Federal Reserve Governor Christopher Waller picked Nov. 17 to sketch where the Fed tokenization debate stands, and his conclusion was refreshingly unglamorous: the boring pipes work, so build on them. Speaking at The Clearing House Annual Conference in New York, Waller argued traditional payment rails have earned their place as the trusted floor beneath tokenized markets – while fellow Governor Michael Barr, speaking a day earlier in Milan, offered a blunt reminder that the AI layer comes with its own stability story.
Waller called the moment a payments revolution “we have not seen for a long, long time,” but kept returning to plumbing over promises: large-value and instant payments “have proven resilient through market events” and remain “a reliable backbone for final funds settlement on which innovative services, including tokenized markets, can develop.”
Fed tokenization vision: traditional rails stay the backbone
Waller’s thesis is integration, not competition. Central bank money, he said, “need not be used in every transaction or on every platform, but its presence at the foundation of the system is essential.” He walked through how stablecoins and tokenized deposits differ in issuer, regulation and links to the traditional system, and noted corporate treasurers will likely run several instruments at once rather than pick a winner.
He was at his most concrete on atomic settlement: imagine a payment and a tokenized share programmed to change hands simultaneously, stripping out settlement delay and risk. Technical capability alone, he cautioned, won’t win institutional trust – systemically important transactions may need conservative designs that favor privacy and control over the developer freedom of fully open networks. No policy changes were announced; the remarks reflect his own views, not the Board’s.
Barr’s warning: AI agents could accelerate bank runs in minutes
Barr’s Nov. 16 speech in Milan supplied the counterweight. AI agents, he warned, could “instantaneously detect yield differentials or stability signals and autonomously move funds at high speed, potentially accelerating runs that unfold in minutes, overwhelming traditional liquidity management and regulatory intervention timelines.” Major banks already report finding AI agents attempting actions in customer accounts without verifiable consent. His prescription: tight authentication, explicit consent checks and hard limits on what agents can do – noting some models treat security restrictions as “obstacles to overcome rather than rules to be followed.”
Barr also cited the Fed’s Nov. 13 Financial Stability Report: off-balance-sheet liabilities tied to the AI buildout now exceed $3 trillion, debt at the seven largest AI firms is on track to pass $800 billion this year, and $100 billion-plus in equity investments into AI labs and clouds creates what he called circular financing.
What it means for stablecoins and tokenized deposits
Both governors arrived at the same destination from opposite directions: tokenized money and AI agents should be integrated with trusted rails, not allowed to quietly route around them. Waller closed by noting tokenization and AI agents in payments could be catalysts for one another – a nice vision, so long as the agents learn to respect the speed limit.
