the technology. Float monetisation on balances a company already holds generally does not require new permissions beyond disclosure, while a pass-through savings product usually needs deposit or e-money permissions, and a self-directed product raises suitability obligations. This is a question for counsel in each target market, not one with a single answer.

What is the difference between building and integrating yield infrastructure?

Building means owning vault contracts, NAV accounting, fee logic, withdrawal queues, risk controls and an audit programme, which is a multi-quarter commitment plus a permanent risk-management function. Integrating means configuring a vault on existing infrastructure and going live in weeks. Almost every fintech that has shipped an earn product recently integrated.

Who runs the strategy behind the yield?

A curator, typically a hedge fund or asset manager, allocates within limits the vault enforces at the chain, protocol, token and function level. A partner can bring its own curator or use one already on the platform. The curator cannot move funds outside whitelisted contracts, which is what separates strategy discretion from custody.

Can customers withdraw instantly?

That depends on how the vault is configured. A liquidity buffer sized to expected withdrawals serves redemptions immediately without unwinding a strategy, and vaults process claimable redemptions daily. A product whose interface implies instant access needs the buffer sized to match that promise.

What happens when interest rates fall?

Yield sourced from short-dated government paper falls with the policy rate, since the product is reselling that rate. Lending and strategy-based sources move on different drivers, so a multi-strategy position is less directly exposed to a single rate cycle. Any product marketed on a headline number should model the downside case before launch.

Keep reading

• How to earn yield on stablecoins in 2026 covers the same yield sources from the depositor's perspective.
• How traditional asset managers are controlling risk in DeFi is the risk-control argument a compliance team tends to want.
• Why every financial product will run on vaults makes the structural case for why balance-holding products converge here.
• Upshift DeFi Yield documents how multi-strategy vaults allocate across whitelisted protocols.
• The Upshift FAQ answers the custody, redemption and fee questions that come up in diligence.

Keep reading

• How to use an onchain vault in a treasury product: the ETP comparison, fees, and what to decide first.
• What it takes to launch a vault for your own clients: the operational path from mandate to live vault.
• The DeFi yield strategies onchain vaults run: the strategy shapes, the risk in each, and measured returns.