The AI honeymoon is over — CFOs want returns now
The mood around enterprise AI shifted in 2026. Budgets kept climbing, but as Computerworld reported — citing Forrester — most organizations still cannot show a sustained return. Pilots look great, then the economics fall apart at scale. The problem is not that the AI fails; it is that companies are applying SaaS-era financial models to a technology that bills by consumption.
Two things follow. AI spend is no longer “additive” — as one CIO put it, “if we want to spend more, we have to move things around.” And analysts increasingly describe AI as infrastructure, not software: something you operate and account for, not a subscription you forget. The money is concentrating on internal automation and private-data work, and pulling back from generic chatbots.
- If AI is infrastructure now, own the infrastructure. A flat, owned model is the one a CFO can actually forecast — no per-token meter that punishes success.
- The work getting funded — internal automation, private-data intelligence — is exactly the work that shouldn’t leave your building.
Reporting: Computerworld. Run your own numbers on the cost calculator, or see flat pricing.