Skip to main content Skip to footer

Blog

Money in motion: How banks win B2B payments when AI agents decide

When agents pick the payment rail on every transaction, banks will have to earn each choice.

10-minute read

November 16, 2026

As corporate flows move to digital currency rails, banks could find up to $80 billion of their payments revenue at risk by 2030. As agents increasingly influence the choice of payment method, optimizing for cost and speed, $126 billion could be exposed.1,2

That’s the threat. But the same shifts open great opportunities. Faster settlement through digital currencies and programmable payments, with AI agents deciding when and where cash moves, could unlock an estimated $1 trillion of trapped liquidity for corporate clients by the end of 2030, worth roughly $84 billion in cumulative value over 2027 to 2030 (see Figure 2). Corporate clients can reinvest that cash to grow, and banks that help them capture it deepen those relationships. AI can also lower banks’ own costs: 98% of banks expect to increase their investment in AI and automation capabilities for payments3.

A significant majority (88%) of the 205 bank executives who participated in our Money in Motion survey see agentic payments as creating more opportunity than risk.

The challenge is getting there. Our study, which also surveyed 230 corporate treasury leaders, exposed clear divides over what each party needs and how fast. Banks that close those gaps decisively will turn disruption into growth, by moving faster on client priorities, influencing the payment decisions agents make, building controls that keep pace with machine-speed payments and helping clients build the business case for adoption.

Risks and responses

Consider the payment revenues risks in more detail. Cross-border corporate fees top the list for bank executives, followed by correspondent banking and FX, reflecting cost and settlement-time advantages of stablecoins and tokenized deposits (see Figure 1).

Figure 1: Potential fee revenue at risk by 2030

Projected fee revenue decline by 2030 (% vs.baseline)
Projected fee revenue decline by 2030 (% vs.baseline)

Due to client demand, banks are showing interest in offering these services, and responding collectively, though these efforts are fragmented and not fully proven. Recent examples (not exhaustive by any means) include the Great British Tokenized Deposit (GBTD), with seven major banks engaged, The Clearing House in the US, with 17 major banks engaged and Quivalis, with 37 European banks engaged4,5,6. Forty global financial institutions and eight central banks have teamed up to develop shared tokenization infrastructure to improve cross border transactions through Project Agora.7

Meanwhile, another question looms: where do banks sit in agent-driven transactions, which typically originate inside corporate systems such as enterprise resource planning (ERP), procurement and treasury management systems (TMS)? Software vendors in these categories are already building and deploying agents as part of their platforms, enabling agents to evaluate payment methods and providers on every transaction in real time, with significant revenue implications. Accenture estimates that $126 billion in cumulative revenue could be moving across payment pools by 2030, with cross-border and e-commerce payments being the most exposed. That makes it more important for banks to choose strategically where to participate in the new rails, so they can retain and grow revenue.

The implications of interoperability and new options

That said, the speed of this shift will be determined by several market developments. Some examples include interoperability across digital currency rails, maturity of the agentic schemes and the wider adoption of new payment options such as account-to-account (A2A) payments. Common industry global standards for agent authentication, intent verification and liability rules remain unresolved. Without these foundations in place, agents have limited ability to route around traditional and new rails.

When those conditions are met, the dynamic changes quickly. Agents could offer corporate clients real-time alternatives and execute payments via A2A or digital currency rails. Whoever provides the agent, whether a bank, an ERP vendor or an external platform, would be best placed to own the client relationship and influence the associated revenue.

This contest is already taking shape inside enterprise software. Oracle’s Payments Agent in Fusion Cloud ERP evaluates payment options and supplier offers, including dynamic discounting and virtual cards, to optimize cash outflows and execute payments, with supply chain financing capabilities on the roadmap.8 SAP Pay embeds payment execution and reconciliation directly within invoice workflows, while Mastercard and Oracle are integrating commercial payment capabilities into enterprise processes.9,10

For banks, the question is whether their products stay visible and selectable as agents compare terms and initiate payments under pre-approved policies. Fifty-nine percent of corporate treasury leaders identify integration with ERP, TMS and other platforms as one of the biggest sources of complexity in payments today, and reconciliation across multiple banks and payment providers adds further friction. Seventy eight percent say better APIs and integration standards would improve interoperability. And the same number (78%) of banks are already investing in API-based integration and open-banking capabilities. Yet progress is slow, and banks’ risk of losing clients rises with every year the gaps remain.

The costs of delay are already visible. In the US, FedNow and RTP, the country’s two instant payment rails, are not interoperable, so many banks run both, adding integration costs and operational complexity.11 That fragmentation also pushes corporates toward independent orchestration layers. Şişecam, which operates across 14 countries with more than 70 bank connections, centralized its payments through Kyriba12. It achieved 95% straight-through processing without manual intervention, plus visibility, payment control and fraud detection across its banking relationships. As such layers become the primary interface for managing payments, banks’ direct influence over payment decisions could weaken.

