Summary
- Franklin Templeton asserts that traditional payment systems are inherently inadequate for agentic AI.
- The firm highlights that blockchain technology offers a significant "speed gap" advantage.
- Sandy Kaul believes that to fully harness the economic benefits of the AI revolution, investments in cryptocurrencies such as Solana and Ethereum are essential.
Sandy Kaul, who leads Digital Assets and Innovation at Franklin Templeton, has a clear message for those who think investing in Nvidia stock is sufficient for AI exposure: It is not.
In a paper released on Tuesday, Kaul posits that agentic AI—software that autonomously acts, pays, and makes decisions without constant user input—will operate on cryptocurrency infrastructure rather than traditional financial systems. Franklin Templeton oversees nearly $1.8 trillion in assets.
"Blockchain will be crucial in enabling agentic AI to realize its full potential for consumer transactions, and the rise of agentic AI is poised to be the 'killer' application that promotes blockchain adoption," the asset manager stated in a social media post.
The argument relies on the capabilities of AI agents, according to Franklin Templeton. These agents extend beyond the functionalities of basic chatbots; they can autonomously shop, make bookings, and complete payments once granted user permissions. A report from the AI consultancy Capgemini, referenced in the paper, describes this evolution as a transition from reactive chatbots to autonomous systems capable of understanding their surroundings, formulating plans, and executing complex tasks independently of constant human oversight.
This is the essence of agentic AI: performing actions autonomously rather than merely providing information.
According to a forecast by Bain & Company, also cited by Kaul, "AI agents are projected to represent 15% to 25% of all U.S. e-commerce sales by 2030."
Conventional payment systems, which are designed for human-paced transactions, are unable to manage the volume of transactions needed to support a world where agentic AI becomes commonplace. The sheer number of transactions would overwhelm banks' processing capabilities.
Blockchain networks, on the other hand, utilize distributed ledgers that can record and finalize transactions simultaneously without the need for a bank intermediary.
“While bitcoin processes approximately 7 transactions per second and Ethereum around 75, newer high-speed chains can achieve maximum speeds between 12,933 transactions per second (TPS) on Aptos, 6,284 TPS on Solana, and 3,252 TPS on BNB Chain,” the post notes. “These speeds are comparable to Visa's network, which handles between 1,700 and 10,000 transactions per second in typical operations.”
“However, this comparison is somewhat misleading. Blockchains both record and settle transactions within that TPS range, whereas the Visa network only records transactions. Settlement on the Visa network can take 1 to 3 business days.”
This disparity in speed is crucial when software is tasked with executing thousands of micropayments every hour. Coinbase has introduced tools that enable AI agents to trade and make payments autonomously. Google has announced a payment protocol for agents set to launch in 2025, supported by the Ethereum Foundation.
The x402 Foundation, comprising 40 organizations including Visa, Mastercard, and AWS, is developing open payment systems for AI and officially launched on July 14. This protocol revives a previously unused HTTP status code from 1991, which was intended for web payments that never materialized, to facilitate direct software-to-software payments over the internet.
If each agent purchases data, computing power, and API access using a blockchain's native cryptocurrency, demand for tokens like Solana and Ethereum could increase in line with transaction volume.
Kaul's assertion is straightforward: "I believe it will become increasingly evident in the coming years that to capitalize on the advantages of decentralized networks and businesses, investors will need to acquire the cryptocurrencies and altcoins issued by those networks."
Agentic commerce—AI systems conducting transactions on behalf of humans, such as purchasing cloud computing resources or booking travel—is anticipated to reach a market size of between $3 trillion and $5 trillion by 2030, according to a report from McKinsey & Company.