Franklin Templeton: Agentic AI Will Run on Crypto Rails, Not Wall Street

Agentic AI Needs Crypto Rails, Franklin Templeton Argues

Sandy Kaul, Head of Digital Assets and Innovation at Franklin Templeton, has a direct message for investors convinced that exposure to artificial intelligence through companies like Nvidia covers their AI investment needs: it doesn’t. According to the $1.8 trillion asset manager, a crucial piece of the AI story unfolds not on Wall Street but on blockchain networks.

Kaul recently published research arguing that agentic AI—the next generation of autonomous artificial intelligence—will require cryptocurrency infrastructure to operate at scale. Unlike traditional AI chatbots that respond to queries, agentic AI systems execute actions autonomously. These systems can shop for goods, book services, and process payments on a user’s behalf after receiving initial authorization, all without requiring approval at each step.

Research firm Capgemini, referenced in Franklin Templeton’s analysis, describes this evolution as the shift from “a reactive, conversational chatbot to an autonomous system that can perceive its environment, devise a plan, and execute multi-step tasks to achieve high-level goals without constant human supervision.” The distinction matters: agentic systems don’t just provide information—they independently take action to accomplish defined objectives.

The Settlement Speed Gap

The technical hurdle Franklin Templeton identifies centers on transaction processing capacity. As agentic AI systems proliferate and operate autonomously, they will generate an unprecedented volume of micropayments and transactions—potentially thousands per hour per user. Traditional payment infrastructure and banking networks, designed for human-speed commerce with human approval workflows, cannot accommodate this transaction density.

Blockchain networks solve this problem through their architecture. Distributed ledgers record and settle transactions simultaneously without requiring a banking intermediary, eliminating the settlement delays inherent in traditional systems. The speed metrics underscore this advantage. Bitcoin processes approximately 7 transactions per second, while Ethereum handles around 75 TPS. Newer blockchains operate at substantially higher speeds: Solana processes 6,284 transactions per second, Aptos reaches 12,933 TPS, and BNB Chain manages 3,252 TPS.

However, a critical distinction exists. Visa records transactions within its throughput window of 1,700 to 10,000 transactions per second but settles them over one to three business days. Blockchain networks both record and settle transactions instantly within their TPS window, eliminating the settlement lag entirely. For systems making thousands of autonomous payments hourly, this difference becomes material to functionality.

The Near-Term Timeline

Adoption timelines suggest this infrastructure question will become urgent quickly. Bain & Company, cited in Franklin Templeton’s analysis, forecasts that AI agents will represent 15 to 25 percent of all U.S. e-commerce sales by 2030. This growth trajectory means the blockchain infrastructure gap will become practically significant within the next several years, not decades.

Early moves indicate the market recognizes this need. Coinbase has already launched tools enabling AI agents to execute trades autonomously, demonstrating initial adoption of blockchain-based AI infrastructure. As agentic AI becomes more prevalent in commerce and financial transactions, the settlement speed advantage of blockchain networks becomes increasingly critical to efficient operation. If agentic AI adoption proceeds as projected, blockchain infrastructure could transition from speculative asset class to essential financial rails.

Source: Franklin Templeton, via Decrypt. Not financial advice.