Four of the largest firms turned the cost of AI into a service, in one week.
The field spent three years selling what AI can do. This week it started charging to count what it costs.
For three years the pitch was capability, and this week it turned into a meter. Accenture launched a practice it calls Tokenomics and put a name to the thing every buyer had been feeling: the agentic Jevons paradox, the pattern where the cheaper a unit of intelligence becomes, the more of it a system consumes, so the total bill climbs even as the price of a single token falls. On the same tape EY published the fifth wave of its AI Pulse survey, a poll of five hundred and thirty-four senior leaders, and found that ninety-eight percent of those investing in AI said token usage and its costs had forced them to reconsider their approach. KPMG advanced its own work on standardizing how tokens are counted, and Cognizant tied its economics to the workforce it is certifying. Four firms, one week, one message: the cost of running AI is now the operating question.
The firms were, for once, right about the diagnosis, and it is worth being precise about it. Only sixty-four percent of the leaders EY surveyed said their organization actively monitors token usage with clear budget guardrails, which means a third of the companies buying AI at scale cannot yet see the bill they are running. Dan Diasio, who leads AI consulting at EY, put the turn in one line: the claim that AI saves time is no longer a sufficient business case when the costs are mounting and hard to ascertain over the long run. A year ago the barrier to AI was belief. This week the barrier is arithmetic.
So the honest question is not whether the diagnosis is correct. It is who ends up owning the cure. A tokenomics practice is a service that reads your meter, tunes your routing, and hands you a report every quarter on a bill you still do not control, because the meter belongs to the supplier and the discipline belongs to the firm you retained to run it. The point was underlined on the fourth of August, when JPMorgan and Accenture joined a venture to standardize how token use is measured across the industry. Standardizing the meter is useful. It also confirms whose meter it is. Manage the cost that way and you have rented a second layer of dependency on top of the first, and when the model price moves again, and it moved twice this week, the practice bills you to re-tune.
There is a different answer, and it starts one level below the meter. Most of the work a company automates is deterministic, the same steps against the same inputs producing the same output, and deterministic work does not need a model at all once it is written as code the client owns. Encode the procedure, run the deterministic part without spending a token, and reserve a model for the narrow place where judgment is genuinely required, on whichever one is defensible that quarter. Now the meter runs on a sliver of the work instead of all of it, the bill is bounded because most of the work was taken off the model, and the record of what the system did belongs to the company rather than to the firm reading its usage.
This is the quiet significance of the week. When the four largest professional-services firms simultaneously reframe AI as a cost to be governed, they are conceding the ground this publication has argued from the start: the model was never the moat, and the spend was never self-justifying. Their answer is to sell the governance as a retainer. The durable answer is to make the governance an owned asset, a procedure the client keeps and a record it can inspect, so the cost is not managed after the invoice but designed out before it. The field spent this week learning to count the cost. The product is the thing that does not run it.