Blockchain is in a trough, but AI agents may be the force that pulls it out

Blockchain is in a trough. Jerry Murdock says so plainly, and few observers of the technology cycle would argue with the diagnosis. The hype that once surrounded distributed ledgers has cooled into something closer to ambient skepticism, with the technology neither dead nor dominant. Murdock’s separate contention is that this trough is a temporary position, not a final verdict, and that the force capable of moving blockchain out of it will be autonomous AI agents that need to settle payments with each other.

That is a specific, checkable bet. It depends on a premise that is already taking shape: AI agents operating autonomously need a payment rail that does not require human sign-off on every transaction. Traditional financial infrastructure was not built for that. It was built for human counterparties, with settlement cycles, authentication layers, and oversight mechanisms that assume a person is somewhere in the loop.

The structural mismatch is visible even at small scale. An autonomous agent completing a complex task might need to pay for data, computing time, and downstream model calls in the course of a single job. That granularity of settlement, at machine speed and without human intervention, is what existing rails handle poorly. Blockchain’s programmability and its capacity to execute transactions without a human intermediary in the loop is precisely what makes Murdock’s framing coherent rather than merely hopeful.

Blockchain for agent payments. Blockchain I is in the valley of disillusion right now. Jerry Murdock

His call is interesting because it depends on two things going right at once. First, AI agents must actually proliferate to the point where machine-to-machine commerce becomes a material economic activity rather than a theoretical construct. Second, blockchain infrastructure must be ready, cheap, and fast enough to serve as the settlement layer when that commerce arrives. Neither is guaranteed. The agent economy is developing quickly, but quickly in technology terms can still mean years of infrastructure build-out before transaction volume reaches the level that makes any payment rail indispensable.

The case against Murdock’s call is not that blockchain is unsuited to the task. It is that the window for becoming the dominant rail may be contested. Closed or permissioned systems could serve some portion of agent-to-agent settlement without delivering the broader blockchain revival his framing implies. The question is whether the openness and programmability of public blockchain infrastructure turns out to matter enough that agents, and the developers building them, choose it over alternatives that carry less historical baggage and fewer scalability constraints.

There is also a timing question that the trough itself raises. Technologies in the valley of disillusionment have a way of sitting there longer than advocates expect. The trough is defined by the gap between a technology’s actual capabilities and the inflated expectations that preceded it. Closing that gap requires real adoption producing real outcomes, not just a new use case that sounds compelling in the abstract. If agent-to-agent payments become that use case, the timing of the blockchain recovery is tied directly to the pace of the agent economy’s growth. Slow agent adoption means a longer trough.

What makes the prediction worth tracking is that the underlying premise, that AI agents will need to transact with each other at a scale and speed that existing financial plumbing cannot handle, is not speculative in the way blockchain predictions have historically been. The agents are arriving. The payment problem is structural. Whether blockchain is the answer that wins, or one answer among several, will be visible in the infrastructure choices developers and companies make in the coming years. Murdock has named the bet. That is enough to hold it to account when the evidence arrives.