Ethereum is fighting a war on two fronts. While ETH suffers its worst year-to-date performance on record — down 34% to approximately $2,000, with DeFi TVL sinking to a three-month low of $51.7 billion in early February — co-founder Vitalik Buterin has unveiled the most ambitious architectural res...
"As we keep scaling, the centralisation force towards sophisticated actors is increasing, and FOCIL prevents these actors from censoring transactions." — Jihoon Song, Ethereum Foundation Researcher
Ethereum is fighting a war on two fronts. While ETH suffers its worst year-to-date performance on record — down 34% to approximately $2,000, with DeFi TVL sinking to a three-month low of $51.7 billion in early February — co-founder Vitalik Buterin has unveiled the most ambitious architectural reset since Ethereum's proof-of-stake transition. The plan spans two interconnected initiatives: a "cypherpunk principled" protocol rebuild centered on censorship resistance and zero-knowledge proofs, and a radical AI governance framework that would deploy personal AI agents as voting proxies across decentralized organizations.
These are not distant research proposals. FOCIL — Fork-Choice Enforced Inclusion Lists — was officially scheduled for the Hegotá hard fork in H2 2026 at the most recent All Core Devs meeting. The AI governance proposal, published February 21, lays out a production-ready architecture for replacing human voters with AI stewards trained on individual values. Together, they represent Buterin's answer to Ethereum's dual crisis: a base layer losing its neutrality to MEV-extracting block builders, and a governance model collapsing under voter apathy.
The economic stakes are enormous. Ethereum still commands approximately 57-68% of all DeFi total value locked, making it the settlement layer for the majority of on-chain economic activity. Whether these proposals succeed or fracture the validator community over sanctions compliance will determine whether Ethereum retains that dominance.
Buterin's vision is not to scrap Ethereum and start over. Instead, the plan calls for bolting a new, deeply interoperable layer onto the existing architecture — what he describes as building a "cypherpunk principled non-ugly Ethereum" that gradually strengthens core properties: censorship resistance, zero-knowledge compatibility, and lean consensus design.
The centerpiece of this initiative is FOCIL (EIP-7805), now confirmed for inclusion in the Hegotá hard fork scheduled for H2 2026. FOCIL is an enforcement mechanism: if a proposed block ignores valid transactions sitting in inclusion lists, the chain can fork away from it. The result is what Buterin calls "guaranteed rapid inclusion" — virtually any valid transaction would be included within one to two slots, even in adversarial conditions.
Layer 2 developer Tim Clancy described FOCIL as the "single most important [proposal] for Ethereum," arguing it "delivers a capability that Ethereum must have to continue delivering on its mission of being the most neutral blockspace."
Paired with FOCIL is EIP-8141, an account abstraction upgrade also slated for Hegotá. This embeds smart wallet logic directly into Ethereum's base layer — multisig security, transaction batching, custom validation, and gas-sponsored privacy transactions — eliminating the need for intermediary relayers. Together, FOCIL and EIP-8141 represent a fundamental hardening of the base layer ahead of what Buterin envisions as a five-year architectural migration.
The Ethereum Foundation formalized its 2026 development priorities into three tracks, announced on February 20:
Scale: The Foundation merged its previous "Scale L1" and "Scale Blobs" initiatives into a single coordinated effort. The target: push Ethereum's gas limit "toward and beyond" 100 million, up from the current 60 million (which was itself doubled from 30 million during the 2025 Pectra and Fusaka upgrades). This track also oversees continued blob parameter expansion for L2 data availability and the delivery of scaling components within the Glamsterdam network upgrade, expected in H1 2026.
Improve UX: A dedicated track focused on native account abstraction (EIP-7701 and EIP-8141), enabling smart contract wallets to operate like standard accounts. The Foundation is also expanding the Open Intents Framework, which aims to streamline asset transfers and liquidity movement across rollups — tackling a fragmentation problem that has plagued the L2 ecosystem.
Harden the L1: The most philosophically charged track. Beyond FOCIL, this includes post-quantum cryptography research (a gradual transition away from ECDSA toward quantum-secure signatures), execution-layer safeguards, expanded testing infrastructure, and censorship resistance research for both transactions and blob data.
Two major hard forks will deliver these changes: Glamsterdam in H1 2026 (up to 22 EIPs focused on L1 scalability) and Hegotá in H2 2026 (targeting FOCIL, native account abstraction, higher gas limits, and post-quantum readiness). The Foundation also disclosed leadership changes: Tomasz Stańczak stepped down as co-executive director, with Bastian Aue assuming interim duties.
FOCIL's most dangerous implication is legal, not technical. By forcing validators to include all valid transactions, the protocol would override the voluntary censorship that currently keeps U.S.-based validators out of regulatory crosshairs.
The data is stark: when Tornado Cash faced OFAC sanctions, approximately 90% of Ethereum validators refused to include related transactions. FOCIL would eliminate that choice. Under U.S. sanctions law, validators forced to process transactions from sanctioned addresses could face prosecution, even when acting in good faith.
