Why Traditional IT Valuation Metrics Fail in the Age of Generative AI

A new Reuters Breakingviews column is putting a finger on a problem quietly bothering corporate IT departments: the familiar yardsticks for measuring what the largest software vendors are worth were not built for the AI era. In a piece dated August 14 and headlined "Wanted: New value metrics for IT giants in AI age," the column argues — as its title suggests — that investors and operators are looking for fresher ways to read the health of the companies whose platforms now run the agentic and generative AI stack.

The measurement gap

The historical IT revenue story was tidy. Per-seat licensing, maintenance attach, services renewals — these gave procurement and finance a clean reconciliation between what they spent and what showed up on a vendor's income statement. AI deployments are starting to blur that symmetry, with consumption-based pricing, bundled "copilot" offerings, and the swing from pilot to production that does not map cleanly onto the old categories. Items that once lived in neat rows on a waterfall chart are now sharing real estate with token charges, inference commitments, and platform credits.

Here is the catch for IT leaders: if the finance counterpart cannot tie AI spend to a clearly defined operational outcome, the next budget conversation becomes harder, not easier. Procurement teams begin asking about run-rate versus measurable returns, and "value" turns into a negotiation metric instead of a celebration one. For organizations, the workflow friction of untangling which spend was growth and which was migration is real, and it usually lands back on the CIO's desk right when the board is asking the hardest questions.

What corporate IT should watch

In practice, the column's premise lands directly on the people who have spent two years piloting AI tools with somewhat fuzzy attribution. A few things worth tracking:

• Vendor disclosure language. Watch whether major platforms start calling out AI-attributed revenue, agent seats, or token consumption as their own disclosed line. The first mover here often ends up setting the de facto market standard for how everyone else has to break out their numbers.
• Internal accounting discipline. Treat pilot spend as an investment line, not an operating one, while the measurement framework matures. That gives finance a clean before-and-after and reduces compliance headaches at audit time.
• Wiring outcomes in from day one. Tie every rollout to a metric a line-of-business owner already cares about — cycle time, defect rate, handle time, customer response time — so that when new value metrics arrive from the analyst community, the team is already fluent in the language and can defend the spend without translation.

The deeper point is not that the IT giants lack data; it is that the old definitions of vendor value — users, seats, renewal rates — were built for a world where software largely sat on a shelf and was licensed by the box. AI demands a tighter feedback loop between what the technology does in production and what the organization gets back. Until that loop shows up explicitly in earnings calls and in the financial filings the industry reads each quarter, expect more commentary of this kind, more silence from CFOs asked the awkward questions, and more pressure on whoever owns the AI line item inside the company.