Who else owns rails in the agent era?

Adyen — the enterprise cross-border alternative

Adyen is the closest direct competitor to Stripe in the integrated-stack frame. The product surface is similar: card processing, alternative payment methods, risk and fraud, terminal payments, marketplaces, and payouts. The differences are positional. Adyen built up from enterprise downward (Uber, Spotify, Microsoft, Meta as flagship customers) where Stripe built up from developers and SMB upward. The enterprise-first DNA shows in Adyen's single-platform architecture (one stack globally, one contract, one settlement currency choice) and in their ability to handle very-large-merchant complexity.

Where Adyen beats Stripe: cross-border presence outside the US, particularly in continental Europe and parts of Asia-Pacific; local-payment-method coverage (iDEAL in the Netherlands, SEPA across the EU, Sofort/Klarna integrations, regional wallets in Asia); enterprise contract terms and dedicated solution engineering; merchant-of-record options that simplify tax handling. Where Adyen loses: developer experience for small and mid-market accounts is meaningfully behind Stripe, the time-to-integrate for a small team is days where Stripe is hours, and the documentation surface is built for solution engineers rather than for autonomous developer agents.

For the agent era specifically, Adyen's profile is interesting but mixed. The enterprise cross-border position is valuable because a meaningful slice of agent commerce will be B2B procurement that crosses borders. The developer-experience gap is a problem because the agent on the seller side is going to be an autonomous developer agent (Cursor, Claude Code, the next generation) and those agents prefer documentation surfaces built like Stripe's. Adyen has work to do at the LLM-consumable-docs layer to keep pace.

Source: Adyen investor day disclosures 2024-2025; Stripe Annual Letters; analyst coverage from BTIG and Mizuho.

PayPal, Braintree, and the Venmo distribution lever

PayPal is the legacy incumbent. The brand is two decades old, the user base is enormous, and the consumer-side checkout experience is broadly recognized. The product portfolio includes the consumer PayPal wallet, Braintree (the developer-payments brand acquired in 2013), Venmo (the US peer-to-peer wallet acquired with Braintree), and a growing set of merchant services. PayPal's revenue per active account has been trending down for years and the company is in the middle of a strategic repositioning under newer leadership.

Braintree is the relevant brand for developer-facing payments. The platform competes with Stripe and Adyen on integrated stack but has lost mindshare since the mid-2010s. Braintree's product surface is competent; its problem is positioning. Most new builders default to Stripe without considering Braintree, and PayPal has not invested in changing that default narrative the way Stripe has invested in maintaining theirs.

Venmo is the more interesting strategic asset. With approximately 90M US users and a deep peer-to-peer pattern, Venmo is one of the few US wallets with consumer brand at scale. The question is whether Venmo's distribution can be turned into a checkout default for AI-driven commerce, particularly for personal-shopping agents that act on behalf of consumers. PayPal has been pushing in this direction with Venmo for Business and various agent-friendly hooks, but execution has been slow. The asset is real; the conversion to relevant volume is unproven.

Square, Cash App, and the SMB / individual-operator angle

Block (parent of Square and Cash App) plays a different position. Square is the SMB point-of-sale and software business: terminal hardware, inventory and CRM software, payroll, capital, banking-as-a-service. Cash App is the consumer peer-to-peer wallet that has grown into a quasi-bank with debit card, Bitcoin trading, and the Afterpay buy-now-pay-later acquisition. The two sides cross-pollinate: Cash App users discover Square sellers; Square sellers receive Cash App payments.

For the AI-built economy specifically, Square's position is strongest in the individual-operator slice that Stripe is also pursuing. A solo creator running an AI-assisted services business, a microbusiness selling AI-generated goods, a contractor invoicing for AI-augmented work; these accounts are exactly the cohort Square has been serving for a decade. Square's banking and capital products may be more familiar to that operator than Stripe Atlas and Stripe Capital, depending on geography and channel.

Where Block is weaker: international presence is thin, enterprise positioning is essentially non-existent, and the developer-platform layer is less polished than Stripe's. The agent-era thesis for Block depends on Cash App's consumer base remaining sticky as more buying gets delegated to agents and on Square's SMB stack staying relevant for the kind of operator the AI economy produces. Both look defensible but not expansive; Block is a meaningful player in the agent era, not the candidate to win it outright.

Crypto rails: USDC, stablecoins, and the L2 problem

Stablecoin volume has been quietly compounding. USDC and USDT together moved trillions of dollars in 2025, with a meaningful slice now flowing on layer-two networks (Base, Arbitrum, Optimism, Polygon) rather than only on Ethereum mainnet. The use cases that have stuck are cross-border payments, B2B settlement in markets with weak local rails, treasury operations for crypto-native businesses, and a growing slice of retail payments in countries with high inflation or strict capital controls.

