Why AI Agent Activity May Not Boost Public Blockchain Token Value

Fidelity Bitcoin Moves Highlight Stablecoin Infrastructure Over Public Tokens in AI Strategy

The report’s core distinction is structural: transaction volume can grow while the durable margin accrues to stablecoin issuers, custodians, and closed payment systems. For smart-contract engineers, this shifts the question from throughput to control of the settlement layer.

The signal matters because the proposed demand is not necessarily chain-native. An AI agent will select the cheapest, fastest, and most regulation-friendly rail. That rail may be a public blockchain. It may also be a private or closed system.

Activity is not ownership

Fidelity Digital Assets’ Max Wadington identified six risks to the AI-agent thesis in a report released on August 19, according to TechStock². The main failure mode is a broken economic invariant: more transactions do not guarantee higher value for the token securing the network.

The execution path is straightforward:

1. An agent initiates a payment.

2. A settlement provider selects the rail based on cost, speed, and compliance constraints.

3. A stablecoin issuer, custodian, or payment operator captures fees or control over the account relationship.

4. The public token may receive only indirect demand, if any.

That is not an argument against public chains. It is an argument against treating transaction count as a sufficient valuation model.

The market response already showed divergence. On Thursday, FETH outperformed FBTC by 4.49 percentage points and exceeded Circle’s performance by 4.20 points, despite Fidelity’s analysis suggesting that infrastructure providers could capture more of the economic value. Circle shares rose 6.02% before the market opened, while the stock continued to trade at a reported price-to-earnings ratio of 47.74. Analyst ratings were split between 13 Buy, five Hold, and three Sell recommendations, with price targets ranging from $37 to $175.

Those figures describe price beta. They do not establish long-term margin ownership.

Fidelity is also building the rail

Fidelity’s position is not external criticism. On February 4, Fidelity Digital Assets NA introduced the Fidelity Digital Dollar, or FIDD, designed to track the U.S. dollar on a one-to-one basis. The company provides custody and trading services for the token.

That creates a relevant state mutation in the business model. Fidelity is questioning whether public tokens will capture AI-driven payment growth while simultaneously offering infrastructure that could benefit if stablecoin settlement expands.

For protocol teams, the implementation question is therefore narrower than “will AI use crypto?” The useful questions are:

• Which contract controls issuance and redemption?
• Who performs custody?
• Where does compliance enforcement occur?
• Which component receives the fee?
• Can the payment path operate without the public token?
• What happens when the preferred rail changes?

A system that depends on public-token appreciation but routes settlement through an external stablecoin provider has a value-capture dependency. That dependency should be explicit in the protocol’s threat model and revenue model.

The wider infrastructure competition is also moving beyond token settlement. Seeking Alpha reported that former Bitcoin miners are repositioning their power assets and land holdings toward AI and high-performance-computing infrastructure. Stocktwits separately highlighted AI lease deals involving HUT, CIFR, and WULF. The available source material does not establish the economics of those deals, but the direction is clear: some operators are competing for contracted infrastructure revenue rather than relying only on block rewards.

What developers should verify

Fidelity’s broader 2026 assessment also placed Bitcoin mining in competition with AI and high-performance computing for electricity and data-center investment. The report noted an 8.8% decline in the network’s average 30-day hash rate before a rebound.

That metric is not a forecast. It is a reminder that infrastructure has competing uses. Power, custody, settlement, and compliance are allocation problems before they are token problems.

The practical checklist is rigid:

• Model fee capture by component, not by aggregate transaction volume.
• Separate public-chain demand from stablecoin settlement demand.
• Identify every custodian and privileged operator in the payment path.
• Test whether the protocol’s invariant survives a change in settlement rail.
• Treat regulatory action, cybersecurity incidents, and falling interest income as independent risk variables.
• Do not use AI-agent adoption as a substitute for demonstrated cash-flow ownership.

The reporting is not an investment conclusion. It is a warning about abstraction layers. The transaction may be on-chain while the economic control remains elsewhere. That distinction is the part developers should preserve in architecture reviews.

A separate example of infrastructure-led growth can be found in India’s regional tourism infrastructure push, but the engineering lesson is the same: the visible application is not always the layer that captures the durable value.