Arc vs Ethereum L2s for stablecoins: an L1's assurances against a fragmented rollup landscape

This isn't Arc against one L2 — it's Arc, a single Layer 1 with one validator set and one settlement guarantee, against a landscape of separate Ethereum L2s (Base, Arbitrum, Optimism, Polygon, and others) that each have their own sequencer, their own exit window, and their own pool of stablecoin liquidity that doesn't automatically talk to the others. As of 2 September 2026, every L2 named here is live and carries real, verifiable USDC; Arc does not go live until 16 September 2026. The structural question this guide answers isn't "which is faster" — it's "what are you trusting, and how many times do you have to trust it."

Why "L2 landscape" is the right unit of comparison, not one L2

Arc is a single Layer 1: one validator set, one consensus mechanism, one settlement guarantee that doesn't depend on another chain. The Ethereum L2 world is structurally different — Base, Arbitrum, Optimism, and Polygon are separate deployments, each with its own sequencer operator, its own upgrade and exit mechanics, and its own liquidity pools, all sitting on top of the same Ethereum settlement layer but not sharing state, sequencers, or a unified stablecoin pool with each other. Comparing Arc to "the L2s" means comparing one thing to several different things that happen to share a settlement layer, and the differences between those L2s matter as much as the difference between any one of them and Arc.

Settlement assurance and the sequencer/withdrawal-window question

This is the sharpest structural distinction, and it varies meaningfully L2 by L2, per L2Beat's independent risk assessments:

Base (OP Stack, Coinbase-operated): centralized sequencer, rated Stage 1. If the sequencer censors or stalls, a user can force a transaction directly through Ethereum, but L2Beat notes up to a 12-hour delay before that path is available. Fault-proof challenge resolution takes up to five days, collapsing to one day if contested on both proof paths. TVL around $13.04 billion (L2Beat: Base).

Arbitrum One (Optimistic Rollup): also centralized sequencer, also Stage 1 — L2Beat notes it meets four of six Stage 1 requirements. Users can force a transaction after roughly a 1-day delay if the sequencer misbehaves. Regular contract upgrades carry an 8-day delay on L2 plus a 3-day delay on L1 before taking effect — a real, if imperfect, protection window. TVL around $11.57 billion, with 34% of that carrying additional trust assumptions (L2Beat: Arbitrum).

Optimism (OP Mainnet): centralized operator — described by L2Beat as "the only entity that can propose blocks." Critically, L2Beat notes there is no exit window for users if an unwanted upgrade happens: the Security Council holds instant upgrade power, meaning a malicious or mistaken upgrade could take effect before an affected user has any practical opportunity to exit. TVL around $1.56 billion (L2Beat: Optimism).

Polygon PoS: structurally different from the others — not a rollup at all, but an EVM-compatible proof-of-stake sidechain, with 105 active validators in a closed, capped set (minimum 100,000 POL staked to participate, no new entrants once the cap is hit). Block production is delegated in practice to a smaller set of elected block producers. Polygon has stated plans to transition toward a Validium architecture. TVL around $3.71 billion (L2Beat: Polygon PoS).

The pattern across all four: every one of them concentrates meaningful operational trust in a centralized sequencer or a small operator set today, with upgrade and exit protections that range from real-but-imperfect (Arbitrum's multi-day delay) to effectively none (Optimism's instant-upgrade Security Council). None of them is "trustless" in the way their marketing sometimes implies, and a builder evaluating any one of them needs the specific numbers above, not the general rollup-security pitch.

Arc, as a single L1, doesn't carry this particular risk category at all — there's no separate sequencer sitting between a transaction and the validator set, because the validator set is the whole settlement layer. What Arc substitutes instead is trust in its own permissioned validator cohort (roughly 100 institutions, eleven named publicly) and in Malachite, its Tendermint-based BFT consensus, targeting roughly 780ms finality. That's a different trust concentration — institutional and validator-based rather than sequencer-and-upgrade-committee-based — not an absence of trust concentration. Whether that trade is better depends on whether a builder's counterparties are more comfortable trusting a named institutional cohort or an Ethereum-anchored fault-proof system with a multi-day challenge window.

Fee volatility: native gas token vs. USDC gas, chain by chain

Every L2 discussed here uses a volatile native asset for gas by default: ETH on Base, Arbitrum, and Optimism; POL on Polygon. A stablecoin-denominated payment moving through any of them still pays its network fee in a token whose dollar price floats independently of the payment amount. Circle sells a separate product, Paymaster, specifically to let applications on these chains quote and collect fees in USDC instead — but that's an abstraction layer added on top of each chain's native ETH- or POL-gas design, not a base-layer property any of these L2s ship with by default (Circle multi-chain USDC).

Arc's gas is USDC-denominated at the protocol level, with no separate paymaster integration required to get that behavior — it's the chain's default. For a builder whose entire product is USDC-denominated, that removes an integration dependency and a point of fee-volatility exposure that every L2 named here carries by default.

Liquidity fragmentation: the cost the L2 model doesn't advertise

Because each L2 is its own deployment, native USDC issued on Base doesn't automatically exist on Arbitrum, Optimism, or Polygon — each chain gets its own native USDC deployment, and moving value between them requires a bridge (native or third-party) rather than a single unified pool. A builder deploying across multiple L2s to reach users on each one is managing separate liquidity pools, separate bridge risk, and separate integration work per chain — a real, recurring cost that the "we're all secured by Ethereum" pitch doesn't capture.

Arc, as a single chain, doesn't have this fragmentation problem internally — there's one Arc, one pool of USDC gas and liquidity on it. The trade-off is that Arc also doesn't inherit any of the L2s' existing liquidity or user base; a builder moving to Arc is moving to a fresh pool, not tapping into Base's $13 billion in existing TVL or any other L2's established liquidity.

What's actually live vs. what's a date on a calendar

Base, Arbitrum, Optimism, and Polygon are all live today, with the TVL, sequencer behavior, and USDC deployments cited above independently verifiable through L2Beat and each chain's own explorer. Arc's public testnet has run since 28 October 2025, with mainnet scheduled 16 September 2026 — meaning every claim about Arc's live finality, fee behavior, and validator conduct here remains untested in public until that date.

Which fits which job

A builder needing to reach users and liquidity that already exist today, comfortable with a centralized-sequencer trust model and willing to manage per-chain fragmentation and USDC-fee abstraction (via Paymaster) across one or more L2s, has real, live, well-documented options in Base, Arbitrum, Optimism, or Polygon — each with a different specific risk profile per the numbers above. A builder who wants a single settlement environment with native USDC gas by default and an institutional, named validator cohort rather than a sequencer-and-security-council model, has a case for Arc's 16 September mainnet, unproven as it currently stands. Built on Arc doesn't rank these; the honest read is that the L2 model and the single-L1 model concentrate trust in different places, and which one a builder's counterparties will accept is the real deciding factor, not raw speed or fees, where all of these chains are already fast and cheap.

Built on Arc is an independent directory. Arc is a Circle product; we are not affiliated with, endorsed by, or operated by Circle.

Sources: - L2Beat: Base risk analysis - L2Beat: Arbitrum risk analysis - L2Beat: Optimism (OP Mainnet) risk analysis - L2Beat: Polygon PoS risk analysis - Circle: multi-chain USDC - Circle Announces Founding Validator Cohort and Major Integrations for Arc