Ethereum researchers have demonstrated a way to spread new block data across the network in under one second — five times faster than today’s method — using a technique called segmented broadcasting that could make the network faster and smoother for everyone who uses it.
By Amir Hassan | September 17, 2026
The Hook: From Five Seconds to Under One
On September 17, Ethereum Research published new test results for EIP-8411, a proposal that changes how block data travels between computers running the network. In simulations, the median time for a 1 MiB block payload — think of it as the package of transactions inside a block — to reach half the network dropped from roughly five seconds to about 0.75 seconds, with the slowest cases improving from around six seconds to just over one.
Why should a regular investor care? Faster data spreading means every node sees new blocks almost at the same moment. That reduces the chances of tiny forks (two competing versions of the ledger), makes the chain more predictable for applications, and helps Ethereum stay quick and reliable even as usage grows. It is infrastructure work you never see, but you would feel it if it broke.
On-Chain Evidence: How Segmentation Actually Works
Today, Ethereum’s network forwards data in a “store-and-forward” pattern, like a relay race where each runner must receive the entire baton, inspect every inch of it, and only then pass it on. Because a full block payload must completely cross each network hop before the next one begins, delays stack up.
EIP-8411 breaks the payload into smaller pieces — the current draft proposes 64 chunks. Each piece carries a cryptographic proof, tied to a Merkle root (a kind of digital fingerprint) that the block builder commits to in advance. That lets any computer verify a single piece on its own and start forwarding it immediately, while the rest of the pieces are still arriving. It is like streaming a video: you watch the first minutes while the rest downloads, instead of waiting for the whole file.
Importantly, the prototype keeps Ethereum’s existing networking machinery — the gossipsub messaging format, network mesh, peer scoring — intact. Only how payload pieces are published and forwarded changes, which lowers the engineering risk of the upgrade.
The Core Conflict: Speed Costs Bandwidth
The impressive numbers come from a controlled simulation, and the researchers are candid about the trade-offs. They modeled 500 nodes with realistic geographic latency, 50 Mbps upload and 100 Mbps download capacity, a 1 MiB payload originating from a home-based block builder, and ten randomized network configurations. The tests ran real Prysm client and go-libp2p-pubsub code against a simulated network and virtual clock — real software, but not real mainnet conditions.
The catch: the basic “Tier 1” design, which combines segmentation with batch publishing, requires roughly one-third more received bytes than today’s whole-message approach, because sending many individually identified pieces plus announcement messages adds overhead. Faster delivery, in exchange for more network traffic. More advanced tiers under discussion tackle the duplicate-data problem, but add complexity.
There is also process risk: the proposal is still a Draft EIP in the Ethereum repository, not an agreed upgrade. Ethereum developers were scheduled to discuss its possible inclusion in the Hegotá upgrade at the September 17 ACDC call — the same day the research dropped.
Market Implications: Plumbing Upgrades Pay Off Later
Infrastructure improvements rarely move prices the day they are announced, but they compound. Ethereum competes for developers and users on speed, reliability, and cost. If sub-one-second data propagation eventually lands on mainnet, it strengthens the foundation for higher-throughput usage — payments, tokenized assets, and DeFi — without the network becoming more fragile. Ethereum currently trades around 2,463 USD, up roughly 3 percent on the day, though daily moves have little to do with research posts like this one.
For ETH holders, the right lens is long-term: projects that keep upgrading their core plumbing tend to hold up better when demand returns. The fact that this work is happening at the networking layer — invisible to most users — is a sign of a maturing platform fixing bottlenecks before they bite.
The Verdict
EIP-8411 is a promising but early-stage upgrade. The five-fold speedup is real in simulation, the design smartly reuses Ethereum’s existing networking stack, and the costs — extra bandwidth and implementation complexity — are understood. Whether it ships depends on the ongoing developer discussions around the Hegotá upgrade. Nothing changes for users today, and that is precisely the point: the best infrastructure upgrades are the ones you only notice by never having to think about them.
The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.
Disclaimer: This article is for informational purposes only and does not constitute financial advice.
5 seconds down to 0.75 for block propagation is massive tbh. fewer micro forks means less reorg anxiety for anyone running anything mempool sensitive
this is the boring infrastructure work that actually compounds. nobody tweets about segmented broadcasting but youd feel it the day it broke
five seconds down to under one for propagation. whoever was front running you by seeing blocks first just lost most of their edge
thats the real story imo. segmented broadcasting mostly hurts the hft style extractors, regular users just get cleaner inclusion
the edge doesnt disappear, it compresses. extractors will just pay for colocated relay access instead, same fight one layer up
colocated relay access kinda already exists, thats basically what mev boost relays are. agree the fight just moves up a layer
mostly agree, but a 0.75s median still leaves a window for the fastest operators. under one second is better, its not neutral
colocated validators still win any sub second race yeah. but shrinking the gap from 5s to 0.75s cuts their edge by a lot, direction matters here
1 MiB across half the network in under a second, worst case from 6s to just over 1. devs quietly shipping while everyone argues about prices
Simulations are one thing. What I want to see is EIP-8411 on a live global network with thousands of geographically spread nodes before cheering.
fair on wanting mainnet data, but those testnet numbers came from geographically spread nodes, not a single lab rack. one real congestion event will tell us everything
spread nodes sure, but testnet blocks are tiny. the real test is a full block during an NFT mint when every peer is drowning
exactly, a 1 MiB sim is a lab coat test. run it during a hyped mint with 200k pending txs and then show me the p99 numbers
p99 is the number that matters and they only published medians. still, worst case 6s down to 1s in sim is a big enough cushion to survive a mainnet haircut
Fair ask, but even if mainnet lands at 2x instead of 5x, cutting worst case propagation from 6s to under 2 kills a whole class of stale block nonsense
the 6s worst case is where the real damage hides. even 2x on mainnet means stale block orphans stop being a daily tax on validators and exchanges
block data spreading 5x faster and half the thread is about price. this is why the good shipping happens while nobody is looking
5s down to 0.75s median with worst case barely over 1s. the practical win is fewer reorgs for fast settlement, not raw latency
fewer reorgs yes, but lets be honest the win only counts if mainnet nodes at actual scale reproduce the 0.75s. sims with 1 MiB payloads are the easy case
fewer reorgs is also fewer accidental double spends at exchanges. this quietly saves venues money on confirmation waits
5x faster blobs means L2 data lands before slot 2 even starts. rollup finality gains alone justify the EIP imo
0.75s median on 1 MiB is a lab number and i still want it. the real test is a congested block during a mint, if it holds there the reorg tax mostly disappears
a congested mint is exactly the right test but even a 3x improvement under load kills most of the stale block tax. you dont need the full 5x for it to matter
Under one second median means solo stakers see blocks nearly as fast as the big farms. Small protocol change, decent decentralization argument.
speed parity with the big farms is the quiet decentralization win. a solo staker that sees blocks as fast as a datacenter has less reason to give up and delegate