Fedwire vs Stablecoin Costs: $1M Settlement Benchmark 2026

A $1 million payment rarely becomes expensive because of the payment network itself. The meaningful costs usually sit around the rail: FX spread, custody, compliance, liquidity, reconciliation, and the work required when a transaction fails or arrives late.

Central takeaway: For a domestic USD-to-USD payment, Fedwire usually wins on simplicity. Stablecoins become compelling mainly for cross-border, after-hours settlement when both parties already have compliant wallets, usable liquidity, and operational controls.

The $1 Million Payment-Rail Benchmark

Payment pricing has three layers:

• Network fee: What the payment system charges a participating institution.
• Participant or customer fee: What a bank, sponsor, or service provider passes on to its customer.
• All-in cost: Fees plus FX, custody, compliance, liquidity, reconciliation, and failure risk.

Those layers are easy to confuse. A quoted $0.045 Fedwire, FedNow, or RTP charge is a participant-level network fee. It is not necessarily what a corporate customer pays.

The official 2026 Fedwire Funds Service fee schedule lists a $0.045 transfer fee. The FedNow fee schedule lists the same per-credit-transfer charge. The Clearing House’s RTP pricing also lists a $0.045 network fee. These figures describe what participants pay the system, not a guaranteed retail or corporate price.

On a $1 million transfer:

• $1 equals 0.01 basis points
• $10 equals 0.10 basis points
• $100 equals 1 basis point
• $1,000 equals 10 basis points

That arithmetic puts network pricing in perspective. A 10-basis-point FX spread costs $1,000—more than 22,000 times a $0.045 network fee. A $25 bank wire fee costs 0.25 basis points. Even a $3 blockchain transaction fee is financially trivial unless it accompanies expensive conversion or operational work.

Fedwire remains the uncomplicated choice for domestic, high-value USD settlement. Its operating window is a limitation, but banks, treasurers, and auditors understand the process and the settlement finality.

CHIPS optimizes a different problem. Its value comes from multilateral netting, which can reduce the intraday liquidity a bank needs for recurring high-value flows. That advantage matters to a treasury operation processing thousands of payments, not necessarily to the company sending one wire.

FedNow and RTP offer 24/7 instant domestic credit transfers. FedNow’s $10 million figure is a system maximum; participating institutions can configure lower customer or transaction limits. The practical constraints are often bank policy, fraud controls, credit-push-only behavior, and sponsor pricing rather than the network fee. The Clearing House reported 422.2 million RTP payments totaling $1.679 trillion through September 2026, a useful scale indicator but a time-sensitive figure that should be read with its September 30, 2026 reporting date.

Stablecoin Settlement: The Real Cost Stack

Stablecoins add a 24/7, programmable settlement layer, but the blockchain transfer is only one step.

Consider a company sending $1 million of USDC to a foreign operating account. The full route may include:

• USD-to-USDC conversion
• Institutional custody or wallet-management charges
• Blockchain gas
• Wallet screening and sanctions monitoring
• USDC redemption or exchange
• Conversion into local currency
• Local payout and banking fees
• Prefunded liquidity
• Accounting and reconciliation
• Manual review for flagged transactions

The following ranges assume USDC, an institutional custody model, and a relatively liquid chain such as Ethereum Layer 2 or Solana. They assume institutional, not retail, FX pricing. Ethereum mainnet, retail exchange spreads, or a managed wallet provider with per-transaction pricing can produce materially higher costs.

These are operating scenarios, not universal price quotes. A company with direct institutional minting, its own wallet infrastructure, and deep local liquidity may land below the ranges. A company buying USDC through a retail exchange and manually reconciling wallet activity may land far above them.

For a one-off domestic USD-to-USD transfer, Fedwire generally wins. Stablecoins become more interesting when the alternative involves correspondent banks, weak local banking hours, multiple time zones, or repeated weekend funding. They can also reduce settlement coordination when the recipient already accepts USDC and can convert it efficiently.

Before choosing the route, ask:

1. How does the recipient redeem into bank money?
2. Who carries issuer, depeg, and chain-outage risk?
3. What happens when wallet screening blocks a payment?
4. How much liquidity must be prefunded?
5. Can the ledger reconcile wallet activity automatically?
6. What does a failed or misdirected transfer cost to recover?

