Smart-Account Wallets in 2026: The Race to Go Native

For a decade, account abstraction was the upgrade Ethereum kept promising and never quite delivered. The idea is simple to state: let a crypto wallet be a smart contract that runs its own rules, instead of a bare private key that can do exactly one thing. The execution took years. In 2026 it arrived twice over, and then the fight over how to finish the job broke into the open.

Smart-account wallets now power tens of millions of accounts, delivered through two upgrades bolted onto Ethereum rather than built into it. ERC-4337 added account abstraction at the application layer in 2023. EIP-7702, live since the Pectra hard fork in May 2025, let ordinary wallets borrow smart-contract powers without moving funds. Between them they have processed well over a billion transactions. Neither, though, is the finish line.

The finish line is native account abstraction: the feature built directly into the protocol, so that no bolt-on is needed at all. Three well-funded camps are now racing to define it. The Ethereum Foundation is aligned with Vitalik Buterin’s EIP-8141; Coinbase’s Base has committed to ship a rival design called EIP-8130 in September 2026; and Paradigm, the crypto investment firm, is pushing a third path called Tempo. Whoever wins gets to decide how your wallet signs a transaction, who pays your gas, and whether a handful of intermediaries sit between you and the chain. This guide explains what a smart-account wallet is, how the current approaches work, why the industry is not satisfied with them, and what the race to go native means for the wallets and exchanges you already use. ETH traded around $1,880 as this went to press.

What a smart-account wallet actually is

Every Ethereum address is one of two things. The first is an externally owned account, or EOA: an address controlled by a single private key, usually derived from a twelve or twenty-four word seed phrase. An EOA can hold assets and sign transactions, and that is roughly all it can do. It cannot set spending limits, batch several actions into one, recover from a lost key, or let anyone else pay its gas. The rules are fixed by the protocol and identical for everyone.

The second is a smart-contract account, better known as a smart account. Here the address is governed by code. That code decides what counts as a valid signature, whether to require more than one approver, how much can be spent and by whom, and what happens if a key is lost. Because the logic lives in a contract, it can be upgraded, extended with plugins, and tailored to a person, a treasury, or a game.

The practical payoff comes down to four powers that an EOA cannot offer. Batching lets a wallet approve a token and swap it in a single click rather than two separate signatures. Gas abstraction lets someone else, an app or a paymaster, cover the network fee, or lets the user pay it in a stablecoin instead of ETH. Custom authentication lets a wallet accept a phone’s passkey or a hardware security module rather than a seed phrase. And recovery lets a user regain access through trusted guardians or a time delay instead of losing everything with one misplaced backup.

A fifth power is quietly becoming the reason many people meet smart accounts at all: the session key. A session key is a temporary, tightly scoped permission, for example the right to make moves in one game for the next few hours and nothing else, that a wallet can grant without exposing the master key. It is what lets an on-chain game feel like a normal app, with no signature popup on every action, and it previews how automation and AI agents will transact on a user’s behalf within limits the user sets. None of this is possible with a plain EOA, which can only hand over everything or nothing.

From one private key to programmable accounts

The concept dates to Ethereum’s early years. Buterin has noted that the community has been, in his words, talking about account abstraction ever since early 2016, via Cointelegraph. The problem was always the same: changing how accounts work at the base layer is invasive and risky, so the feature kept getting deferred in favor of simpler wins.

The breakthrough was to stop asking the protocol for permission. ERC-4337, finalized in March 2023, delivered account abstraction entirely in smart contracts and off-chain infrastructure, with no change to Ethereum’s consensus rules. Two years later, EIP-7702 took a different shortcut: it let an existing EOA temporarily point at smart-account code, so a normal wallet could act like a smart one without abandoning its address or history. Both are workarounds. The endgame, native account abstraction, would make the two-track approach unnecessary by supporting smart accounts in the protocol itself. Understanding why that matters means understanding how the workarounds actually function.

