Indian CFOs pressed to prove real ROI from AI spend, not just pilots

This column captures a growing tension inside enterprises: boards and CEOs feel they have to “do something with AI,” but CFOs increasingly want proof that those experiments actually move the P&L. The data point that only 8% of firms can properly track AI costs is striking — it suggests that even as models and tooling race ahead, basic financial plumbing and value attribution are lagging badly. In other words, we may be in an AI deployment bubble where perception of transformation outstrips measured impact.

From an AGI race perspective, that misalignment matters because it influences how long boards will tolerate multi‑billion‑dollar AI bets with fuzzy payback. If CFOs start killing underperforming pilots or pushing for standardised ROI dashboards, capital could shift away from vanity use cases toward a smaller number of deeply integrated systems that demonstrably change productivity. That’s arguably good for serious AI progress: fewer proof‑of‑concepts, more real deployments with feedback, but also a harder funding environment for speculative projects.

In India specifically, where policymakers like Piyush Goyal are pitching AI as a net job creator, the CFO lens is a useful counterweight. It suggests that the country’s AI story will be decided less by grand narratives and more by whether mid‑market firms can wire AI into ERP, finance and operations in a way that shows up in cash flows.