Ethereum Paymaster: Who Actually Pays Gas?
The paymaster contract pays the gas — but only because someone fronted the ETH. That distinction is where 90% of retail bettors get liquidated on account abstraction plays. The Ethereum Paymaster is not a charity faucet. It is a smart contract that agrees to cover gas on behalf of a user, then settles that debt through a token, a subscription, or an off-chain credit line. The real question is not who pays at the protocol level. It is who eats the risk when the settlement layer breaks.
ERC-4337 Mechanics Nobody Explains Honestly
Under ERC-4337, a UserOperation gets bundled by a Bundler, and the EntryPoint contract executes it. The paymaster calls validatePaymasterUserOp, locks up stake in the EntryPoint, and then postOp settles the bill. The paymaster must hold a deposit in EntryPoint — that deposit is the actual gas tank. If the paymaster runs dry, the UserOperation reverts and the Bundler eats the loss. Bundlers are not charities. They will blacklist your paymaster address within one block. This is the cold, hard audit reality.
Most paymaster designs fall into four buckets: sponsored (dApp pays), token (user pays in ERC-20), subscription (prepaid credit), and reputation-based (postpaid trust). The gas sponsorship model looks generous until you realize the dApp is buying user acquisition at $2 to $40 per transaction on mainnet. That math only works when the lifetime value of the wallet exceeds the gas burn. In casino and betting verticals, the LTV is high enough to justify it — but only if the paymaster enforces per-address rate limits. Without limits, a single sybil attacker can drain a $50,000 deposit in under an hour.
The Hidden Credit Risk in Paymaster Staking
Here is the part auditors whisper about. Paymasters must stake ETH in EntryPoint to prevent griefing. That stake is slashable. A malicious bundler or a poorly coded postOp can trigger a slashing event, and the paymaster loses principal. In 2024, at least three production paymasters on L2s were drained through postOp revert manipulation. The attacker did not break cryptography. They exploited the fact that postOp runs after the user op succeeds, and if postOp reverts, the whole bundle reverts — but the paymaster's deposit is already committed. This is not a bug. It is a design tradeoff that favors bundlers over paymasters.
For betting operators, this means the paymaster is a counterparty risk instrument, not just a gas tool. If you sponsor gas for 10,000 wagering wallets, you are extending unsecured credit to every one of them. The settlement can happen in USDC, in your own token, or in nothing at all if the user churns. Smart operators cap exposure per wallet at 0.001 ETH equivalent and rotate paymaster contracts weekly. Dumb operators launch one paymaster, fund it with 20 ETH, and watch it get farmed by bots within 48 hours.
Who Really Pays When the Paymaster Fails
The Bundler pays first. Then the Bundler sues the paymaster operator. Then the paymaster operator discovers that their legal entity is a DAO with no assets. This is the circular liability chain that makes institutional adoption slow. In practice, the only paymasters that survive are those backed by treasury reserves that can absorb 100% loss on any single bundle. That means a minimum of 5 ETH liquid per 1,000 daily user ops, plus a 30% buffer for gas spikes. If you cannot meet that, you are not running a paymaster. You are running a subsidized honeypot. 😅
The algorithm is brutal and simple: deposit ÷ (avg gas per op × ops per hour) = hours until insolvency. No governance token, no revenue share, and no partnership announcement changes that arithmetic. The paymaster pays gas only until it does not. The moment the deposit hits zero, the EntryPoint rejects every op, the Bundler drops your bundle, and your users see failed transactions. That is the terminal state of every undercapitalized paymaster — and it always arrives faster than the spreadsheet predicts.