ERC-4337: account abstraction without touching the protocol

ERC-4337 introduced a parallel transaction system that lives above Ethereum rather than inside it. Instead of a normal transaction, a smart-account user signs a structured message called a UserOperation that describes what they want to do. These messages collect in a separate mempool. Specialized actors called bundlers gather them, wrap them into a real transaction, and submit them to a single audited contract, the EntryPoint, which verifies each operation and executes it against the user’s account contract.

Two supporting roles make the system useful. Paymasters are contracts that agree to cover gas, either sponsored by an app or paid by the user in a token such as USDC. Account factories deploy a user’s smart-account contract on first use, often at a deterministic address so it can receive funds before it exists on-chain. The EntryPoint has been revised carefully over time, from v0.6 through v0.7 to v0.8 in 2025, which added native support for EIP-7702 accounts, per the eth-infinitism release history.

Adoption is no longer theoretical. By mid-August 2026, ERC-4337 accounts had generated more than 1.24 billion UserOperations across nearly 64 million accounts with on-chain activity, with paymasters covering roughly $13.4 million in gas, according to the analytics dashboard BundleBear. The cost is architectural: bundlers and the alternative mempool are extra infrastructure that Ethereum’s base protocol does not know about, and a few operators handle most of the volume. That concentration is the thread that unravels later in this story.

Concentration is easy to overlook because the system works. A short list of operators, led by names such as Pimlico and Alchemy, produce most successful UserOperations, and a similarly short list of paymasters sponsors most of the gas, per BundleBear’s operator dashboards. That is efficient, and it is exactly the sort of quiet centralization that account abstraction was meant to avoid. It is why the next two sections matter: a smart-account model that depends on a handful of well-run companies is only as censorship-resistant as their willingness to include your transaction.

EIP-7702: a smart-account upgrade for the wallet you already have

ERC-4337 works best for accounts created as smart accounts from day one. It does little for the hundreds of millions of existing EOAs, including nearly every address behind a MetaMask or hardware-wallet user. EIP-7702, authored by Buterin, Sam Wilson, Ansgar Dietrichs and Matt Garnett and shipped in the Pectra upgrade on 7 May 2025, closed that gap.

It adds a new transaction type, 0x04, that lets an EOA sign an authorization pointing its address at a smart-contract implementation. Once set, the EOA keeps its address, balance and history but executes the delegated code, gaining batching, gas sponsorship and custom rules. The delegation is revocable and can be pointed back to nothing. Marius van der Wijden, an Ethereum core developer, described it plainly to DL News: the change adds a transaction type that allows existing wallets to emulate the functions of account abstraction wallets. Alex Jupiter, a senior product manager at MetaMask, framed it as the moment the ecosystem lined up behind one unified account abstraction roadmap.

Uptake has been fast. By mid-August 2026, BundleBear counted more than 222 million EIP-7702 authorizations, roughly 48 million live delegated accounts, and more than 92 million set-code transactions. Ambire was the first wallet to ship it; MetaMask, Rabby, Trust Wallet, OKX and others followed. The feature also drew immediate abuse, a point the security section returns to. What matters here is the shape of the result: two overlapping standards, one for new accounts and one for old, both stitched onto a chain that still, at its core, thinks in terms of single private keys.

The relay problem: why neither workaround is the endgame

The uncomfortable truth about both ERC-4337 and EIP-7702 is that they lean on middlemen. A smart-account user does not usually broadcast to Ethereum directly; the operation passes through a bundler, or through a relay that promises to include it. That is convenient, and it is a weak point. Buterin flagged it directly in 2026, warning that relay dependency is, as reported by Bitcoin.com, a source of vulnerability and fragility, since if a relay goes offline or if a relay operator refuses to process a specific transaction, the end user has no alternative path to inclusion.

