Account Abstraction (ERC-4337 and EIP-7702)
Executive summary
Account Abstraction promised to upgrade Ethereum's user experience from awkward externally-owned-account (EOA) flows to programmable smart-contract wallets supporting gasless transactions, batched operations, social recovery, and policy-driven security. ERC-4337, the off-chain mempool standard that shipped on Ethereum mainnet in March 2023, became the foundation for AA-native wallets including Coinbase Smart Wallet, Safe with AA, Daimo, Argent, and Privy/Dynamic-powered embedded wallets. By 2026, AA wallets collectively held tens of billions of dollars and processed hundreds of millions of UserOperations, but the predicted sweep of EOAs did not happen: most active users remained on EOA wallets like MetaMask, partly due to inertia and partly due to the complexity AA introduced. The Pectra hardfork in 2025 shipped EIP-7702, a hybrid path that allows EOAs to temporarily delegate to smart-contract logic, accelerating adoption by removing the migration friction. The 2026 consensus is that AA is not a replacement but an upgrade path, with EIP-7702 being the practical bridge.
What Account Abstraction actually is
Ethereum has historically supported two account types: externally-owned accounts (EOAs) controlled by a private key, which can initiate transactions and pay gas, and contract accounts which can hold logic but cannot initiate transactions. Account Abstraction (AA) collapses this distinction by allowing accounts to be smart contracts that can initiate, pay for, and validate transactions according to programmable logic. The benefits are substantial: gas can be paid in any token (or by a third party), transactions can be batched into atomic operations, signing schemes can be customized (multi-sig, social recovery, biometric, hardware-secured), spending policies can be enforced at the account level, and account upgrades become possible without changing the address. ERC-4337, authored by Vitalik Buterin and others and finalized in March 2023, achieves this without an L1 protocol change by introducing an off-chain mempool of UserOperations (intent-style transaction objects), a Bundler (an EOA that aggregates UserOperations into bundles and posts them to a singleton EntryPoint contract), and a Paymaster (a contract that can sponsor gas for UserOperations meeting policy criteria). The architecture is complex but flexible. By 2026, ERC-4337 had been deployed across Ethereum mainnet and most major L2s with stable bundler infrastructure provided by teams like Pimlico, Stackup, Alchemy, and Biconomy.
The smart-contract wallet ecosystem
The AA wallet landscape consolidated around several major players by 2026. Safe (formerly Gnosis Safe) - the largest smart-contract wallet by TVL, holding $100B+ in assets across multi-sig deployments - extended its functionality with Safe-AA modules that bring 4337 compatibility to existing Safe accounts. Coinbase Smart Wallet, launched in 2024, became one of the largest consumer-facing AA wallets, offering passkey-based accounts with no seed-phrase requirement, gasless transactions on Coinbase-supported L2s, and seamless onboarding from email or Apple/Google login. Daimo built a Venmo-style consumer payments app on top of AA, demonstrating that AA enabled UX flows (instant gasless transfers, no gas-token requirement) competitive with Web2 payment apps. Argent, an early AA pioneer dating back to 2018, evolved its smart-contract wallet across Ethereum L2s and Starknet with social recovery as a hallmark feature. Embedded wallets from Privy, Dynamic, Magic, and Web3Auth - which provision smart-contract wallets transparently to users authenticated via Web2 credentials - became the dominant onboarding path for new users in 2024-2026, particularly in agent-payments and game contexts. Aggregate AA wallet count crossed tens of millions by Q1 2026, though active usage remained skewed toward power users and consumer-app integrations.
Gasless transactions, paymasters, and the UX upgrade
The headline AA UX feature is gasless transactions: users can transact without holding ETH (or the chain's native gas token) by having a Paymaster contract sponsor gas. Paymasters can be funded by the application (sponsoring user gas as a customer acquisition cost), by the user paying in an alternate token (USDC paymasters that swap stablecoins for ETH at execution time), or by hybrid arrangements. By 2026, paymaster infrastructure was mature: Pimlico, Stackup, Biconomy, Alchemy Account Kit, and ZeroDev each offered turnkey paymaster services, and most consumer crypto apps integrated some form of gas sponsorship. The economic implication is meaningful: applications that absorb gas costs convert higher than apps that require users to manage gas tokens, particularly for new users. Coinbase Smart Wallet's gasless Base transactions, Daimo's gasless USDC transfers, and Friendtech-era social-app gas-sponsorship all demonstrated the conversion impact. Batched operations - multiple actions atomically combined into a single transaction - are the second major UX upgrade: a user can approve and swap in one operation rather than two transactions, eliminating the awkward 'approve, then swap' two-step flow that has been a UX pain point since 2017. By 2026, batched operations became the default for AA-native DeFi flows, with savings in both gas cost and user friction.
Why AA didn't sweep the market
Despite the technical and UX advantages, AA did not displace EOA wallets at the pace its proponents predicted. Several reasons. First, the existing EOA infrastructure (MetaMask, Trust Wallet, hardware wallets like Ledger and Trezor) had massive installed bases and switching costs. A user with 5 years of MetaMask history is unlikely to migrate to a different wallet for marginal UX gains. Second, AA introduces complexity that some users prefer to avoid: smart-contract wallets are slower, more expensive in some operations, and have less universal support across protocols. Third, security models differ: smart-contract wallets are upgradeable and depend on the wallet implementation's audit quality, which some institutional users are uncomfortable with. Fourth, deposit/withdrawal flows from CEXs can be more complex with smart-contract addresses, particularly when using L2s. By 2026, the practical reality was that AA wallets dominated new-user onboarding (especially via embedded wallets in consumer apps and agent-payments contexts), while EOA wallets retained dominance among existing power users and DeFi natives. EOA wallets remained the majority of active accounts through 2026, with the AA share growing slowly but consistently. The agent-economy use case (covered in the AI x Crypto deep-dive) was a particular accelerator for AA adoption.
