The Architecture Shift
Until 2023, Ethereum operated under a simple model. Validators assembled transactions from the mempool, ordered them, and produced blocks. The validator had full control over transaction inclusion and ordering. This was centralized in a single entity (the validator) but predictable.
Then came Proposer-Builder Separation (PBS). The model split into three roles.
The proposer (a validator chosen by the protocol) decides which block to include in the chain. The builder (a specialized searcher or pool operator) assembles bundles of transactions and produces block candidates. The relayer (Flashbots Relay or equivalent) acts as matchmaker between proposers and builders.
This separation was designed to solve MEV problems. By fragmenting control, the theory goes, no single entity can extract MEV efficiently. In practice, PBS created new architecture and moved the chokepoint.
Under the old model, a validator could sandwich a payment transaction to extract slippage. The validator had total control. Under PBS, the builder has total control of transaction ordering within blocks. But the builder cannot force the proposer to include a specific block. And the proposer cannot see the block contents until deciding to include it.
The result is that neither the builder nor proposer has unilateral control. But this also means neither has a commitment to including your transaction in a specific time window.
For payment infrastructure, this matters. Your settlement order may be delayed, reordered, or excluded based on builder incentives and relay policies.
Understanding PBS Architecture and Settlement Risk
The proposer maximizes profit by choosing the highest-bidding block. The builder maximizes MEV by reordering transactions to extract slippage, liquidations, and toxic flow. The relayer delivers the block but doesn't enforce ordering. None are obligated to prioritize your payment. If your USDC-to-DAI swap conflicts with a more profitable arbitrage, the builder orders the arbitrage first to capture price impact. This is rational builder behavior.
The PBS Model on Ethereum
Post-Dencun (March 2024), PBS is Ethereum's standard. Builders gather transactions from mempool and private pools, assemble blocks to maximize MEV, and bid for inclusion through Flashbots Relay. The proposer only sees bid amounts, not block contents. Time from submission to inclusion spans roughly 0-30 seconds in the mempool, 4-8 seconds for builder assembly, 2-4 seconds for auction, and 1-2 seconds for distribution. Total elapsed time is 7-44 seconds, but only if the builder includes you. If not, you wait 12+ seconds for the next builder.
You cannot guarantee sub-60-second settlement for regular transactions. Builders deprioritize regular payments in high-MEV environments. To guarantee faster settlement, use a private relay (Flashbots Protect, MEV-Blocker) or pay MEV-Burn fees (3-10 basis points).
Flashbots Relay and MEV-Burn
Flashbots Relay is the dominant PBS relay. It holds approximately 90% of MEV extraction via relay route. Flashbots committed to MEV-Burn (Ethereum Improvement Proposal 7002), which burns a portion of MEV, reducing the maximum extractable value. But Flashbots itself maintains a database of transaction flows and builder bids. Gatekeeping power is still concentrated.
MEV-Burn reduces the incentive to extract MEV but does not eliminate it. Under MEV-Burn, builders can still extract MEV but must burn the profit. This incentivizes builders to include transactions from high-fee sources (payments willing to pay a premium) over regular transactions. For payment processors, the practical outcome is that regular transactions (no premium) are deprioritized, while transactions willing to pay MEV-Burn fees (basis points above market) get inclusion in the next block. Cost of MEV-Burn-compliant routing on Ethereum is 3 to 10 basis points per transaction, depending on MEV opportunity available.
L2 Sequencers and Deterministic Ordering
Layer 2s avoid PBS by using centralized sequencers. Arbitrum orders transactions deterministically in published-in-advance slots, making it MEV-resistant and predictable, though you trust Offchain Labs as the operator. Optimism uses a centralized proposer with FIFO ordering, also deterministic. Polygon uses a single validator per block, less decentralized but more predictable than Ethereum PBS. Solana rotates slot leaders every 400 milliseconds and orders by priority fee, transparent and fast.
Settlement Timing by Chain
For time-sensitive payments, choose your chain carefully. Ethereum L1 with PBS guarantees sub-60 seconds for regular transactions, but sub-10 seconds only with private routing at 3-10 basis points cost. Arbitrum guarantees sub-15 seconds with deterministic sequencing. Optimism guarantees sub-20 seconds with FIFO ordering. Polygon guarantees sub-5 seconds with low cost. Solana guarantees sub-500 milliseconds with transparent fee ordering.
Builder Concentration and Centralization Risk
The builder market on Ethereum shows high concentration. Lido staking node operators dominate block building. The top three builder groups control 65% of blocks built. If these builders collude or fail simultaneously, PBS breaks.
This is an understated risk. In 2025, two major builders (Beaverbuild, MEV-Blocker) experienced outages that lasted 20 to 40 minutes. Payment processors using these builders as fallback routes saw settlement delays.
Mitigation. Use multiple builder routes. If your primary builder (Flashbots) is congested, immediately fall back to a secondary builder (Beaverbuild) or a tertiary route (direct Ethereum transaction, higher gas cost). This adds latency (10 to 30 seconds) but guarantees eventual inclusion.
When to Route Through MEV-Resistant Sequencers
For payment infrastructure, the decision to use Ethereum PBS vs. a sequencer-based L2 comes down to three factors. First, settlement timing requirements. Must completion happen in under 15 seconds? Use an L2 sequencer (Arbitrum, Optimism, Solana). Can you tolerate 30 to 60 seconds? Use Ethereum with private routing. Second, cost tolerance. Is 3 to 10 basis points per transaction acceptable? Use Ethereum private routing (Flashbots Protect, MEV-Blocker). Can you absorb only less than 1 basis point cost? Use Arbitrum or Polygon. Third, trust model. Do you prefer cryptographic consensus (Ethereum PBS with distributed builders)? Or do you accept a centralized operator (Arbitrum sequencer, Solana leader)? Both have trade-offs.
Most payment processors route high-value, time-sensitive payments through Arbitrum (deterministic sequencer, low cost, moderate trust assumption). They route regular payments through Ethereum with MEV-burn protection. They use Solana for mass-market micropayments where latency is not critical but cost must be minimal.
Final Thoughts on PBS and Settlement Design
PBS is permanent on Ethereum. Centralization of builders is not a bug; it is a feature of the profit motive. Builders can invest in hardware and infrastructure because they capture MEV. This attracts institutional capital and improves Ethereum's transaction throughput.
The cost is that individual transactions are no longer independently settled. They are bundled into blocks by builders pursuing MEV. This is not a failure; it is a design choice. For payment infrastructure, the answer is to accept it, price for it, and route around it when necessary.
The networks that minimize PBS (Arbitrum, Solana) will capture payment volume. They offer deterministic ordering. But they centralize sequencing risk. Both models are valid. Choose based on your settlement requirements and tolerance for trust assumptions.