Franklin Templeton says agentic AI could become the “killer use case” that blockchain and crypto have been searching for over many years, as autonomous AI agents begin to require instant, programmable, and verifiable payment infrastructure without continuous human intervention.
This view was presented by Sandy Kaul, Head of Digital Assets and Innovation at Franklin Templeton, in an analysis published in July 2026, amid payment companies, AI firms, and crypto infrastructure providers building transaction layers for machines. Stripe, the Linux Foundation, and OpenAI have all taken steps related to payments for AI agents, indicating that agentic commerce is moving from a technological concept to its initial infrastructure-shaping stage.
— Franklin Templeton Digital Assets (@FTDA_US) July 21, 2026
Franklin Templeton’s Thesis
Sandy Kaul, Head of Digital Assets and Innovation at Franklin Templeton, does not view agentic AI merely as a new application layer of artificial intelligence, but as a test for whether blockchain can serve real economic needs beyond digital asset trading.
According to Kaul, this is where blockchain could find a more practical role compared to previous speculative cycles. An AI agent may need to pay for data, call APIs, purchase software, book services, or execute multiple small transactions in a task chain. Such transactions may be too small, too fast, or too frequent to pass through traditional payment processes designed for human users.
The asset manager has built a clearer presence in the tokenization space through the Franklin OnChain U.S. Government Money Fund, linked to the BENJI ecosystem. The fund recorded total net assets of $753.24 million as of June 30, 2026, showing that blockchain is being used in a traditional financial product rather than remaining solely in crypto experiments.
The Case for Blockchain-Based AI Payments
When AI agents execute transactions autonomously, the payment system must know whom the agent represents, how much it is authorized to spend, and leave an auditable transaction history when needed.
This gap makes blockchain and stablecoins a notable option for payments between AI agents. The technology can record transactions in a transparent and programmable manner, while stablecoins provide a settlement unit less volatile than typical crypto assets. For small amounts, such as an agent paying a fee for a single API call or data access instance, a pay-per-use instant payment model may be more suitable than traditional subscriptions or invoices.
Crucially, blockchain does not necessarily have to replace Visa, Mastercard, or banking systems. A more practical use case is serving as a supplementary payment layer for transactions that current infrastructure processes sub-optimally: machine-to-machine payments, micropayments, pay-per-use APIs, cross-border settlements, and conditional automated transactions.
Market Signals Behind the Thesis
Franklin Templeton offered this assessment at a time when the market had already shown more concrete signals regarding payments for AI agents. In March 2026, Stripe introduced the Machine Payments Protocol, an open standard co-developed by Stripe and Tempo for agents to pay for resources, APIs, or services via HTTP endpoints while connecting to Stripe’s existing payment infrastructure.
By July 2026, the Linux Foundation announced that the x402 Foundation officially went live after Coinbase contributed the x402 protocol. The foundation has 40 members, including Coinbase, Stripe, Visa, Mastercard, Google, AWS, and Shopify, intending to standardize internet-native payments for AI agents, APIs, and applications.
OpenAI is also bringing agentic commerce closer to mainstream users. Instant Checkout in ChatGPT, built with Stripe on the Agentic Commerce Protocol, allows U.S. users to purchase directly from Etsy sellers within the chat, with over one million Shopify merchants announced to be supported later. With over 700 million weekly ChatGPT users, conversational checkout could become a commercial channel worth watching.
Market forecasts are also reinforcing this story. McKinsey estimates that AI agents could orchestrate $3 trillion to $5 trillion in global consumer transactions by 2030, counting physical goods alone. Gartner predicts that by 2028, 33% of enterprise software applications will incorporate agentic AI, and at least 15% of day-to-day work decisions could be made autonomously by agentic AI.
Risks and What Comes Next
However, “killer use case” remains a thesis that needs to be verified by real-world adoption. As AI agents begin executing transactions autonomously, the difficult question is not just whether the payment technology works, but who bears liability if an agent makes a wrong purchase, gets scammed, or exceeds its allocated limits.
These risks will directly affect the pace of deployment. Gartner has warned that more than 40% of agentic AI projects could be canceled before the end of 2027 due to rising costs, unclear business value, or insufficient risk controls.
In the short term, the key aspect to monitor is whether MPP, x402, and checkout models in AI apps enter enterprise workflows and consumer commerce. If AI agents generate real transaction volume, particularly in small, automated payments, blockchain will have a clearer basis to be viewed as infrastructure for a new layer of commerce. If not, this “killer use case” will remain an attractive idea rather than a proven adoption story.