For agentic commerce specifically, stablecoin rails have one structural advantage: programmability. A stablecoin transfer is a smart-contract call that can carry arbitrary structured data, enforce conditional release, and integrate with on-chain identity primitives. An agent that wants to transact with another agent under specific conditions (release on delivery, escrow with arbitration, multi-party settlement) has more native affordances on a stablecoin rail than on a card rail. The agent-to-agent payment primitive on stablecoins is meaningfully ahead of the equivalent primitive on card.

The L2 fragmentation problem is the major drag. There are now dozens of layer-twos and emerging layer-threes, each with its own liquidity profile, bridging risk, and developer ecosystem. An agent attempting to transact on the cheapest, fastest, most-liquid stablecoin venue has to navigate the fragmentation, which usually requires aggregation infrastructure (Stargate, LayerZero, Across) that adds its own trust assumptions. The fragmentation will likely consolidate over the next 24-36 months as a handful of L2s emerge as the practical defaults; until it does, stablecoin commerce remains operationally complex.

Stripe's stablecoin posture (USDC support since 2024, expanded global payouts on stablecoin rails) is the practical hedge. The optionality is cheap. If stablecoin volume continues to compound, Stripe is positioned to be the bridge between card-rail and stablecoin-rail commerce. If stablecoins stay a niche, Stripe loses nothing material on the investment.

Source: TRM Labs annual stablecoin report 2025; Chainalysis Crypto Adoption Index; Circle disclosures and USDC quarterly reports.

Central bank digital currencies

Central bank digital currencies are the slowest-moving but largest-tail-risk piece of the payments map. China's DCEP (digital yuan) is the most advanced in production deployment, with hundreds of millions of wallets and meaningful retail volume inside China. The EU's digital euro is in advanced design phase with a target deployment window in the late 2020s. The US has limited retail-CBDC appetite politically but has been working on a wholesale-CBDC infrastructure for interbank settlement (Project Cedar, FedNow as a related but non-CBDC parallel).

The relevant question for Stripe and the broader payments stack is whether a meaningful retail CBDC ships in a major market and whether merchants are required (or strongly incentivized) to accept it. In China, that is already the answer; the digital yuan is operationally a parallel rail that merchants accept alongside Alipay and WeChat Pay. In the EU and US, the political and architectural questions are unresolved and shipping is years out.

If a major retail CBDC does ship at scale, the take-rate economics for card rails get pressured. CBDCs are typically designed to settle near-instantly at near-zero cost, with no intermediation by card networks. Sellers facing customer-led pressure to accept CBDCs would have a strong incentive to do so, and the cost differential would compress the value proposition of card-network-intermediated processing. Stripe and its peers would respond by becoming the integration layer for CBDCs (the same way they became the integration layer for cards) but the per-transaction economics would shift unfavorably.

The probability-weighted scenario for the next five years is: continued slow rollouts, increased adoption in markets with weak alternatives (parts of Asia, Africa, Latin America), limited disruption to card-rail economics in the US and Western Europe. The longer tail is meaningful, though, and any serious payments-stack thesis has to acknowledge the CBDC risk as a real input.

B2B AP/AR rails: a different game entirely

The B2B accounts-payable and accounts-receivable side of payments runs on entirely different unit economics from card-led consumer commerce. Annual B2B payments volume globally is multiples of consumer card volume, but very little of it flows over card rails. The dominant rails are ACH and wire in the US, SEPA in the EU, and various local equivalents elsewhere, with the AP/AR workflow itself (invoice receipt, approval, payment, reconciliation) being the value-added layer above the rails.

The publicly-listed AP/AR pure-plays include BILL (formerly Bill.com) and AvidXchange in the US. The private leaders include Tipalti (cross-border, large-merchant), Ramp (corporate cards plus AP automation, growing fast), and Brex (originally cards, now broader spend management). The product surface looks like accounts-payable software with payment rails built in; the revenue model is some combination of SaaS subscription, per-transaction fees, and float on customer funds.

For agent-driven commerce specifically, the B2B AP/AR side is one of the highest-leverage agent applications. An agent that handles a company's vendor management, invoice approval, payment scheduling, and cash-flow optimization is a coherent product that delivers measurable value. Ramp and Brex have both been pushing agent-driven workflows aggressively; the early signal is that this is one of the cleanest fits between agent capability and enterprise willingness to pay. The take-rate economics here are typically higher than consumer card processing because the value-add is the workflow, not the rail.

Stripe's exposure to this side is partial. Stripe Billing handles recurring invoicing; Stripe Connect handles marketplace payouts; the broader AP/AR workflow is not Stripe's primary focus. The plausible reading is that the B2B AP/AR layer remains a separate competitive field with its own winners, and Stripe is unlikely to be the dominant player there. For an investor or operator looking at agentic-commerce exposure, BILL, Ramp, Brex, and their peers are the relevant comparison set rather than Stripe.

