AI-native banking infrastructure & programmable settlement for autonomous financial agents

Differential insight (one line)

The realizable profit pool in agentic settlement is migrating from issuer float (Tether/Circle) and from bank-charter ownership (Augustus) toward whoever supplies institutional liquidity and yield to coalition stablecoins and BIN-sponsor banks, a role a licensed liquidity/broker/transfer-agent operator can occupy without a charter or an authorization-layer land grab.

Consensus → Δ

Consensus: Augustus's OCC approval signals an imminent AI-native bank that will issue its own stablecoin and capture reserve-float economics.

Δ: CEO Dabitz (Jul 2026) explicitly disavowed issuance; Augustus repositions as neutral clearing infra connecting fiat rails to third-party stablecoins, shifting its model to fee/spread orchestration, not seigniorage. [conf: med, category D statement, contradicts earlier filing language]

Consensus: The "regulated trust and control layer" for agent payments is greenfield territory for new entrants.

Δ: Visa and Mastercard built proprietary agent-identity tokens within a day of each other (Apr/Oct 2025), then wrapped them inside Google's AP2 mandate standard as launch partners, enclosing the authorization layer before Augustus-style entrants could compete for it. [conf: med-high, network press + technical spec]

Consensus: GENIUS Act (signed Jul 2025) settles the stablecoin regulatory framework, de-risking reserve-yield and yield-sharing models.

Δ: Effective date triggers on the earlier of Jan 2027 or final rulemaking; a year later, rules are still being drafted, and the anti-yield-sharing provision is a live political fight (OUSD coalition, White House CEA review), not settled law. [conf: med]

Why-now

A narrow 2026-2027 window exists before three things lock in: OCC's expanded permission envelope for digital-asset-holding banks becomes precedent, card networks fully consolidate the agent-authorization chokepoint via AP2, and GENIUS Act rulemaking forecloses affiliate-routed yield-sharing. After that, the cheap entry points (charter novelty, undefined liability rules, unresolved yield-routing) close.

Binding constraint

Regulation. OCC approval is conditional and non-operative; Fed stock approval and FDIC insurance are independent, undisclosed-timeline veto points (historically ~20% of conditionally approved de novo charters fail to open). Separately, GENIUS Act's affiliate-yield rebuttable presumption is unresolved, which directly gates any yield-distribution business model. Technology (AP2, x402) and distribution (card networks, coalition wallets) are not the bottleneck; regulatory finality is.

Wedge

Do not compete for a bank charter or the authorization layer, both are slow (multi-agency, 2026-2027) or already enclosed (Visa/Mastercard/AP2). Instead, position as the institutional liquidity and yield backend serving coalition stablecoins (OUSD-style) and BIN-sponsor