Concentration makes the concern concrete: a small number of bundler and paymaster operators handle most ERC-4337 traffic, which means a small number of parties could, in principle, censor or stall it. Two protocol efforts aim to remove the dependence. FOCIL, short for fork-choice enforced inclusion list, would randomly draft validators as transaction includers each block, so no single operator controls what gets in. Native account abstraction would let wallets submit directly to the public mempool with no bundler in the middle at all. The two are designed to work together, and both point to the same conclusion: to fix the middleman problem, smart accounts have to move into the protocol.

Native account abstraction: three roads, three camps

Native account abstraction means the base protocol understands smart accounts on its own, with no external bundlers and no side mempool. Standard Ethereum nodes would validate and execute these accounts directly. The prize is large enough that it has produced not one design but a competition, and in 2026 that competition became a genuine contest among three camps with different philosophies.

The first road belongs to the Ethereum Foundation and Buterin. It runs through EIP-7701, which proposes maximally flexible accounts that can define arbitrary validation logic, and its formalization EIP-8141, an omnibus that Buterin says wraps up and solves every remaining problem that account abstraction was intended to address, plus more, per Cointelegraph. This track prizes long-term flexibility and post-quantum readiness; its designers call it a walk-away architecture that should not need future hard forks as cryptography evolves. The cost is complexity, especially in the mempool rules needed to stop denial-of-service attacks.

The second road belongs to Coinbase and its Base network. EIP-8130, a Coinbase proposal, offers what it calls account abstraction by account configuration: accounts register an on-chain verifier and choose from a fixed menu of signature types (secp256k1, the P-256 curve behind passkeys, WebAuthn and BLS), with protocol-level multisig and token gas payment. The philosophy is to ship robust features quickly with bounded complexity rather than chase full generality. Base is not waiting for consensus. It has said it will launch EIP-8130 in its Cobalt upgrade in September 2026, working with Optimism, Coinbase and WalletConnect to extend it across OP Stack chains later in the year, per the Base engineering blog.

The third road belongs to Paradigm, whose Tempo design (transaction type 0x76) is the most minimal of the three. It standardizes a small set of primitives, atomic batching, validity windows, gas sponsorship, two-dimensional nonces, and native passkey support, while deliberately refusing arbitrary validation logic and token-based gas payment. Tempo optimizes for shipping fastest with the least new protocol surface, leaving complex cases to ERC-4337 and EIP-7702.

All three are under review in Ethereum’s All Core Devs process, and none has a locked slot. Buterin has tied the Foundation’s version to a future upgrade he refers to as Hegota and suggested it could happen within a year, but that is an aspiration, not a schedule. The one firm date on the board belongs to Base and September.

One motivation sits behind all three efforts: preparing Ethereum for the day today’s signatures are no longer safe. A related proposal, EIP-7851, would let users permanently retire an old-style key after migrating to a quantum-resistant smart account, with a cancellation window as a safety net. Native account abstraction is the vehicle that makes such a migration possible at all, since a protocol that only understands single ECDSA keys has nowhere to put a new signature scheme. That is the deeper reason the Foundation’s camp is willing to pay for complexity now rather than ship the smallest thing that works.

What going native changes for wallets and exchanges

For end users, the visible change is subtraction. If smart accounts live in the protocol, the bundler disappears, the side mempool disappears, and the wallet talks to Ethereum the way an ordinary account does today, only with batching, sponsorship and flexible signatures built in. Base’s own numbers hint at the efficiency gain: it says a native USDC transfer under EIP-8130 costs about 46,000 gas, roughly 63 percent less than the ERC-4337 equivalent, and a sponsored passkey transfer about 68,700 gas, per its engineering post.

For exchanges and wallet providers, the shift is strategic. Coinbase is not merely supporting account abstraction; through Base and EIP-8130 it is trying to set the standard that every other wallet adopts, which is why the effort bundles in Optimism and WalletConnect. Node operators face real work, since supporting a new transaction type means updating client software. And existing smart-account users would go through a one-time migration rather than a disruptive switch. The competitive subtext is hard to miss: the party that defines native accounts shapes the default experience for the next wave of users, and all three camps know it.

