Ethereum co-founder Vitalik Buterin published a technical note on the social platform X late on Saturday, 5 September 2026, arguing that recent work on Ethereum transaction formats has produced a clearer split between transaction actions and dependencies, and that Ethereum Improvement Proposal (EIP) 8141 is the natural interface for both native account abstraction and later scaling work.
The note is not a new proposal filing. It is an interpretation of design work already underway around Frame Transactions, and it lands days after core developers advanced the proposal within the schedule for a future network upgrade.
Buterin’s main note, posted at 23:20:53 Greenwich Mean Time (GMT), followed an earlier update about 31 minutes prior in which he wrote that “a lot of important progress on Frames (EIP-8141) has been quietly happening over the last few months,” and pointed readers to the official specification and to a 5 September 2026 essay by the pseudonymous Ethereum developer lightclient.
In a follow-up at 23:27:19 GMT, Buterin added that “I think 8141 gets close to optimal” and described a transaction as “just a series of calls,” while noting that the public transaction pool, known as the mempool, would still need to remain more restrictive to resist denial-of-service abuse.
Actions and Dependencies
In the main post, Buterin defined an action as an effect a transaction has, such as a call that sends Ether (ETH). He defined a dependency as a fact about the transaction or the state that must be true for the transaction to be valid.
His examples of dependencies were a signature, a Merkle proof of an Unspent Transaction Output (UTXO), and a Zero-Knowledge Succinct Non-Interactive Argument of Knowledge (SNARK) or a Zero-Knowledge Scalable Transparent Argument of Knowledge (STARK).
According to the post, dependencies can be processed in parallel. State-touching dependencies can be reasoned about by a mempool when the accessed state is statically declared. Pure dependencies, meaning those that cannot access state, can be processed once at the mempool layer and potentially replaced by a STARK, which would allow both execution and the underlying data to be omitted later on.
Buterin argued that actions and dependencies can in principle all be expressed as calls, including calls to precompiles, which are built-in low-level contracts on the network. That framing would leave a minimal transaction format: a list of calls, flags marking whether each call is static or pure, and a small set of fields such as origin and nonce. He wrote that this approach would maximize cross-compatibility even where different Ethereum Virtual Machine (EVM) chains support different features.
He contrasted the direction with the 2015-era Ethereum design, when serial execution and single-key Elliptic Curve Digital Signature Algorithm (ECDSA) accounts were sufficient. Current scaling work, he said, requires moving past that model because a fully dynamic execution and state layer is “not friendly to scaling.”
He estimated that more than 90% of Ethereum activity by volume does not need that level of dynamism, and proposed that contracts, accounts and transactions should declare what is dynamic and what is statically analyzable, with the more restrictive and analyzable path receiving the lowest gas cost. He described the direction as a mixture of the original Ethereum model and a more Bitcoin-like model, and noted that Bitcoin has effectively had account abstraction from its earliest days.
What EIP-8141 Actually Specifies
EIP-8141 is titled “Frame Transaction.” The official draft, maintained by the Ethereum Foundation, was created on 29 January 2026 and lists ten authors: Buterin, lightclient, Felix Lange, Yoav Weiss, Alex Forshtat, Dror Tirosh, Shahaf Nacson, Derek Chiang, Toni Wahrstätter and Stavros Vlachakis. The proposal is in Draft status on the Standards Track under the Core category. Discussion is hosted on the Ethereum Magicians forum.
The specification defines a new EIP-2718 transaction envelope with type identifier 0x06. Published constants in the current draft include an intrinsic cost of 12,000 gas, an additional 475 gas per frame, a maximum of 64 frames per transaction, and a designated entry point at address 0xaa.
Frames are purpose-labeled calls. VERIFY frames handle authentication and payment approval, SENDER frames execute user operations, and DEFAULT frames handle deployment or post-operation logic. The original 29 January 2026 draft used a higher intrinsic cost of 15,000 gas and a maximum of 1,000 frames, indicating that key parameters have already tightened during review.
