Ethereum Real-Time Blockspace

Ethereum’s 12-second block time is not just a latency problem — it is a financial primitive failure. Making block execution real-time (100ms mini-blocks) eliminates ~94% of MEV and creates an entirely new commodity market for block space. (→ [[ethdenver]])

The Two Problems

  1. MEV extraction: ~$300M/year extracted via front-running and sandwiching. Validators earn ~30bps extra yield from MEV on top of their ~3% consensus rewards. This extraction is structural, not incidental — it arises from the ability to reorder transactions within a 12-second window.

  2. Latency: Ethereum is slower than almost every competing chain. Speed matters both for UX and for the financial primitives Ethereum is trying to host (DEX trading, lending liquidations, cross-chain bridges).

The Block Space Commodity Framing

A single Ethereum block is worth ~$100. For that $100, you theoretically control $650B of assets for 12 seconds. Block space is underpriced relative to what it secures.

Parallels to traditional commodity markets:

  • Price discovery: what is a block actually worth?
  • Risk management: producers and consumers of blockspace need to hedge
  • Cartel visibility: MEV searchers are the “cartels” of blockspace — commodity markets expose and price this

Forward Block Markets

Ethereum’s FOCIL/PBS/ePBS work allows buying blocks ~64 slots (12.8 minutes) ahead. This unlocks:

ActionDescription
SpeculateBuy and sell block rights without building
Build the blockPhysically deliver the block (the actual pipeline)
Strip into piecesSell top-of-block separately from rest
Combine blocksSynthetically create longer 24s windows
Build emptyWalk away from the $100 — your prerogative

Moving from a $100 block to a block worth thousands (or millions) via real-time markets would massively expand Ethereum’s revenue model.

Real-Time Mini-Blocks (ETHGas Proposal)

The core proposal: split each 12-second block into 120 × 100-millisecond mini-blocks.

  • Transactions land at the end of each 100ms slot with a confirmed state update
  • Ordering is first-come-first-served within the slot; no reordering possible
  • Network latency (~50ms) means attackers cannot query the state and re-submit a competing transaction within the window → ~94% of MEV disappears
  • Trades are settled in real-time; no waiting for 12 seconds to know your execution price

Residual MEV

A ~100ms window still exists. The sheer network latency means that the theoretical attack surface is present but practically inaccessible — you cannot observe, compute, and submit within a single 100ms slot given global network propagation.

Impact on DeFi Primitives

AMMs (Uniswap model)

Study: ETH/USDC pool with 5bps fee at 12-second blocks vs. real-time:

  • Real-time allows fees to drop from 5bps → 1bps while LP revenue increases 2.7×
  • Effect: LPs earn more money because trading frequency multiplies (3–5× per 12-second interval)
  • For volatile alt coins: LP fees potentially 15–20× current levels
  • Net benefit to Uniswap alone: $3–3.5B/year

Lending Protocols

  • Can liquidate at 100ms granularity instead of 12-second intervals
  • Lower liquidation threshold needed → higher leverage available to borrowers
  • Increased capital efficiency throughout the lending market

Wallet UX

Standard transaction options become: Slow / Medium / Fast / Instant — where Instant means “confirmed by the time you lift your finger.”

Cross-Chain Liquidity

Real-time finality removes latency-based incentives to stay on faster L1s. L2 liquidity migrates toward the L1 where certainty is fastest.

Throughput Context

From $100 to $5B in net ecosystem value (Kevin Lepsoe’s estimate) by combining:

  • Real-time execution (100ms)
  • Block size increase (30M → 300M gas in 2026, per ZK proving progress — see ZK Proving Infrastructure)
  • Blockchain trilemma: ZK proving addresses throughput; real-time mini-blocks address latency

As of ETHDenver 2026: ~1% of the network running real-time. Target: 5–10% within months.

Connections

Open Questions

  • Can a real-time builder actually maintain the MEV-free property at scale, or does the 100ms window become exploitable with purpose-built low-latency infrastructure?
  • How does real-time block building interact with ePBS — does forward block purchasing still make sense?
  • What is the equilibrium price for a “real-time” block vs. a regular block in a forward market?
  • Does this require a hard fork or can it be implemented as a builder-level optimization?