The fraud problem: Current controls are not built for machine speed

Corporates rank AI-driven fraud and impersonation as their single greatest operational risk in adopting agentic payments. Banks share this concern, yet only 14% report fraud systems trained on both human and machine transaction profiles.

Fraud systems trained on behavioral patterns cannot flag a transaction that is behaviorally normal and genuinely authorized. And agent-initiated payments compress the time available to identify and stop fraudulent transactions. Banks will need controls that verify intent and mandate alongside credentials and transaction behavior, not just flag anomalies after the fact.

The promising foothold

Corporate clients still see their banks as key players in payment decisions. When respondents in our survey were given the option to select more than one answer, 63% of treasury leaders said they expect internal treasury teams to remain the most influential players, 60% point to banks and 52% to ERP and TMS providers. That data signals that banks have a real foothold in the decision path.

Banks are well positioned to compete because they already hold trusted roles in payment execution, liquidity and risk management, and 92% of corporates trust their bank to execute payments. But trust alone does not defend revenue.

Right now, most banks are investing behind the scenes, in infrastructure, authorization, risk control and orchestration. Only 39% plan to offer client-facing AI agents, leaving a gap at the point closest to the customer.

The airline industry shows where that gap can lead. Airlines still fly the planes, but aggregators decide which flight a customer sees first and capture the margin on choice. Banks could likewise keep providing the funding and executing payments while ERP vendors or software platforms steer payment selection through their AI agents, weakening banks’ influence over client relationships and their ability to cross-sell and up-sell.

Pivot to capture the opportunity

By investing strategically in digital currencies and agentic commerce, banks can lower their operating costs, develop differentiated client propositions and capture revenues that are currently out of reach.

The largest prize sits in the cash behind B2B payments. Digital currencies allow for continuous and programmable settlement, and agents decide when and where cash moves at speed. Together they can release liquidity trapped in multiple areas, including these examples: accounts prefunded for cross-border payments, invoices awaiting settlement and operating balances that earn little. Across these areas alone, we estimate $1 trillion of liquidity could be unlocked by the end of 2030 (Figure 2 offers detail).

Figure 2: Scenario of liquidity unlocked by 2030 and cumulative interest income, 2027–2030

Area

Scenario

Liquidity unlocked by 2030Cumulative interest income, 2027–2030
B2B cross-border prefunding25% of the roughly $430 billion tied up in prefunded accounts is released~$110 billion~$9.6 billion

Corporate working capital

Customers pay one day faster on about $15 trillion of receivables~$275 billion~$24 billion

Corporate idle operating cash

25% of a $2.2 trillion pool of excess operating cash is put to work by 24/7 and automated sweeping~$571 billion~$49 billion
Total ~$1 trillion~$84 billion

Source: Accenture analysis using the Accenture Payments Revenue Model, supplemented by Federal Reserve Z.1 data. Illustrative estimates, not forecasts. See endnote 3.

Prefunding benefits accrue first to banks, which hold the pre-positioned balances, and reach corporate clients through faster and cheaper payments. Working capital and operating cash sit with the clients themselves, and banks that offer the sweep and early-payment terms keep the relationship. Because stablecoins cannot pay interest under the US GENIUS Act, capturing idle cash depends on tokenized deposits and money market funds.

Beyond liquidity, corporate clients will also pay for better services. We estimate that global commercial payments generate approximately $1.25 trillion in annual revenue, and corporate clients are willing, on average, to spend 8.1% of their annual payments costs on value-added services, implying a total addressable pool of roughly $100 billion.13

Integrated treasury and payments orchestration is the strongest demand signal, selected by 80% of corporate treasury leaders, followed by real-time cross-border liquidity services (39%), AI-driven payment intelligence and advisory (37%) and agentic payments governance and control (36%). If banks captured only one-third of the orchestration segment, a conservative assumption given their existing client relationships, the annual revenue opportunity could exceed $25 billion.

Trust that is not converted into a competitive position in the decision path is goodwill, not advantage. It does not defend revenue on its own. And priorities vary by sector. Retail companies emphasize AI-driven payment intelligence and supply-chain visibility, travel and airline businesses real-time cross-border liquidity, and health and public-service organizations embedded compliance and regulatory reporting. Banks can tailor propositions accordingly (see Figure 3).

Figure 3. Value-added services to address payment and treasury challenges

Value-added services to address payment and treasury challenges
Value-added services to address payment and treasury challenges

Four moves banks can make now on digital currencies and B2B agentic commerce

01

Prioritize the areas where corporate demand is moving fast

Accelerate product development and market launch in the areas with the greatest demand and revenue at risk: cross-border payments and treasury management. Integrated treasury and payments orchestration are the strongest demand signal, selected by 80% of corporate treasury leaders.

