Ethereum Co-founder Vitalik Buterin: Future Forks Will Fully Transition to Recursive STARK
Ethereum co-founder Vitalik Buterin, in a remote keynote speech at ETHShanghai 2026, highlighted the recursive STARK memory pool proposal EIP-8288 as a core topic, stating it aims to alleviate the conflict between quantum-resistant signatures, privacy transactions, and mainnet scalability.
Currently, quantum-resistant signature verification requires about 100,000 to 300,000 gas, while privacy transactions, with careful engineering, require about 300,000 gas. If quantum resistance is added, it could rise to about 10 million gas. He warned that if everyone verifies such objects directly on-chain, the mainnet throughput would drop from about 25 transactions per second to about 0.25 transactions. The proposal is to keep heavy computations in the memory pool: nodes collect dependencies and generate recursive proofs approximately every 500 milliseconds, with each block on-chain verifying only one STARK of about 100 to 300kb, plus about 96 bytes for each statement.
He hopes this draft will enter the I-star fork after Hegota and serve as the next step for Frame Transactions. The effects include reducing the cost of quantum-resistant privacy to tens of thousands of gas, allowing new signatures like Falcon and ML-DSA to not require changes to the Ethereum Virtual Machine, as long as the client first wraps a layer of STARK. Recursive proofs require a unified instruction set, with RISC-V currently preferred, effectively writing a standard ISA into the protocol. The foundation aims for comprehensive quantum resistance readiness by December 2029.
Block time and finality belong to another parallel route. The foundation's technical blueprint from early 2026 proposed gradually reducing block time from 12 seconds and finality from about 16 minutes to seconds-level block time and about 6 to 16 seconds for finality; Vitalik himself has described tiered slots of 8, 6, 4, 3, and even 2 seconds. The speech also mentioned that as proof costs decrease, the interval for layer two to submit to the mainnet is expected to shorten from several minutes to about one minute or less. He used artificial intelligence to illustrate that formal verification is accelerating, allowing more complex cryptographic systems to be machine-checked.
The market mechanism is the repricing of protocol rules. Buyers want quantum resistance and privacy without blocking mainnet applications and layer two, while sellers are clients that can perform aggregated proofs in the memory pool and the proof market. The event-driven aspect comes from the summit speech pushing existing drafts back onto the upgrade agenda. Beneficiaries are STARK provers and the RISC-V toolchain, while those under pressure are old wallets and sorters that still stuff heavy verification into the execution layer. Ethereum will not change block time on this day; instead, it will change what cryptography is written as default in the next one to two years' forks.
Source: Public Information
ABAB AI Insight
In the past year, Vitalik has shifted Ethereum's mainline focus from "throughput slogans" to "quantum resistance and privacy without blocking the mainnet." Zero-knowledge proofs have moved computation off-chain for the past decade, and recursive STARK writes this migration into the memory pool instead of adding precompiles to the virtual machine. Hegota narrows the scope, and I-star follows EIP-8288, breaking down major changes into schedulable forks to avoid another overly large merge upgrade.
Capital and engineering resources follow the proof market. Layer two has already settled based on validity proofs; if the mainnet mandates that each block only consumes one recursive proof, provers will transition from optional infrastructure to protocol necessities. RISC-V as a standard ISA means clients, audits, and formal tools will revolve around the same instruction set, and AI-accelerated verification will only strengthen this standardization. Money will flow towards hardware and software that can produce STARKs in real-time, rather than inflating another layer of "new virtual machine precompile politics."
Benchmarking against Bitcoin's minimal scripting, Solana's execution piled on a single high-performance machine, and various layer twos using different proof systems, Ethereum has chosen to slim the mainnet and thicken proofs. The industry phase is shifting from a scalability race to a cryptographic default value race: whoever first reduces the unit cost of quantum resistance and privacy to that of ordinary transactions will define the next round of account abstraction.
The structural change is a technological substitution. On-chain verification will no longer validate heavy objects one by one but will verify proofs that "verification has occurred." The mechanism is: quantum safety makes signatures and proofs larger and more expensive, and without aggregation, privacy and security will directly consume throughput; the recursive memory pool shifts costs from the execution layer to parallel node computation, allowing scalability, privacy, and quantum resistance to move in the same direction without having to choose between three options.
ABAB News · Law of Cognition
The most expensive security, if verified on-chain one by one, will first kill throughput.
Once the protocol specifies the proof format, provers become the new miners.
Fork scheduling is more decisive than white papers in determining which cryptography becomes default.