Ethereum co-founder Vitalik Buterin has shared three upcoming technical changes that may shape the network’s future. These updates focus on tightening limits within the protocol to improve security and performance.
Ethereum’s recent development roadmap shows a clear shift toward enforcing limits on transaction behavior. Buterin explained that new protocol caps aim to reduce attack surfaces and improve client efficiency.
This process started with earlier updates. In 2021, Ethereum implemented EIPs 2929 and 3529, which increased gas costs for certain operations and reduced gas refunds. This helped cut down on disk overloads and reduced spam attacks. Later, in 2024, the SELFDESTRUCT function was restricted under the Dencun upgrade, closing another complex vulnerability.
Now, in 2025, a hard cap of 16,777,216 gas per transaction is in place. This change blocks oversized transactions that could lock up nodes or cause unpredictable behavior. As Buterin noted, these limits create a network where “worst-case behavior is strictly bounded.”
Contract Code Access to Be Capped
One of the changes involves limiting the number of contract code bytes accessed during a single transaction. Under current conditions, large contracts can slow down nodes and increase costs.
Buterin suggested that in the short term, interacting with large smart contracts will become more expensive. Over time, this should lead developers to use more efficient structures, such as binary trees, and push for chunk-based pricing. This shift would create more uniform contract sizes and reduce strain on the network.
Boundaries for ZK-Proof Processing
Another planned update targets the use of zero-knowledge (ZK) proofs, which are increasingly used by layer-2 solutions. Buterin raised concerns that, without boundaries, complex proofs could slow down block processing.
The proposed change would set limits on ZK-EVM prover cycles. This would prevent block builders from packing blocks with resource-heavy proofs and reduce risks of network congestion.
“Repricing proofs are becoming more and more important,” Buterin said.
Adjustments to Memory Usage Costs
The third change involves setting clearer limits on EVM memory use. While memory expansion is already controlled to some degree, Buterin believes further changes are needed.
He explained that attackers can still push client software into unstable conditions. A hard cap on memory would simplify client execution engines and help teams model worst-case scenarios more effectively.
