How New L1 is built

The big idea: agree on order first, run later

Most chains run every transaction before they agree a block is valid. New L1 splits these jobs:

1. Ordering. Validators agree on the exact order of transactions, without running them.
2. Execution. The transactions run shortly afterwards, in the background.
3. Commitment. The results are published in a later block and voted on.

Because a fixed order fully determines the outcome, the docs put it this way: execution reveals the state rather than deciding it. This means consensus never has to wait for heavy computation, which keeps blocks fast.

The gap between ordering and execution adapts to load. It shrinks when the network is quiet and grows when it is busy, up to a limit set by governance.

Four stages of a block

Every block moves through four states: ordered, executed, committed and execution-finalized. Ordering alone never finalizes a block. Finality needs the execution result committed on-chain and voted on.

Consensus

An enhanced version of Parlia, BNB Chain's proof-of-staked-authority protocol, with BLS votes across blocks for fast finality. See the article on blocks and pre-confirmation.

Execution

A parallel EVM that runs standard Ethereum-style contracts. Multi-Lane sits on top to reserve capacity for selected apps.

Storage

Instead of the Merkle-Patricia trie used by Ethereum and BSC, New L1 keeps state in a flat key-value store and commits to it with a cumulative hash (called LtHash). Updating it costs the same no matter how large the state grows.

The trade-off: tools that rely on trie proofs, such as eth_getProof and snap sync, are not available.

Networking

There is no global mempool. Transactions are sent straight to the current and next block producers. Blocks and smaller "sub-blocks" are gossiped across the network.