Passkeys and the slow death of the seed phrase

The single most consumer-visible feature smart accounts unlock is the passkey. A passkey is a cryptographic credential stored in a phone’s secure hardware, Apple’s Secure Enclave or Android’s Keystore, and unlocked with a face scan or fingerprint. It is the same WebAuthn technology that already replaces passwords on major websites. The catch for crypto is that passkeys use the P-256 curve (secp256r1), which Ethereum could not verify cheaply, since the chain is built around a different curve.

RIP-7212 fixed that. It defines a precompile, a piece of native chain functionality, that verifies P-256 signatures, cutting the cost from roughly 300,000 gas to about 3,450, per Alchemy. It was the first rollup improvement proposal adopted by layer-2 networks, with Arbitrum and Polygon among the early implementers. That is what makes passkey wallets practical: Coinbase’s Base Account (the successor to Coinbase Smart Wallet) and a growing list of others let a user create and operate a self-custodied wallet with nothing but a passkey, no seed phrase to write down. Every native account abstraction proposal treats P-256 and WebAuthn as first-class, which tells you where the industry expects consumer wallets to land. For readers weighing whether to keep a seed-phrase device anyway, our hardware-wallet comparison lays out the trade-offs.

Removing the seed phrase does not remove the recovery problem; it moves it. A passkey lives in one device’s secure hardware, so a wallet built on it needs a plan for the day that device is lost. The common answers are social recovery, where a set of guardians can jointly restore access; time-delayed recovery, which gives the real owner a window to cancel an unauthorized attempt; and cloud-synced passkeys that let Apple or Google back up the credential. Each trades a little self-custody for a lot of resilience, and each introduces a party or a process a user should understand before trusting it with real money. The right choice depends on how much value the wallet holds and how much complexity its owner will tolerate.

Who pays for gas? The paymaster economy

Gas abstraction is the feature users notice first and understand least. Someone still pays the network fee; the smart-account model just makes that someone flexible. In sponsored mode, an app covers gas so a new user can transact with an empty wallet, a familiar pattern from onboarding flows and games. In token mode, the user pays gas in a stablecoin, and a paymaster fronts the ETH.

The economics are not free. Circle’s Paymaster, which lets any ERC-4337 account pay gas in USDC and is live on Arbitrum and Base, adds a surcharge of roughly 10 percent on top of the underlying gas cost, the paymaster’s fee for taking on the ETH exposure and execution risk, per Circle. That is the honest price of gasless: convenience funded by whoever sponsors it, or a premium paid by the user who would rather not hold ETH. As stablecoins become a default unit of account for on-chain payments, the ability to pay fees in dollars rather than a volatile gas token is less a gimmick than a requirement. Native proposals from Base and the Foundation both fold token gas payment into the protocol, which would cut out the paymaster middleman for the simplest cases.

The wallet landscape in 2026

The clearest way to understand smart accounts is to look at who ships them. The dominant name in value terms is Safe, the multisignature standard that began as Gnosis Safe and now custodies more assets than any other smart-account system. In its second-quarter 2026 report, the Safe Ecosystem Foundation said its accounts processed nearly 130 million transactions, held $27.24 billion in self-custodied assets including $6.48 billion in stablecoins (about $1 of every $48 in global stablecoin supply), and passed 63 million total accounts. Lukas Schor, the foundation’s president and a Safe co-founder, said the activity held up even as crypto prices fell sharply in June.

Below Safe sit a range of consumer and developer wallets. Coinbase’s Base Account brings ERC-4337 and passkeys to a mainstream audience and underpins the company’s native account abstraction ambitions. MetaMask, the largest self-custody wallet by users, added MetaMask Smart Accounts on the EIP-7702 path. Ambire was first to ship 7702 and leans into it. Argent, the pioneering smart-contract wallet, rebranded to Ready in 2025 and refocused on mobile. The table below sketches the field.

