​

Three primitives power stablecoin subscriptions in 2026: pull-payments using ERC-20 allowance, allowance-cap with periodic top-ups, and intent-based deposits that settle into a chosen stablecoin. Sablier and Superfluid handle streaming flows. Stripe Billing added native stablecoin subs. Pick by churn risk and customer wallet behavior.
​

Recurring billing is the use case that stablecoins struggled with for years. Cards do it cleanly with a saved-card-on-file and a merchant-initiated charge. Onchain, the wallet has to authorize each pull, the user has to keep the wallet funded, and a missed payment cannot be solved by retrying a stored credential.
​

2025 to 2026 changed that. The patterns now work. This article walks through the three primitives, the dunning logic, and a comparison of when each fits.
​

The original onchain subscription pattern. The user calls approve() on the stablecoin contract, authorizing the merchant's contract to spend up to N tokens. The merchant calls transferFrom() on the schedule, pulling the recurring amount.
​

Strengths: native ERC-20 primitive, supported by every wallet, no new infra. Used by Loop Crypto, Coinflow, and roll-your-own merchants.
​

Weaknesses:

Best fit: web3-native users, low-friction SaaS where users already understand approvals.
​

A refinement. Instead of a high or infinite allowance, the merchant requests a tight cap (one period's worth) and the dapp prompts the user to re-approve before the next pull. Higher friction, lower trust risk.
​

The trade-off: more UX prompts but no scope creep. Some wallets (Rainbow, Coinbase Wallet, Family) now expose the allowance management UI directly, making the periodic re-approve flow tolerable.
​

Best fit: high-value subscriptions where users care about exposure.
​

A different model entirely. Instead of discrete pulls, the user opens a stream that drips the subscription amount per second. The merchant can withdraw the accrued balance at any point. Stop the stream, stop the bill.
​

Superfluid (per the directory listing in our 2026 stablecoin payment providers audit) handles continuous payroll, vendor payments, and per-second subscriptions. The contract holds a deposit, accrual is real-time, the receiver can drain anytime.
​

Sablier (sablier.com) is in the same family: token distribution with vesting, payments, and airdrops, per-action fees, transparent pricing. Heavier on the vesting side; payment-streaming exists.
​

Strengths: pro-rated billing is free, cancellations are instant, no failed-charge dunning loop because the user can never owe more than they staged.
​

Weaknesses: gas cost per stream open / close on some chains, less familiar UX, requires upfront token deposit.
​

Best fit: payroll, B2B vendor payments, premium-tier SaaS, anything where the customer wants pro-rated cancel.
​

The newest pattern. The user does not approve anything; the merchant generates a payment intent (a quote + a deposit address + a target settlement stablecoin), and the user sends funds from any wallet, any chain, any token. The intent system normalizes the deposit into the merchant's chosen stablecoin.
​

Daimo Pay (docs.daimo.com) is the cleanest example. Eco Routes provides the intent-based cross-chain settlement layer that powers similar flows.
​

The subscription variant: a hosted billing page with a recurring intent. The user comes back monthly (or on a reminder), pays from whatever wallet they want, the merchant settles into USDC on Base or Solana.
​

Strengths: zero approval risk, any-chain any-token UX, lowest customer setup friction.
​

Weaknesses: not strictly recurring; user has to act each period. Pairs well with reminder emails or wallet push notifications. Closer to invoice-based than true subscription.
​

Stripe shipped stablecoin subscriptions through Stripe Billing (docs.stripe.com/billing/subscriptions/stablecoins). The flow: a SetupIntent collects an authorization from the customer's wallet, the Stripe billing engine fires recurring withdraw calls, and the merchant receives the agreed amount in USDC on Ethereum, Solana, Polygon, or Base.
​

For merchants already on Stripe, this is the lowest-friction path. The trade-off is platform lock-in and Stripe's fee on the stablecoin leg, which is published at 1.5% in their docs.
​

Card dunning is a known craft: retry on day 1, 3, 7, fall back to email, recover 30 to 60% of failed charges through retry logic. Onchain dunning is different because the reasons for failure are different.
​

Common failure modes:

The dunning playbook:

Streaming subscriptions skip most of this because the failure mode is the deposit running dry, which is visible to the user in real time.
​

A short decision tree: