On September 16, 2026, Circle brought its Arc blockchain to public mainnet after a testnet phase that processed more than 700 million transactions (The Defiant). Arc is a layer-1 chain built to settle payments, foreign exchange and tokenized-asset activity, with transaction fees paid in USDC rather than a native gas token. Instead of opening to permissionless validators, it launched with eleven named founding validators (BlackRock, DTCC, Galaxy, ICE, Mastercard, MoneyGram, SBI Group, Standard Chartered, Sumitomo Corporation, Visa and Worldpay) running proof-of-authority consensus (a model where a fixed, named set of parties produces blocks, rather than anyone who stakes or mines), alongside roughly 20 SOC 2-certified operating partners (SOC 2 is a common third-party audit standard for how a service provider handles data security) (The Defiant). Circle also completed a genesis mint of 10 billion ARC tokens (60% to ecosystem uses, 25% to Circle itself, 15% to reserves) while stating this is "a technical milestone rather than a commitment to publicly launch ARC"; the token is not yet tradable (The Block). More than 100 applications went live day one, including Aave, Morpho and Uniswap (Decrypt).

Two of the six weighted dimensions and one hard gate apply here, and the honest answer for all three, three days in, is: not yet measurable, which is itself the finding.

Security, Decentralization & Durability (18/100) is built from node count and client diversity, governance, audit quality, and a track record through at least one full market cycle (framework). Arc's node count is eleven institutions, chosen by Circle, running one consensus client, with Circle retaining control over protocol development, compliance policy and validator membership. That is a legible, auditable structure (worth more than an opaque one) but it is concentration, not diversity: eleven parties that already know each other, several with existing commercial ties (BlackRock manages a large share of USDC's reserves; Visa and Mastercard are payment-rail partners), are not eleven independent sources of failure. And the track-record component starts at zero on day three, regardless of how much institutional weight sits behind the launch. A framework that scores durability by time survived under stress has nothing yet to score.

Hard gate 8 (custody that can freeze, mint, or seize) asks the right question about the wrong subject if pointed at Arc itself, because Arc is infrastructure, not a covered asset. But the question travels to whatever settles on it. USDC already carries Circle's blacklist function. A chain whose validator membership and upgrade path also run through Circle's control puts a second form of unilateral authority in the same hands as the stablecoin paying for its gas, rather than distributing that authority the way a permissionless base layer would. Nothing here is a new power over any covered asset; it's the same operator's authority showing up in a second place, which is a governance fact worth naming even though it fails no covered asset's gate today.

Tokenomics & Value Accrual (28/100) doesn't apply yet, and that's worth raising anyway. The August 20 read on the SEC's disclosure proposal argued that periodic, filed data is worth more to this dimension than a founder's claim, because supply and allocation numbers only count once they're checkable against something. ARC's 60/25/15 split is exactly that kind of claim today, reported by two independent outlets, but a press disclosure, not a filing, and not attached to a trading market yet. If ARC ever lists, insider share, vesting and float-versus-FDV (circulating supply against fully diluted value, how much of the total token count is actually tradable versus locked up) become gate 4 and gate 5 questions the moment there's a chain to check them against.