Optimum Says mump2p Covers 7.8% of Staked Ether on Ethereum

Optimum said Nov. 3 that operators using its mump2p block-delivery service represented 7.8% of staked Ether (ETH), broadening the service’s reach as independent mainnet testing offers limited evidence of faster delivery.

That share rose from more than 3% at the service’s Oct. 20 mainnet launch. At launch, 16 node operators with about 1.4 million ETH in connected stake were using it, covering roughly 22,000 validator keys. The Nov. 3 update gave a stake share but no new operator count.

mump2p gives validators another path for block data alongside Ethereum’s GossipSub network. Its gateway runs beside existing Ethereum software and distributes coded pieces of a block without changing the network’s consensus rules. Receiving a block sooner could leave a validator more time to process it and submit an attestation, a vote on the chain head.

Optimum co-founder Muriel Médard said Nov. 9 that about 90% of blocks were reaching participating validators roughly twice as fast. The claim concerns block delivery to those validators; it does not establish an improvement in transaction processing across Ethereum. An earlier block arrival also may not translate into a gain after the validator’s software processes it.

An independent test ran from Aug. 27 through Sept. 7, before the Oct. 20 launch, using an early gateway build on 36 Ethereum mainnet full nodes. Some software clients recorded earlier arrivals, but none of the client results met the test’s 0.05 statistical-significance threshold. Some arrival gains did not persist after client processing.

The test could not assess votes or rewards because no validators were attached to its nodes. Its timing and early software version limit what its results can establish about the current deployment.

A separate Optimum evaluation involving Lido operators examined 180,327 slots on the Hoodi testnet from Sept. 17 to Oct. 15. The 95th-percentile arrival time was 1,207 milliseconds at mump2p gateways, compared with 1,614 milliseconds at separate public Xatu sentry nodes. The 407-millisecond difference compares those two groups’ percentile readings, not paired block deliveries to the same validator.

Hoodi’s conditions differ from Ethereum mainnet’s. The testnet comparison therefore does not establish the same arrival-time difference on mainnet or an effect on validator performance.