AI Agents and the Future of Payments: Why Stablecoins Won’t Replace Credit Cards
For years, the stablecoin sector has sought a “killer app” – a use case demonstrating their value beyond cryptocurrency speculation and powering everyday commerce globally. The emergence of artificial intelligence (AI) agents initially appeared to provide that catalyst. A scenario from Citrini Research posited that AI agents bypassing card fees could wipe out billions in market capitalization from Visa, Mastercard, and American Express in a single day. However, the reality of integrating AI agents with payment systems is proving more complex, with credit cards retaining a significant role, at least in the near future.
The Initial Promise of Stablecoins for AI Agents
The core thesis driving interest in stablecoins for AI agent payments stemmed from the need for fast, cheap, and programmable money. AI agents, operating at machine speed, require efficient transaction methods. Stablecoins, pegged to fiat currencies like the US dollar, offered a potential solution by reducing transaction costs associated with traditional card networks. This sparked excitement about a future where AI-driven commerce would be seamlessly powered by digital dollars.
The Limitations of Stablecoins: Trust and Infrastructure
Despite their potential, stablecoins face significant hurdles in replacing established payment methods. Credit cards offer robust features like chargebacks, fraud protection, and dispute resolution – a trusted infrastructure built over decades and relied upon by hundreds of millions of users. Currently, stablecoin payments lack comparable safeguards. Widespread adoption of stablecoins remains limited; relatively few consumers currently hold them, while the vast majority have access to credit cards.
According to Crone Consulting LLC, cryptocurrencies, including stablecoins, are projected to account for only 1% to 5% of gross merchandise value in online and offline retail by 2030, with traditional payment methods handling the remainder. Christian Catalini, a professor at the Massachusetts Institute of Technology (MIT) who was involved in the development of the Diem stablecoin project (formerly Libra), cautions against overestimating the impact of stablecoins, stating, “If you are exaggerating…traditional traders will most likely dominate [agentic trading].”
Card Networks Adapt to the AI Landscape
Recognizing the potential of AI, card networks are proactively adapting. Visa has launched “Visa Intelligent Commerce for AI agents,” currently in testing, and Mastercard’s “Agent Pay” is already live for US cardholders. Rubail Birwadker, Global Head of Growth at Visa, emphasizes the importance of separating the attention surrounding stablecoins in the AI space from the current reality, noting, “Most people don’t have a stablecoin wallet.”
Several startups initially focused on stablecoin-only agent payments are now integrating card networks and banking integrations. X402, an open standard for agent payments incubated by Coinbase, is exploring adding traditional payment methods. Skyfire, a platform for AI agent transactions, has added Visa and is integrating Mastercard, and is also in discussions with banks. Amir Sarhangi, CEO of Skyfire, acknowledges the continued importance of credit cards for consumers, citing familiarity with rewards programs and purchase protection, stating, “On the consumer side, at least for the foreseeable future, credit cards will continue to be an crucial factor.” He believes stablecoins may discover greater traction with corporate clients.
A Hybrid Approach: Stablecoins and Traditional Rails
Even among corporate users, traditional rails maintain a foothold. Sean Neville, co-founder of USDC stablecoin and now running Catena Labs, envisions AI software utilizing single-use cards with spending limits. Catena Labs is testing a system with banks to facilitate secure money movement, lending, and transactions for AI software. Neville suggests a complementary relationship, stating, “It’s true that using stablecoin rails makes a lot of sense for some use cases, but that doesn’t mean you can’t use traditional rails for agentic flows.”
The emerging architecture appears to be a hybrid model: AI agents initiating payments with virtual cards on the front end, while stablecoins settle transactions on the back end. This allows card networks to maintain their customer relationships and fraud protection mechanisms, while leveraging the efficiency of stablecoins for settlement.
Niche Applications for Stablecoins
Stablecoins are finding a niche in specific areas where traditional payment systems fall short. For micro-transactions – such as a developer accessing financial data through an API 40,000 times at a fraction of a cent per request – stablecoins offer a viable solution, as traditional payment processors struggle to underwrite the risk and fees associated with such small transactions. Similarly, agent-to-agent transactions, involving data scraping, API service consumption, and cross-border software workflows, benefit from the economics and global accessibility of stablecoins.
Erik Reppel, Head of Engineering at Coinbase and founder of x402, notes that the platform is currently processing $8,000 to $60,000 in daily transaction volume, primarily in USDC. He predicts that while the total transaction count on crypto rails may be higher, the dollar value will likely remain comparable to that on fiat rails.
The Role of Agent Autonomy
The level of autonomy granted to AI agents will influence the choice of payment rails. Cosmo Jiang, General Partner at Pantera Capital, argues that “The real differentiator is autonomous agents.” Agents with limited human oversight are more likely to adopt stablecoins, while those requiring human intervention may continue to rely on traditional methods.
the future of AI agent payments is likely to be a blend of both stablecoins and traditional payment networks. Stablecoins may not replace the existing financial system, but will quietly operate behind the scenes, providing the infrastructure for a new era of automated commerce.