The firms actually building for the new model are the ones paying real money to do it.

Between OpenAI's announcement and its close, what gave it away was which firms took charges and named contracts rather than published positions.

Positions are cheap. A manifesto costs nothing and commits a firm to nothing. In the five weeks between OpenAI's deployment venture being announced and closing, the firms that made the most credible moves were the ones that paid for them. The tell was a charge on the income statement or a contract value on the record, not a stage appearance.

Cognizant is the clearest case. It agreed to buy Astreya, an AI infrastructure and data-center services firm, for about $600 million, and opened a restructuring program, Project Leap, carrying an estimated $230 million to $320 million in costs, most of it severance. In the same breath it said it still plans to hire more than 20,000 graduates this year. That is not a firm trimming at the edges. It is a firm rebuilding its cost base around a different delivery model and paying cash to do it.

ISG shows the other half of the pattern, revenue rather than restructuring. On its first-quarter call it disclosed about $21 million of AI-related revenue, roughly a third of the total and up from $12 million a year earlier, anchored by a record governance contract worth up to $17 million that oversees about $300 million of a global manufacturer's technology spend. Quarterly profit jumped more than 80 percent. Governance advisory stopped being a white paper and became a priced, recurring line, which is the one category that grows as more delivery players crowd the market.

The contrast is the whole point. The firms publishing without a charge or a named contract are making moves that do not require their own thesis to change. The firms taking the charge are betting it already has. When the reprice fully arrives, the second group will be holding receipts.