Position ledger

The AI value gap is a governance failure, not a measurement impossibility. Attribution at task level is not technically impossible; it is organisationally uncomfortable.

Assessed on 6 October 2026. The State of Tokenomics found that organisations with a named owner and metered, attributable spend were far more likely to show value to the CFO, which is consistent with the position. The findings are correlations, so they support it rather than prove it. Status: holding.

Where it appears

Total AI invoices will rise for at least the next three years regardless of unit prices, because usage expansion dominates price decline.

Assessed on 15 October 2026. No verified evidence on enterprise AI invoice totals either way. Current provider price lists show per-token prices that are not falling uniformly: one major provider lists output prices from $0.50 to $50 per million tokens after cutting its small-model price tenfold on 7 October 2026, and one Gemini model has a scheduled price doubling on 1 January 2027. That is consistent with, but does not test, a claim about total spend. Unverified token-volume figures were not used. The falsification test stands: if total AI invoices fall in 2027 or 2028, this position was wrong.

Where it appears

The token is the meter, not the value. Token counts measure consumption and make AI priceable; they do not show that anything valuable happened, and rewarding token volume rewards consumption.

Assessed on 8 October 2026. The market has moved towards this position since June. The Tokenomics Foundation's launch release (4 August 2026) describes tokens as a unit of usage and proposes measuring cost to serve per call rather than per token. Large employers including Uber have dropped internal token-usage leaderboards (Fortune, August 2026), and practitioner commentary from FinOps X 2026 makes the same distinction. This is now close to consensus, so the position is holding but no longer distinctive.

Where it appears