The AI story moved from the byline to the earnings call.

Cognizant told investors a one-trillion-dollar market could become a five-trillion-dollar one, Infosys disclosed its AI revenue rose while it cut its outlook, and Genpact rewrote its own description into an AI consulting firm.

For most of the year the firms made their AI case in thought leadership. This week they made it on the earnings call, which is a more honest venue, because an investor gets to ask what the number actually is. Cognizant offered the largest frame. On its second-quarter call, chief executive Ravi Kumar told investors the roughly one-trillion-dollar system-integration market could expand into a five-trillion-dollar one as AI moves from pilots into production, and described the firm as an AI builder with more than eight thousand AI engagements and over forty percent of its software development already AI-assisted. Revenue came in at five and a half billion dollars, up four and a half percent, with financial services growing twelve percent.

Infosys told a subtler story in the same window. Its AI revenue mix rose to more than eight percent of quarterly revenue, up from around five and a half percent, which is genuine progress, and it cut its full-year growth outlook to between one and a half and three percent in the same breath. The two facts belong together. A rising AI mix on a falling growth guide is what the deflation thesis predicts: the technology is real, it is landing in the revenue, and it is compressing the price of the surrounding work faster than it is adding to the top line. Even Genpact contributed a small tell, rewriting its own one-line self-description in its results into an AI strategic consulting and digital transformation firm, up from strategy and operations. When a company edits its own definition, it is telling you where it thinks the multiple is.

None of this is hype, and it is important to say so. The addressable-market reframe is directionally right, the engagement counts are real, and the AI revenue is showing up on the disclosures. The firms are not making up demand. They are, however, describing an opportunity in which the client rents the outcome. An eight-thousand-engagement AI-builder practice is eight thousand relationships in which the intelligence, the tooling, and the record live with the builder, and the client holds a deliverable and a dependency.

The reframe is the opening, not the threat. A five-trillion-dollar market for turning AI into production systems is precisely the market Consulting-as-Code was built to serve, from the other side of the ownership line. The same work, encoded as auditable software the client owns, with the record of what it did kept by the client rather than the builder, converts the rented engagement into an owned asset, and the recurring fee into a one-time build the company can run and re-run itself.

So take the earnings calls at their word. The market is as large as they say, the shift from pilots to production is as real as they say, and the proof has finally moved to a venue where it can be checked. The only thing left to decide is who owns what gets built. The firms answered this week: they do. The product answers differently.