Ethereum Paymaster Gas: Who Pays?
The paymaster pays, and anyone who tells you otherwise has never actually inspected a UserOperation bundle at the mempool level. That is the blunt mechanics of ERC-4337: a smart contract voluntarily fronts the gas, the bundler gets reimbursed, and the end user signs off with a clean conscience and an empty wallet. But the moment you ask who ultimately eats the cost, the clean abstraction cracks wide open, and the answer becomes a ruthless game of incentive engineering.
The Sponsor Contract Is the Real Payer
Strip away the marketing language and a paymaster is nothing more than a contract holding a deposit in the EntryPoint. When a UserOperation arrives, the EntryPoint calls validatePaymasterUserOp, checks the sponsorship policy, and deducts gas from that deposit. The user's balance never moves. The sponsor's balance bleeds. Whoever deployed and funded that contract is the entity actually paying — a dApp subsidizing onboarding, a gaming operator covering entry fees, or a relayer network absorbing costs to capture downstream volume. There is no magic money printer here, just a treasury with a burn rate.
🔥 This is where operators get lazy. They fund a paymaster, watch the deposit drain, and assume the economics hold because volume looks healthy. Then a bot swarm discovers the sponsorship rule and drains the deposit in an afternoon of worthless UserOperations.
Spoofing the User Pays Nobody
Every serious gasless transaction design eventually collides with the abuse surface. A paymaster that sponsors indiscriminately is a faucet with no valve. The standard defense is a signed sponsorship policy — off-chain attestation, allowlists, per-address rate limits, or stake-based reputation checks inside the validate function. If the paymaster validation logic does not reference a cryptographic proof of eligibility, the sponsorship is open to griefing. Signature verification costs gas too, and that overhead is paid from the same deposit, which means a poorly scoped policy burns funds on rejecting invalid requests just as fast as approving valid ones.
Bundler Economics and the Refund Pipeline
Bundlers are not charities. They simulate the UserOperation, estimate the preVerificationGas and verificationGasLimit, and only include the bundle if the paymaster's deposit covers the worst-case path. The postOp refund flows back to the bundler's account, and any surplus returns to the paymaster. The subtle trap is the penalty regime: if a paymaster's validate function fails during execution after passing simulation, the EntryPoint slashes its stake. A sponsor that underprices its own validation logic does not just lose gas — it loses collateral. The genuine payer, in the failure case, is the stake, and that stake is denominated in ETH sitting idle in a contract that a single malformed UserOperation can partially liquidate.
Who Really Carries the Bag
Trace it to the end and the payer is whoever funded the paymaster deposit or posted the stake, minus whatever value the sponsorship generated in downstream revenue. Subsidized onboarding is a customer acquisition cost. Sponsored gameplay is a marketing line item. Relayer-funded gas is arbitrage against future fee capture. The ERC-4337 stack does not eliminate gas cost; it relocates it, hides it behind a validation function, and hands the bill to the contract owner. The moment the sponsorship policy has a gap wider than the deposit runway, the payer becomes the next exploit's exit liquidity at a gas price of exactly whatever the attacker chooses.