Surcharges vs. Amex: How AI Can Cut Payment Costs

Holiday checkout lines are the real-time stress test for payments. In December, volumes spike, average ticket sizes climb, and tolerance for “extra fees” drops fast. That’s why American Express CEO Steve Squeri’s blunt take on surcharging landed the way it did: “Surcharging in general is a bad customer experience.”

He’s not wrong. But here’s the part the industry tends to dodge: surcharges are a symptom, not the disease. They show up when merchants feel boxed in by payment acceptance costs, especially on premium cards where interchange is higher.

This post sits in our AI in Payments & Fintech Infrastructure series for a reason. If you run payments, risk, or fintech infrastructure, you should treat the surcharge debate as a cue to fix the underlying mechanics: cost transparency, smart routing, and automated optimization. That’s exactly where AI earns its keep.

Why surcharges are spreading (and why Amex is in the spotlight)

Surcharges persist because card acceptance costs are real, uneven, and hard to explain at the counter. When merchants can’t (or won’t) adjust pricing broadly, they reach for the most visible tool: add a fee at checkout.

The tension gets sharper with premium cards. American Express swipe fees average roughly 1.43% to 3.3%, while Visa and Mastercard average around 1.15% to 2.6% (figures widely cited in industry consumer finance coverage). When margins are thin—think restaurants, specialty retail, and services—even a 50–100 bps difference can matter.

Squeri’s comments came amid fallout from a proposed settlement involving Visa and Mastercard that would end the long-standing “honor all cards” rule. If that restriction loosens, analysts expect merchants won’t decline premium cards (wealthy shoppers are still shoppers). Instead, they’ll price-discriminate—often via a scaled surcharge.

Here’s the uncomfortable truth: merchants and networks are both rational. Merchants want cost control. Networks want premium rewards economics. Consumers want rewards and predictable pricing. The friction shows up at the register.

“Bad experience” isn’t a strategy

Calling surcharges “bad” is accurate but incomplete. If your business model depends on premium economics, you also need an answer to:

• How do we reduce the cost pressure that triggers surcharging?
• How do we make pricing more transparent before checkout?
• How do we keep authorization performance high while controlling cost?

This is where AI-driven payment systems stop being a buzzword and start being infrastructure.

The real driver: cost opacity at the decision point

Surcharging thrives in environments where the true cost of acceptance is hard to predict in real time. Merchants often don’t know the effective cost per transaction until later—after interchange qualification, downgrades, network fees, processor markups, chargebacks, and reconciliation.

So they do what’s simple: apply a blanket policy (“3% for Amex”) and move on.

The problem is that blanket policies are frequently wrong in both directions:

• They overcharge on transactions that would have been cheap to accept
• They undercharge on transactions that qualify poorly (downgrades) or carry higher risk

AI can’t change interchange tables by magic. But it can reduce the waste around them—waste that comes from bad routing, bad data, preventable downgrades, manual exception handling, and low-confidence fraud controls.

A strong stance: most surcharge programs are a sign the payments stack isn’t being managed as a system. It’s being managed as a monthly bill.

How AI reduces the need for surcharges (the practical version)

AI reduces surcharging pressure by lowering effective cost and stabilizing acceptance outcomes. Not by one big trick, but by dozens of small optimizations that compound.

AI-driven routing and orchestration: pick the cheapest “good” path

Answer first: AI-based transaction routing reduces cost by choosing the best path that still meets authorization and customer experience requirements.

In a modern stack, you may have multiple options per payment:

• Multiple acquirers
• Multiple gateways
• Multiple fraud and step-up options
• Alternative rails (where applicable)

AI models can recommend or automate routing decisions based on:

• Historical approval rates by BIN, MCC, amount, geography, time of day
• Cost curves (blended rates, network fees, processor fees)
• Real-time latency and outage signals
• Risk signals (device, velocity, identity confidence)

The result you actually care about: lower cost per approved transaction—not just lower cost per attempt.

Interchange optimization: prevent avoidable downgrades

Answer first: AI helps merchants qualify for better interchange by detecting data quality issues before submission.

A big chunk of “acceptance cost” comes from preventable issues:

• Missing tax/shipping fields where required
• Late settlement
• Incorrect POS entry modes
• Inconsistent customer data
• Poorly handled recurring credentials

AI (plus good rules) can flag likely downgrade triggers before settlement, not after. That can shave meaningful basis points over large volume—enough to reduce the perceived need for surcharging.