For organizations, the same technology serves a different purpose. A DAO or a company treasury uses a Safe not for convenience but for control: multiple signers, spending policies, and modules that enforce governance. That is a discipline of its own, and the failure modes are distinct from a consumer wallet’s, as our guide to multisig best practices explains.

The modular standards war: ERC-7579 vs ERC-6900

If a smart account is defined by code, the natural next question is how that code gets extended. Two competing standards define how modules, plugins that add features such as session keys, spending limits or recovery, attach to an account. ERC-7579 is the minimal approach, a thin interface that has become the de facto choice, adopted by Safe, ZeroDev, Biconomy, Rhinestone and OpenZeppelin. ERC-6900, championed by Alchemy, is more prescriptive and heavier.

The distinction sounds academic but has real consequences. A modular standard is what lets a wallet add a game’s session key today and a payroll module tomorrow without redeploying, and what lets a security firm audit a plugin once and have it reused everywhere. Whichever standard wins shapes how quickly the ecosystem can build and vet the features that make smart accounts worth using. It is a quieter fight than the native account abstraction race, but it runs on the same logic: settle the plumbing, and the products follow.

The security ledger: blind signing, drainers and the Bybit lesson

Smart accounts change the security picture, and not only for the better. On the positive side, features like spending limits, session keys with narrow scopes, and social recovery can contain the damage from a single mistake. On the negative side, programmability enlarges the attack surface, and the same batching that saves a click can hide malice in a bundle a user does not fully read.

The defining incident is the Bybit hack of 21 February 2025, the largest in crypto history at roughly $1.5 billion. It was not a flaw in Safe’s contracts. Attackers compromised a developer machine and injected malicious JavaScript into the Safe interface shown to Bybit’s signers, so the screen displayed a routine transfer while the payload quietly swapped the account’s implementation to an attacker’s contract. The signers, using hardware wallets that show only a transaction hash rather than its meaning, approved it blind. The FBI attributed the theft to North Korea’s Lazarus Group. Taylor Monahan, a principal security researcher at MetaMask, called it, via The Block, custom malware designed for this specific hack and these specific signers; Ido Ben Natan of Blockaid described blind signing plus targeted malware as perhaps the fastest-growing threat in crypto.

The retail version of the same problem is the wallet drainer. Here the good news is real: Scam Sniffer reported that phishing losses fell 83 percent in 2025 to $83.85 million, with victims down 68 percent. But EIP-7702 introduced a fresh wrinkle, since a single malicious delegation can hand an account’s execution to an attacker; two 7702 phishing cases in August 2025 cost victims $2.54 million between them. The common thread across the boardroom hack and the retail drain is the signature the victim did not understand. The industry’s answer is clear signing, standardized as ERC-7730, which forces wallets and hardware devices to display what a transaction actually does in human terms. Our deeper look at why the signature is the exploit walks through how these attacks work, and our custody guide explains why holding your own keys is not the same as understanding what you sign.

Smart accounts and the SEC

For US readers, the regulatory question is narrower than it sounds. A smart-account wallet you control yourself is self-custody, and self-custody sits largely outside the securities regime aimed at exchanges and custodians. That boundary got firmer in 2026: SEC staff indicated that software allowing users to transact from their own self-hosted wallets is not, by itself, acting as a broker, as reported by CoinDesk. Meanwhile the market-structure bill known as the CLARITY Act advanced through House and Senate committees, though it had not reached a floor vote as of August 2026, leaving Chair Paul Atkins’s SEC to signal it is ready to write rules.

The grey zones are where smart accounts add features that look like services. A paymaster that fronts gas, a recovery provider that holds a guardian key, or a cloud service that stores a passkey backup all introduce a third party into what was a purely self-custodial relationship, and regulators have not fully mapped where custody or money-transmission rules attach. Compliance obligations such as the Travel Rule fall on exchanges and other intermediaries, not on the protocol or a self-custodied contract, a line our explainer on the FATF enforcement gap draws out. The safe reading for now: a smart account you alone control is still yours in the eyes of US regulators; add a helper who holds a key or fronts a fee, and the picture gets more complicated.