EIP-7702 and the hybrid path
The Pectra hardfork in May 2025 included EIP-7702, a proposal authored by Vitalik Buterin in 2024 that allows EOAs to temporarily delegate execution to a smart-contract address. The mechanism: an EOA can sign a special transaction that designates a smart-contract code template; for the duration of that transaction (or until revoked), the EOA executes as if it had that contract's code. This bridges the EOA and AA worlds in a critical way: existing EOA users get AA-style features (batching, gas sponsorship, social recovery delegation, session keys for gaming) without migrating to a new address. EIP-7702 effectively makes AA features available as opt-in upgrades to existing accounts. The implementation through 2025-2026 saw rapid adoption: MetaMask shipped 7702-based features (batching, gasless transactions via paymaster delegation), Safe added 7702 support for delegating Safe-style logic to plain EOAs, and several wallet providers began offering 7702-based session keys for app-specific functionality. By Q1 2026, 7702 was the dominant new AA-adjacent feature, and the predicted EOA-vs-AA dichotomy was effectively dissolved into a spectrum where EOAs could opt into AA features as needed. The strategic implication is that pure-AA wallets (where the wallet is a smart contract from inception) and 7702-augmented EOAs converge on similar feature sets, with the choice mostly a matter of wallet provider and user preference.
Recent milestones (2024 to 2026)
Major milestones in this stretch: Coinbase Smart Wallet launch in June 2024 with passkey-based accounts and gasless Base transactions; Safe-AA module rollout through 2024 bringing ERC-4337 to existing Safe wallets; Daimo's growth as a consumer payments app on Base; Argent's expansion across Starknet with native Cairo smart-contract wallets; embedded wallet providers (Privy, Dynamic, Magic) crossing tens of millions of provisioned wallets through 2024-2025; Pimlico, Stackup, Alchemy, Biconomy, and ZeroDev maturing as bundler/paymaster infrastructure providers; the Pectra hardfork in May 2025 shipping EIP-7702; MetaMask's 7702-based Smart Account features rolling out through 2025; the broader paymaster-infrastructure economy reaching multi-billions of UserOperations processed cumulatively by Q1 2026; the agent-economy use case driving AA adoption in B2B and autonomous transaction contexts; and the gradual EOA-to-7702 migration of some power users seeking specific AA features. ERC-4337 v0.7 and v0.8 specifications standardized through 2025 brought additional improvements (signature aggregation, post-validation hooks). By 2026, AA was no longer a technology debate; it was a deployed infrastructure layer with multiple coexisting implementation patterns.
Risks, key management, and outlook through 2027
AA risks fall into three buckets. First, smart-contract risk: AA wallets are smart contracts, and bugs in wallet implementations have led to fund loss in past incidents (Wintermute's 2022 Profanity-related loss, several Argent and other wallet incidents). The tail risk of a widely-used wallet implementation's exploit is significant. Second, key management risk: AA wallets often use passkeys, social recovery, or session keys, each of which has distinct failure modes. Passkey loss without recovery setup leads to permanent fund loss; social recovery depends on the recovery contacts' integrity; session keys can be over-permissioned and lead to unauthorized transactions. Third, paymaster and bundler centralization: most production AA infrastructure is operated by a small number of providers (Pimlico, Alchemy, Stackup, Biconomy), which is a centralization risk if these providers are compromised or coerced. Mitigations include multi-provider bundler routing and decentralized paymaster pools, but these are early. Looking forward to 2027, the realistic trajectory is that AA features become ubiquitous - every major wallet ships some AA functionality via 4337 or 7702 - while pure-AA wallets remain a strong but minority share. Embedded wallets continue to dominate new-user onboarding. Agent-economy adoption further accelerates AA usage. The deeper open question is whether L1 itself adopts native AA (RIP-7560 or successor) post-Pectra, which would make AA the default account model rather than an opt-in upgrade. That is plausibly a 2027-2028 milestone.
Watch points
- EIP-7702 adoption rate among existing EOAs and the share of transactions using 7702 features
- Coinbase Smart Wallet active user growth and Daimo-style consumer apps
- Bundler and paymaster centralization metrics across major AA infrastructure providers
- Native L1 AA roadmap (RIP-7560 successor) and any timeline updates
TL;DR
Account Abstraction shipped via ERC-4337 in 2023 and EIP-7702 in 2025's Pectra hardfork, with Coinbase Smart Wallet, Safe-AA, Daimo, and embedded wallets (Privy, Dynamic) onboarding tens of millions of users; pure-AA wallets did not sweep EOAs but the 7702 hybrid path effectively made AA features universal by 2026.
Get DeFi Intel research in your inbox
Weekly long-form coverage of papers, incidents, jurisdictions, chains, tokens and the people building them. Free tier covers headlines; Pro adds the analyst-grade breakdowns.