Optimum, an MIT-incubated networking startup, announced on October 20, 2026 that its mump2p block-propagation protocol is live on Ethereum mainnet. Sixteen node operators — managing approximately 22,000 validator keys and more than 1.4 million staked ETH, or roughly 3% of all staked supply — are ...
"We don't mandate new infrastructure across our entire validator set unless we're confident it raises the bar for our stakers and for the network." — Rok Kopp, Co-founder & Chief Growth Officer, ether.fi
Optimum, an MIT-incubated networking startup, announced on October 20, 2026 that its mump2p block-propagation protocol is live on Ethereum mainnet. Sixteen node operators — managing approximately 22,000 validator keys and more than 1.4 million staked ETH, or roughly 3% of all staked supply — are now running the software. Liquid restaking provider ether.fi has made mump2p mandatory for all 12 of its node operators. Named participants include Blockdaemon, Everstake, P2P.org, Stakefish, and Bitcoin Suisse.
Optimum's self-reported metrics claim blocks arrive 2.42 times faster than via Ethereum's default GossipSub protocol, with 7.88 times lower arrival-time variance. In nine out of ten measurements, mump2p users received the block before GossipSub delivered it. However, independent testing conducted by StereumLabs for the Lido DAO found results "not statistically significant" across most consensus clients. The gap between vendor claims and third-party verification is the central tension in this story.
Ethereum's peer-to-peer networking layer uses GossipSub, a protocol where nodes forward messages to a randomized subset of peers. Under normal conditions, a block takes approximately 500 milliseconds to one second to reach most validators across the network. The protocol operates within Ethereum's fixed 12-second slot window, during which a proposer must broadcast a block and attesters must vote on it.
The economics of this window are unforgiving. Approximately 84.4% of validator rewards come from attestations, according to the Ethereum specification. Validators that receive blocks late — even by hundreds of milliseconds — risk attesting to the wrong head, missing the optimal attestation window, or in the worst case, missing the slot entirely. A missed attestation costs exactly as much as a successful one pays, creating a direct penalty-to-reward symmetry that makes propagation speed economically material.
A July 2026 disclosure revealed a vulnerability in GossipSub where a single crafted PRUNE message from an unauthenticated peer could crash a validator node. While patched, the incident highlighted the fragility of a networking layer that the entire staking infrastructure depends on.
Ethereum's 1.2 million-plus active validators operate across more than 80 countries. For geographically distributed operators, GossipSub's multi-hop relay model creates systematic latency disadvantages. Operators in regions farther from network-dense areas — typically North America and Western Europe — face structurally higher propagation times, which translates into structurally lower reward capture.
Optimum's mump2p protocol is built on Random Linear Network Coding (RLNC), a data transmission technique developed by MIT professor Muriel Médard over two decades of academic research. Instead of relaying identical copies of a block from peer to peer, RLNC divides the block into fragments and transmits coded combinations. Receiving nodes can begin forwarding partial information before assembling the complete block. A node needs only enough linearly independent pieces — not the original fragments — to reconstruct the full message.
The implementation runs as a sidecar gateway alongside existing Ethereum consensus clients (Prysm, Lighthouse, Nimbus, Teku, and Lodestar). It operates on top of libp2p, the networking toolkit Ethereum clients already use. Ethereum's default GossipSub continues running in parallel, meaning mump2p adds a supplementary propagation channel rather than replacing the existing one. Consensus rules, execution logic, and signing keys remain unchanged.
Optimum raised $11 million in a seed round led by 1kx in mid-2025, with participation from Robot Ventures, CMT Digital, Spartan, Finality Capital, and others. The company's angel investors include the Polychain CTO, the DeFiance Capital founder, Bitget's CEO, and a co-founder of Polygon.
As of October 20, 2026, Optimum reports the following deployment metrics:
| Metric | Value | |---|---| | Node operators running mump2p | 16 | | Approximate validator keys | 22,000 | | Staked ETH under management | 1.4 million+ | | Share of total staked ETH | ~3% | | Self-reported block arrival speed vs. GossipSub | 2.42x faster | | Self-reported arrival variance reduction | 7.88x lower | | Self-reported block win rate | 9 out of 10 measurements | | Self-reported "Total Value Boosted" | ~$3.7 billion |
Ether.fi's mandate covers all 12 of its node operators. The remaining four operators joined independently. Named operators beyond ether.fi's set include Blockdaemon, Everstake, P2P.org, Stakefish, and Bitcoin Suisse — all institutional-grade infrastructure providers.