Account Abstraction Lineage
Account abstraction, the goal of letting Ethereum accounts define their own validation logic rather than relying on a single hardcoded signature scheme, dates on the network to EIP-86, filed in 2016. The more recent EIP-7701, titled “Native Account Abstraction,” was withdrawn as superseded by EIP-8141.
On 28 February 2026, Buterin described EIP-8141 as “an omnibus that wraps up and solves every remaining problem that AA was intended to address (plus more).” In a later post on 23 March 2026, he argued that frames also make privacy protocols more first-class because they move general-purpose programmability into the verification stage, which previously relied on hardcoded ECDSA. The Crypto Times covered that argument in Ethereum’s Quantum Defense? Vitalik Buterin Backs EIP-8141 Upgrade.
Hegotá Scheduling and Open Caveats
Core developers moved EIP-8141 from Considered for Inclusion (CFI) to Scheduled for Inclusion (SFI) for the Hegotá upgrade during All Core Developers Execution (ACDE) call 244 on 27 August 2026, after it had been marked CFI at ACDE 233 on 26 March 2026. Forkcast, the community tracker for Ethereum upgrades, records that timeline. Hegotá is planned for 2027, following the Glamsterdam upgrade targeted for the fourth quarter of 2026.
Independent call notes by researcher Christine Kim state that SFI does not freeze the specification, that competing Layer 2 work around EIP-8130 has raised fragmentation concerns, and that some participants said formal ratification language still needed to be made explicit.
Related Research Already on Record
Several pieces of related research named in the 5 September post already exist as primary documents. EIP-8250, titled “Keyed Nonces for Frame Transactions,” was created on 16 April 2026 by Thomas Thiery, Toni Wahrstätter, lightclient, and Buterin. It replaces a single linear sender nonce with independent nonce keys so that transactions on disjoint non-zero keys are replay-independent.
Discussion is hosted at the Ethereum Magicians forum. Buterin’s 5 May 2026 explanation included a worked example of 500 billion in-protocol nullifiers over eight years at 2,000 transactions per second.
Native UTXO-style state has been discussed as a more restrictive, lower permanent-state option for payments. In a 16 August 2026 post, Buterin argued that Ethereum should keep “the best of UTXO-style state, dynamic state, and everything in between,” and credited Bitcoin’s Utreexo accumulator research. The Crypto Times followed that discussion in Vitalik Buterin Credits Bitcoin’s UTXO Ideas as Ethereum Explores Hyperscaling.
A recursive-STARK mempool design, in which nodes replace individual validity proofs with recursive proofs of sets of objects, was posted to the Ethereum Research forum on 12 January 2026. A later direction, describing signatures and STARKs as replaceable dependencies, appears in GitHub pull request 11772.
On 4 July 2026, Buterin’s Lean Ethereum note placed new state types and recursive STARK verification inside a three-to-four-year rebuild, sketching a 2030 profile of roughly 2 terabytes of present-day dynamic state and about 100 terabytes of newer, more restrictive state, a plan The Crypto Times covered in Vitalik Buterin Unveils Lean Ethereum Roadmap for Next Era.
What the Post is Not
The 5 September comment should be read against that record rather than as a shipping announcement. EIP-8141 is not live on Ethereum mainnet, and gas figures, frame limits, and opcode lists have already changed since January 2026. Public mempool admission is expected to remain conservative at first. Fork Choice Enforced Inclusion Lists (FOCIL), tracked as EIP-7805, remains the separately scheduled inclusion-list headliner for Hegotá. Buterin previously described the two proposals as complementary in a 19 February 2026 post. The Crypto Times reported on the FOCIL track in Ethereum Advances FOCIL for Upcoming Hegota Fork.
The analytical point in the weekend note is a narrow one. EIP-8141 is being treated as a general-purpose transaction envelope on which later state types and proof-based mempool compression can sit. Whether that envelope remains the Hegotá design, and whether the split between action and dependency becomes an explicit protocol rule rather than a design metaphor, still depends on specification work that the EIP itself labels unfinished.
Also Read: Arthur Hayes Predicts Ethereum Will Hit $10,000 by Year-End