Fraud optimization: stop paying for false positives

Answer first: AI lowers the hidden cost of payments by reducing false declines and unnecessary step-ups.

Merchants don’t just pay interchange. They pay for:

• Fraud losses
• Chargeback ops
• Manual review
• Lost revenue from false declines

When fraud systems are blunt, merchants compensate by raising prices, adding fees, or steering customers to cheaper methods. Better AI risk scoring (and smarter step-up logic) reduces both fraud and friction.

A useful one-liner for operators: Every false decline is a surcharge you didn’t label.

Cost transparency at checkout: quote the total cost earlier

Answer first: AI can make checkout pricing predictable by estimating effective acceptance cost per transaction and feeding it into UX rules.

This is where “bad experience” can be prevented.

Instead of surprising customers with a last-second surcharge, merchants can:

• Present payment options with clear, compliant messaging
• Offer incentives for lower-cost methods (where allowed)
• Use customer segmentation (without being creepy) to tailor options

The best checkout experiences don’t argue with the customer. They guide them.

What the “honor all cards” shift changes for merchants and fintechs

Answer first: Ending “honor all cards” increases merchants’ ability to price-discriminate, which raises the operational value of AI-powered orchestration.

If merchants can treat premium cards differently, they now have a choice:

1. Add/raise surcharges
2. Decline certain card products (rare in practice)
3. Improve cost structure so surcharges become unnecessary

Option 3 is the only one that doesn’t tax your brand.

For fintech infrastructure teams, this policy shift increases demand for:

• Multi-acquirer orchestration
• Real-time cost and margin analytics
• Automated compliance controls (by state, by brand rules, by channel)
• Experimentation frameworks (A/B tests on routing, step-up, and checkout flows)

If you’re building payment infrastructure in 2026, you’re not just “processing payments.” You’re running a real-time decision system.

A practical playbook: reduce surcharge pressure in 90 days

Answer first: You can often reduce the drivers of surcharging within one quarter by focusing on measurable levers: approvals, downgrades, fraud friction, and routing.

Here’s what works in the real world.

1) Build a cost-per-approved-transaction baseline

Don’t start with interchange averages. Start with your actual unit economics:

• Cost per approved transaction (CPA-T)
• Approval rate by brand/product/BIN
• Downgrade rate and top downgrade reasons
• Chargeback rate and manual review rate

If you can’t measure CPA-T, you can’t evaluate whether surcharging is even “working.”

2) Add a routing layer (even if it’s simple at first)

Start with deterministic rules, then evolve to models:

• Route by geography and issuer performance
• Route by cart value thresholds
• Route by fallback logic during latency spikes

AI becomes powerful when it’s improving a routing system you already have.

3) Use AI to target the waste, not the customer

A lot of surcharge conversations become moral arguments about who “should pay.” Skip that. Target waste:

• Reduce false declines (lift approvals without lifting fraud)
• Prevent downgrades (fix data and timing)
• Cut operational cost (automate disputes and reconciliation triage)

4) If you surcharge, redesign the experience

Sometimes you’ll still surcharge—especially in low-margin categories.

If you do, treat it like UX debt you’re paying down:

• Disclose early in the flow, not at the last click
• Keep it consistent across channels
• Train support and store staff with a single script
• Monitor abandonment and NPS shifts weekly, not quarterly

People also ask: can AI really eliminate surcharges?

Answer first: AI can reduce the business need for surcharges, but it won’t erase them everywhere because interchange and network rules still exist.

Where AI makes surcharges less likely:

• High-volume merchants where small basis-point improvements compound
• Orchestrated stacks with multiple acquirers and strong data
• Businesses with meaningful fraud/false-decline problems

Where surcharges may persist:

• Extremely low-margin services with limited pricing power
• Merchants with constrained provider contracts (single acquirer, weak reporting)
• Environments with heavy regulation or strict brand rule constraints

The real win is choice: when your infrastructure is optimized, surcharging becomes optional rather than necessary.

The future looks like intelligent acceptance, not checkout penalties

Squeri’s argument is basically a customer experience argument: surcharges feel punitive. I agree. But I’d go further: surcharges are also a data problem and an infrastructure problem.

In the AI in Payments & Fintech Infrastructure series, we keep coming back to the same theme: payments are becoming software-defined. The winners won’t be the companies that argue hardest about fees. They’ll be the ones that build systems that make fees less painful—through smarter routing, better fraud decisions, and cleaner interchange qualification.

If you’re responsible for payments performance in 2026, here’s the question worth sitting with: Are you managing card costs with policies—or with real-time intelligence?