What to look for in a smart-account wallet

The features that distinguish smart accounts are also the features that determine how much you are trusting. A short checklist helps.

• Recovery model: does the wallet use guardians, a time delay, or a cloud passkey backup, and who could collude to seize the account?
• Signature transparency: does it support clear signing (ERC-7730) so a hardware device shows what you are approving, not just a hash?
• Standard and portability: is it built on ERC-4337, EIP-7702, or both, and can you move or revoke the delegation later?
• Gas model: can you pay in a stablecoin, and what surcharge does the paymaster charge?
• Modules and audits: which ERC-7579 or ERC-6900 modules are enabled, and have they been independently audited?
• Exit rights: can you always withdraw and revert to a plain account without the provider’s cooperation?

None of these questions requires deep technical knowledge, and all of them separate a wallet that gives you smart-account convenience from one that quietly reintroduces a middleman. As native account abstraction moves from proposal to protocol over the next year, expect these choices to get simpler, but expect them to still be choices.

The bottom line

Smart-account wallets have already won the argument that a crypto account should be programmable. The two workarounds that carried the idea to tens of millions of accounts, ERC-4337 and EIP-7702, proved the demand and exposed the flaw: a design that leans on bundlers and relays is fast but fragile. The race now underway to make account abstraction native, contested by the Ethereum Foundation, Coinbase’s Base and Paradigm, is a fight over who removes that fragility and on whose terms. For once, the timeline is short. Base ships in September, the Foundation is aiming for a fork within a year, and the wallet in your pocket is likely to become a smart account whether you asked for it or not. The choice that remains is whose rules it runs.

Frequently Asked Questions

What is a smart-account wallet?

A smart-account wallet is a crypto wallet whose address is controlled by a smart contract rather than a single private key. That lets it enforce custom rules: batching several actions into one transaction, letting an app or paymaster cover gas, accepting a passkey or multiple signers, and offering recovery if a key is lost. It contrasts with an externally owned account (EOA), which can only sign and send with one fixed key.

What is the difference between ERC-4337 and EIP-7702?

ERC-4337 builds account abstraction entirely in smart contracts and off-chain infrastructure (bundlers, paymasters, a special EntryPoint), and works best for accounts created as smart accounts from the start. EIP-7702, live since Ethereum’s Pectra upgrade in May 2025, instead lets an existing EOA delegate to smart-account code, so a normal wallet gains smart features while keeping its address and history. Many wallets now use both.

Are smart-account wallets safe?

They can be safer than a single-key wallet because features like spending limits, scoped session keys and social recovery contain the damage from one mistake. But programmability adds attack surface, and the biggest risk is approving a transaction you do not understand, the blind signing that enabled the $1.5 billion Bybit hack in 2025. Choosing a wallet that supports clear signing (ERC-7730) and independently audited modules matters more than the brand.

Do I still need a seed phrase with a smart account?

Often no. Thanks to RIP-7212, which lets Ethereum verify passkey signatures cheaply, wallets such as Coinbase’s Base Account let you create and use a self-custodied account with only a passkey stored in your phone’s secure hardware. Recovery is then handled by guardians, a cloud backup or a time delay rather than a written seed phrase, though the exact model, and who could access it, varies by wallet.

Will smart accounts become the default on Ethereum?

That is the direction. Native account abstraction would build smart-account support into Ethereum itself, and three proposals are competing to define it: the Ethereum Foundation’s EIP-8141, Coinbase and Base’s EIP-8130 (shipping on Base in September 2026), and Paradigm’s Tempo. None has a locked place in a specific Ethereum upgrade yet, but Buterin has said the feature could arrive within about a year.

By Yuki Tanaka, senior wallets and infrastructure correspondent, HOGE Wire.