Privacy Pools founder Ameen Soleimani has emerged as the most vocal critic: "When you create an EIP to solve the problem of 'filtering out transactions with sanctioned addresses' and your solution is 'to allow validators to impose constraints on builders by force-including transactions in their blocks'... we have a problem, a big problem."
The counter-argument, articulated by Ethereum Foundation researcher Jihoon Song, frames this as existential: as Ethereum scales and block building centralizes further, without mandatory inclusion enforcement, a handful of sophisticated actors could selectively censor any transaction. The theoretical defense — that a single willing validator can still settle any transaction — offers cold comfort to institutional operators evaluating compliance risk.
This is a defining tension. Ethereum is attempting to hardcode censorship resistance at the protocol level while an increasing share of its validator set operates within U.S. jurisdictional reach. The two objectives may prove irreconcilable.
On February 21, Buterin published a governance framework that would fundamentally replace human participation in DAO voting with AI agents. The proposal addresses what he identifies as DAOs' core structural flaw: "limited human attention."
The data supports the diagnosis. Studies show DAO participation rates averaging 5-15% of token holders, with Chainalysis data revealing that just 1% of holders control 90% of voting power across ten major DAOs. Jupiter DAO paused governance entirely, citing "negative feedback loops and community division." The current model is broken.
Buterin's architecture has four components:
1. Personal Governance Agents: AI models trained on an individual's writing, conversations, and stated values. The agent votes autonomously on routine proposals, escalating only high-stakes or ambiguous decisions to the human for review.
2. Public Conversation Agents: Aggregation systems that summarize individual views from many participants, identify commonalities, and convert private positions into shareable formats without exposing private data — an LLM-enhanced version of Polis-style collective intelligence tools.
3. Suggestion Markets: Prediction markets where participants submit proposals and AI agents bet on tokens. Acceptance triggers payouts, creating financial incentives for quality contributions and a spam-filtering mechanism.
4. Privacy-Preserving Multi-Party Computation: For sensitive decisions, personal LLMs submit inputs into cryptographic "black boxes" that process private data without compromising secrecy. Zero-knowledge proofs would be integrated into all governance tools.
The economic implications are significant. If operational, this framework would remove the principal-agent problem inherent in token delegation — where small groups of whales or professional delegates accumulate decision-making power through voter apathy. It would also dramatically reduce the cost of governance participation, potentially unlocking the dormant voting power of millions of passive token holders.
Beneath the near-term upgrades lies Buterin's most radical long-term proposal: replacing the Ethereum Virtual Machine (EVM) with RISC-V architecture. First floated in April 2025, the proposal would have developers compile Solidity and other languages directly to RISC-V, with system operations like SLOAD and CALL handled as syscalls instead of opcodes.
The efficiency argument is compelling: the switch could improve execution efficiency by up to 100x and simplify proof generation, since most zero-knowledge provers already simulate the EVM by translating it into RISC-V internally. Backward compatibility would be maintained by running an EVM interpreter within the RISC-V environment for legacy contracts.
Buterin has framed this as a five-year project — comparable in scope to the Beam Chain consensus layer overhaul — and noted that AI-assisted coding could accelerate the timeline. If successful, the combination of RISC-V execution, native ZK verification, and FOCIL-enforced inclusion lists would produce a fundamentally different blockchain than the one running today.
This architectural ambition arrives during Ethereum's most challenging market period. ETH has fallen approximately 34% year-to-date to around $2,000 — roughly 60% below its level from six months ago, marking its worst start to a year on record. DeFi TVL touched a three-month low of $51.7 billion on February 6, recovering only to around $55.5 billion by mid-February, erasing nearly $20 billion from mid-November levels.
Bitcoin ETFs have seen persistent outflows — $165.8 million on February 19 alone, led by BlackRock's IBIT shedding $164.1 million. Capital is rotating from BTC and ETH products toward Solana and XRP ETFs. Broader macro headwinds continue: President Trump's escalation of worldwide tariffs to 15% announced February 21 has further dampened risk appetite.
Yet Ethereum retains structural dominance: approximately 57-68% of all DeFi TVL and the primary settlement layer for institutional on-chain activity. The question is whether Buterin's technical roadmap can sustain that position through a prolonged bear market.
Vitalik Buterin is betting that Ethereum's survival depends on becoming maximally resistant to censorship, maximally efficient at execution, and maximally inclusive in governance — simultaneously. The cypherpunk rebuild addresses the first two. The AI stewards proposal addresses the third. The RISC-V endgame unifies them.
The risk is fragmentation. FOCIL may force U.S.-based validators to choose between protocol compliance and legal compliance. AI governance agents introduce attack surfaces that have never been tested at scale. And the five-year RISC-V migration could stall if developer attention fractures across too many parallel workstreams.
But the alternative — an Ethereum that grows more centralized, more censorable, and more apathetic in its governance — is the scenario Buterin clearly fears most. With ETH at $2,000 and DeFi TVL eroding, the window for bold architectural decisions may be closing. This week's proposals suggest Buterin intends to move before it does.