A research project at Nethermind measuring how Ethereum execution clients behave when the chain’s state is many times larger than today’s mainnet. We continue an existing community “Bloatnet” — a devnet whose state was already grown to roughly 3.5× mainnet — and push it further, up to 10×, benchmarking the things that decide whether the network can safely raise the gas limit and move toward statelessness.

Where it stands

We took the existing ~3.5× Bloatnet state and grew it through 5×, 7×, and the headline 10× target (2.22 TB, June 2026) — steering the state’s shape (the mix of accounts, storage, and contract code) to match real mainnet, so each size is a fair comparison. Real mainnet (1×) is our measured reference point, not a state we generated. The cross-client benchmark round comes next.

The team

• Marcin Sobczak — lead
• Daniil Ankushin — bloating tooling, measurement, analysis
• Dmytro Belitskyi — benchmarking and client testing

Building on prior work

The ~3.5× state we start from is the Bloatnet initiative — a community effort by the Stateless Consensus team and EthPandaOps, who built the bloated devnet and the spamoor transaction tool used to grow it (public write-up by Carlos Pérez). This project takes that state and carries it all the way through 5×, 7×, and 10× — with mainnet-faithful composition and cross-client measurement.

Mainnet, in numbers

The baseline everything is measured against — real Ethereum mainnet, May 2026:

That last number — the size of the cryptographic proof a stateless client needs per account — is the figure any future upgrade (like Verkle trees) has to beat. Capturing it across every state size is one of the project’s main deliverables.

Why it matters

Follow along

The project lives in the open on GitHub. Results and tooling are published as each milestone lands; the Docs page explains how the rig works, and Results shows the measured numbers so far.

• Code & data — github.com/NethermindEth/state-benchmarks
• Questions or critique — open a GitHub issue