The "Total Value Boosted" metric of $3.7 billion is an Optimum-defined figure whose methodology has not been independently verified. It appears to reference the market value of ETH managed by participating operators, not a direct revenue or fee metric.
The most rigorous public test of mump2p was conducted through Lido's governance process, where Optimum submitted a PERCH (Permissionless Evaluation Request for Client Heterogeneity) proposal requesting that Lido Curated Module operators evaluate the protocol on the Hoodi testnet.
StereumLabs, a Lido node operator, ran an independent evaluation across 36 mainnet full nodes over 11.5 days. Their findings diverged from Optimum's claims:
Multiple Lido-affiliated operators — including InfStones, Ebunker, HashKey Cloud, Blockdaemon, Stakely, FP Validated, Stakefish, and SenseiNode — participated in the evaluation. HashKey Cloud reported approximately 200 milliseconds lower average attestation propagation times, supporting Optimum's claims in that specific case.
Optimum acknowledged that mainnet testing conditions may have confounded results and announced plans for a controlled setup where "only registered operators receive acceleration" after October 26 for cleaner comparisons.
The divergence between Optimum's self-reported metrics and StereumLabs' independent findings remains unresolved. No third-party audit of the mainnet deployment has been published as of October 20.
Optimum's licensing model introduces a structural tension with Ethereum's open-source ethos.
The Optimum Gateway — the sidecar that connects consensus clients to the mump2p mesh — is published under the Microsoft Reference Source License (Ms-RSL). It can be read and reviewed, but the license restricts modification and redistribution in ways that differ from standard open-source licenses.
The core RLNC library is source-available but explicitly cannot be forked or reused. Optimum holds an exclusive license to the RLNC technology in Web3, obtained through its MIT relationship. The company has stated it pays licensing fees to the university for this exclusivity.
This creates a dependency chain: if mump2p adoption grows significantly — from the current 3% of staked ETH toward, say, 20% or 30% — Ethereum's block propagation would become partially reliant on a single company's proprietary library. As one analysis noted: "As more operators route traffic through the same third-party gateway, the network grows more reliant on one company's software working correctly."
The current deployment is permissioned, requiring API keys from Optimum to join the mump2p network. This is standard for early infrastructure rollouts but adds another centralization vector. Optimum has indicated plans to expand beyond Ethereum to other high-throughput blockchains.
Médard herself has acknowledged the broader engineering concern, stating that hardwiring dependency on any single system "creates fragility and dependence" — a statement made in reference to GossipSub but equally applicable to any dominant propagation layer.
Mump2p is currently free for operators. Optimum has not announced a pricing model. The $11 million seed round provides runway, but a venture-backed infrastructure company will eventually need revenue.
The economic value at stake is calculable. Ethereum validators earn 3.9-5.1% APR in 2026, with 84.4% of rewards tied to attestation performance. For the 1.4 million ETH currently under mump2p management — worth approximately $3.3 billion at October 2026 prices near $2,400 — even a marginal improvement in attestation timing across the entire stake could translate to millions of dollars in incremental annual rewards.
The question is whether Optimum captures a portion of that value, and if so, how. Possible models include subscription fees per validator key, a percentage of incremental rewards, or a freemium structure where basic propagation remains free but advanced features (priority routing, analytics dashboards) carry a cost.
For operators managing thousands of validator keys — particularly liquid staking protocols like ether.fi that compete on yield — propagation speed is a competitive variable. An operator that consistently receives blocks 200 milliseconds earlier captures attestation rewards more reliably, compounding into meaningful yield differences across a portfolio of 22,000 keys over 12 months.
This dynamic could create a two-tier network: operators with mump2p access and those without. If the performance gap is real and durable, economic gravity will push adoption. If the gap is marginal — as StereumLabs' testing suggests for some clients — the vendor lock-in may not justify the dependency.
Optimum's mainnet launch represents an infrastructure bet on a proposition that has not yet been independently validated at scale: that Ethereum's networking layer is a bottleneck worth solving with proprietary technology. The 3% staked-ETH footprint is small. Ether.fi's mandate is notable but represents one protocol. The technology's performance claims require third-party verification before the market can price them.
The deeper question is structural. Ethereum's value proposition rests on permissionless, open infrastructure. A propagation layer that delivers measurable advantages but cannot be forked introduces a familiar tension in open-source ecosystems — one that intensifies in proportion to adoption. At 3% of staked ETH, this is a product announcement. At 30%, it would be an infrastructure dependency.