Ethereum’s multi-party block construction goes live on mainnet

Ethereum blocks have traditionally been a one-builder affair. One entity assembles the transactions, one entity wins the auction, one entity gets the slot. As of September 16, that changed. Multi-party block construction, or MPBC, is now running on the Ethereum mainnet, allowing multiple builders to contribute transactions to a single block without requiring any changes to the underlying protocol.

The initiative was coordinated by the Blockspace Forum, which spent much of 2026 herding an impressive coalition of block builders toward this moment. The early numbers are modest but meaningful: roughly 3% of Ethereum blocks in the first 24 hours were constructed using the multi-party approach, and more than 20,000 transactions have been included faster since testing began in late August.

What MPBC actually does

To understand why this matters, you need to understand how Ethereum blocks get built today. After the Merge in 2022, Ethereum adopted a system called proposer-builder separation, or PBS. Validators propose blocks, but specialized builders actually assemble them. Builders compete in an auction, and the winner’s block gets included.

MPBC changes the model so that multiple builders can each contribute transactions to a single block, creating multiple inclusion paths rather than funneling everything through one entity. The result is blocks that are fuller, more diverse, and faster at confirming pending transactions.

The gains in the early data are striking. Some multi-party blocks have included up to 810 additional transactions compared to what a single builder would have produced. In certain cases, gas usage jumped nearly 14-fold.

How we got here

This didn’t happen overnight. The Blockspace Forum ran a series of workshops throughout 2026, culminating in a meeting in Cannes that brought together 32 teams. Those teams collectively account for over 95% of out-of-protocol block contributions on Ethereum.

Foundational research published in May 2026 laid out the design principles. The key constraint was that MPBC had to be strictly additive. No protocol changes, no hard forks, no consensus-level modifications. The system works entirely at the builder and operator layer.

Notably, the design doesn’t depend on relays for its core function. By sidestepping that dependency, MPBC introduces a parallel path that reduces the chokepoint risk.

Alongside the mainnet launch, the team released comprehensive public documentation and a live dashboard tracking multi-party block metrics. Performance indicators, participation rates, and transaction inclusion data are all visible in real time.

Why block construction centralization matters

The current PBS model has structural issues that compound over time. When one builder wins each auction, they effectively control transaction ordering within that block. Builders can prioritize transactions that generate profit for themselves, sandwich trades, front-run orders, or simply exclude transactions they don’t find economically interesting.

MPBC doesn’t eliminate these dynamics entirely, but it dilutes them. When multiple builders contribute to the same block, no single entity has full control over transaction ordering or inclusion. Service transactions get additional pathways into blocks rather than competing for space in a single builder’s priority queue.