For digital currencies, prioritize B2B cross-border capabilities and yield-bearing liquidity services, such as automated sweeps into tokenized deposits or money market funds. These address the prefunded accounts and idle operating cash behind the $1 trillion liquidity opportunity. For agentic commerce, define where your agent sits in the B2B payment flow, whether inside corporate ERP and procurement systems or as a bank-owned agent clients can access, before others set that architecture. With only 39% of banks planning client-facing agents, this is where an early move can differentiate.

Validate demand by sector and corridor, run industry-specific pilots with anchor clients, and be transparent with clients about delivery timelines.

02

Influence the payment orchestration, not only its execution

Make products, pricing and APIs easy for AI agents to discover and select, whether those agents are bank-owned or embedded in ERP and procurement platforms. Banks that control the decision layer have opportunities to cross-sell and deepen relationships. Those that do not might become infrastructure.

03

Start building a resilient infrastructure

Frame your investment case for modernizing infrastructure around what’s coming, not just what’s currently compliant. Address interoperability, fraud prevention and create a minimum viable service at the same time. As agent-initiated B2B transactions accelerate payment flows, banks will need controls that verify intent and mandate alongside credentials and transaction behavior. Resolve on-ramp and off-ramp challenges and improve interoperability between models at the industry level.

04

Help clients build the business case

Stronger proof of business value is the leading accelerator of adoption. Sixty-two percent of corporate clients say they need help making that case, quantifying ROI, understanding which use cases apply to their business and building a phased adoption roadmap. The starting point for most corporates is integrated treasury and payments orchestration. From there, banks should adapt their offering by sector, because what retail, travel, insurance and health each need are materially different. 

Claim a strong stake in the future

The window is narrower than it looks. Agentic payment infrastructure is taking shape now, built largely by non-banks. Digital currency rails are moving from pilot to production, and software vendors already offer the orchestration corporates want. Corporate clients have made clear what they want from their banks: integrated orchestration, real-time liquidity and help building the business case. It’s time for banks to claim a leading position in the next phase of B2B payments.

With contributions from Hannes Fourie, Accenture Research Payments Lead.

About this research

In July 2026, Accenture surveyed 205 banking executives and 230 corporate treasury leaders across four global regions.

Sources and notes

1 All survey figures in this article are drawn from Accenture Money in Motion Research, August 2026 (N=205 banking executives; N=230 corporate treasury leaders) unless otherwise stated.

2 Revenue-at-risk figures draw on Accenture’s global payments revenue model, which combines bottom-up and top-down analysis leveraging secondary data from GlobalData, the World Bank, the European Central Bank and Juniper Research, alongside internal and external expert interviews. These are illustrative estimates, not audited financial projections.

3 Liquidity estimates are illustrative Accenture analysis, not forecasts or audited projections. Prefunding: annual cross-border flows of about $100 trillion (Accenture Payments Revenue Model), 45% routed through prefunded accounts and held for 3.5 days of flow, with 25% released. Working capital: US nonfinancial corporate trade receivables of about $6.0 trillion (Federal Reserve Z.1, Q1 2026), scaled to 2.5 times the US level, a 55-day collection period and collections one day faster. Idle cash: US business checkable deposits of about $3.0 trillion (Federal Reserve Z.1, Q1 2026), 30% excess to daily needs, scaled to 2.5 times the US level, with 25% put to work through digital rails and agents. Value assumes a 3.5% yield; adoption reaches 25%, 50%, 75% and 100% of its 2030 level across 2027 to 2030.

4 UK Finance, Tokenised sterling deposits – GBTD initiative, September 2026.

5 The Clearing House, Major Financial Institutions Unveil Bank-Led On-Chain Money Initiative, June 5, 2026.

6 CoinMarketCap, 37 Banks Back Qivalis Euro Stablecoin Before 2026 Launch, 2026.

7 Bank for International Settlements, Project Agorá, May 27, 2026.

8 Oracle, Oracle Introduces Fusion Agentic Applications for Finance and Supply Chain, April 9, 2026.

9 SAP, SAP Advances Finance Operations with Embedded Payments, New Consolidation Capabilities, and Expanded AI, October 12, 2026.

10 Mastercard, Mastercard simplifies business payments with Oracle integration, September 11, 2025.

11 Banking Dive, Fed, TCH joust over interoperability, May 7, 2024.

12 Kyriba, Şișecam’s awarded playbook: 95% STP, real-time cash, and fraud control with Kyriba, 2025.

13 Accenture, Reinventing commercial payments for profitable growth, September 13, 2023.

WRITTEN BY

Sulabh Agarwal

Managing Director – Global Payments Lead

Srikant Yennamandra

Managing Director – Digital Assets and Currencies, Global Lead