Local-market processors and the geographic surface

Outside the US and Western Europe, the payments map is dominated by local-market processors rather than global platforms. Razorpay in India serves a large slice of the country's online-merchant volume with deep integration into UPI, India's domestic real-time payments rail. Mercado Pago in Latin America (originating from MercadoLibre) is the default for much of Brazil, Argentina, Mexico, and adjacent markets, with strong local-payment-method support and a wallet brand at scale. Paystack and Flutterwave compete for the Africa surface, both with strong local rails and a growing cross-border presence.

China is its own ecosystem. Alipay (Ant Group) and WeChat Pay (Tencent) together handle the vast majority of consumer digital payment volume, with the digital yuan layering on top. Cross-border into and out of China is regulated and operationally complex; the local rails are deep but largely closed to Western processors.

For the agent era, the question is which of these local processors invest in agent-readiness fastest. UPI in India is already structured in a way that an agent can interact with relatively cleanly; the rails are programmatic, the identity layer (Aadhaar-linked) is robust, and the volume is enormous. Mercado Pago has been quieter on agent-specific tooling but the underlying tech is competitive. The African rails are earlier in the curve.

Stripe's posture in these markets varies: present and competitive in India, present but smaller than local incumbents in Latin America, limited in China, growing in parts of Africa. The geographic surface argues against a global single-winner narrative. The likely outcome is that Stripe wins or co-wins in US-and-Western-Europe, Adyen wins enterprise cross-border, and local processors retain dominance in their home markets. The agent layer has to operate across all of them.

Comparison matrix

The matrix below summarizes positioning across the five dimensions that matter for the agent era. Integrated-stack depth is how much of the surrounding tooling (billing, tax, compliance, identity, capital) the vendor owns. Developer experience is the LLM-consumable-docs and clean-API quality. Geographic coverage is the practical breadth of the rail. Agent-readiness covers: agent-aware fraud surface, agent-to-agent commerce primitives, programmatic-issuing, stablecoin support, and LLM-friendly documentation. Pricing pressure is the directional risk to current take rates.

Stripe: deepest integrated stack; best developer experience; strong US and Western Europe; leading on agent-readiness (programmatic issuing, agent-scoped keys, stablecoin support, LLM-optimized docs); moderate pricing pressure as competition intensifies. Adyen: strong integrated stack at enterprise; weaker developer experience; strong cross-border Europe and Asia-Pacific enterprise; behind on agent-readiness at the developer surface; lower pricing pressure due to enterprise-contract stickiness. PayPal/Braintree: legacy integrated stack; developer experience mid-tier; large US consumer footprint; agent-readiness limited; high pricing pressure as differentiation has eroded. Block (Square + Cash App): integrated SMB stack; developer experience mid-tier; strong US, limited international; agent-readiness moderate; pricing pressure mid. Adyen, PayPal, and Block each have credible positions in slices of the market; none of them are positioned to displace Stripe at the developer-first integrated-stack center.

Crypto rails (USDC, stablecoin-native processors): thin integrated stack but strong programmability; developer experience varies; global by definition; agent-readiness structurally high due to programmability; pricing pressure on legacy rails is the main risk vector. CBDCs: integrated stack non-existent (rails-only); developer experience unknown; geographic coverage TBD; agent-readiness TBD; major long-tail risk to card-rail take rates. B2B AP/AR (BILL, Ramp, Brex, Tipalti): workflow-deep stack; developer experience mid-tier; strong US, expanding; agent-readiness strong because the workflow is the agent target; pricing pressure mid because the value is in workflow. Local processors (Razorpay, Mercado Pago, Paystack, Flutterwave): deep local stacks; developer experience improving; dominant in home markets; agent-readiness varies; pricing pressure local-context-dependent.

Strategic read

The picks-and-shovels framing for Stripe holds because Stripe sits at the developer-first integrated-stack center where most net-new AI-built businesses originate. The framing is incomplete because it under-states the value of the other positions. Adyen owns enterprise cross-border. Block owns a slice of the US SMB and consumer wallet space. BILL and Ramp and Brex own the B2B AP/AR workflow. Local processors own their geographies. Stablecoin rails are quietly compounding under everything.

For an operator building an AI-native business, the default integration is Stripe and stays Stripe for almost any consumer-facing or developer-tooling use case. For an operator building B2B agent workflows, the integration set is broader and likely includes Ramp or a peer. For an operator going global day one, Adyen or a local processor becomes part of the mix earlier than would have been true in the prior era.

For an investor, the cleanest exposures are Stripe (private, secondary if available), Adyen (public, established cross-border), Block (public, contested narrative but strong SMB and Cash App assets), BILL (public, AP/AR pure-play), and the B2B spend-management space broadly. Crypto-rail exposure is harder to play directly; the on-ramp companies (Circle, in particular) are the closest pure-plays. The bottleneck shifts from owning one rail to owning the right segment of the rail map, and the right segment depends on which slice of the agent era ramps fastest.