A route that saves five basis points but requires a full-time operations analyst is not cheaper.

Choosing Among Fedwire, RTP, FedNow, and Stablecoins

These systems are not interchangeable.

Fedwire is a high-value institutional settlement rail. RTP and FedNow are immediate domestic credit-transfer systems, often used for supplier payments, insurance disbursements, payroll, and account funding. Stablecoins are most useful where jurisdiction, operating hours, or programmability matter more than domestic simplicity.

All three instant-payment systems are credit-push systems: the sender initiates the payment, and recovery may be difficult once funds are credited. That is manageable in a controlled treasury workflow. It is riskier in a consumer product exposed to account takeover or authorized push-payment fraud.

A routing engine might start with rules like these:

def choose_rail(amount, urgency, cross_border, recipient_ready, fx_bps):
    # Illustrative policy: 15 bps is an assumed break-even point,
    # not a universal market threshold.
    if cross_border and recipient_ready and urgency == "24/7" and fx_bps < 15:
        return "stablecoin"
    if amount >= 1_000_000 and urgency == "business_hours":
        return "fedwire"
    if not cross_border and urgency == "instant":
        return "rtp_or_fednow"
    return "bank_transfer"

The 15-basis-point threshold only makes sense after modeling the alternative. For example, it could represent a corridor where stablecoin infrastructure saves $1,500 per $1 million transfer after avoiding correspondent fees and delay. In another corridor, the break-even point could be 5 basis points—or no point at all.

Real routing also needs bank-specific limits, sanctions screening, recipient verification, liquidity checks, exception queues, and a policy for failed settlement. The goal is not to find one universally cheapest rail. It is to make the choice explicit.

AI SaaS Has the Same Cost Blind Spot

AI software companies make a similar mistake when they focus on token prices and ignore the delivery system around them.

A product may advertise strong software margins while paying for model calls, retrieval, GPU capacity, observability, human review, implementation, and direct support.

AI gross margin = 1 − fully loaded delivery costs ÷ recognized revenue

Here is an illustrative $1 million revenue model. The token assumptions use $2 per million input tokens and $8 per million output tokens—illustrative API rates, not a claim about any particular provider’s current pricing.

That produces a fully loaded gross margin of 62.4%. A 20% increase in output-token use adds $3,200 under these assumptions. A 20% increase in human review adds $30,000. In many deployments, exception handling matters more than model pricing.

Track usage per account, not just at the company level:

• Input and output tokens
• Tool calls and retrieval volume
• Inference or GPU spend
• Review minutes per completed task
• Retry and failed-task rates
• Support tickets tied to delivery
• Implementation hours

A practical warning threshold is when direct usage and review costs consistently exceed roughly 25%–35% of account revenue. That may be acceptable for a strategic customer, but it should be an intentional pricing decision.

Stress-test both payment and AI models against the ugly cases: a chain halt, a 24-hour redemption outage, a bank holiday, a provider outage, higher retry rates, or a surge in manual compliance review. Normal-case unit economics are only half the operating model.

Frequently Asked Questions

How much does it cost to send $1 million over Fedwire?

The 2026 Fedwire participant-level network fee is $0.045 per transfer. A corporate customer may pay a bank or sponsor fee; a practical domestic USD-to-USD total might be roughly $35–$200, excluding unusual liquidity or correspondent charges.

Are stablecoins cheaper than bank wires?

Sometimes, particularly for cross-border and after-hours settlement where both parties already have compliant wallets and local liquidity. For domestic USD-to-USD payments, custody, conversion, compliance, and reconciliation often make stablecoins more expensive than Fedwire.

Is FedNow cheaper than Fedwire?

Their network fees are similarly small, but they serve different jobs. FedNow provides 24/7 instant credit transfers, with a system maximum of $10 million that participants may set lower. Fedwire is designed for high-value institutional settlement during its operating window.

What is a healthy gross margin for AI SaaS?

For many inference-heavy products, 60%–70% fully loaded gross margin is a useful operating target. The calculation should include model usage, infrastructure, human review, implementation, and direct support—not just the software platform.