Companies may be finding real value in AI without yet producing returns that match the scale of industry investment. Technology entrepreneur Azeem Azhar described an informal test of that gap during a talk with about 160 IT vice presidents in Las Vegas.
When Azhar asked who could point to measurable results from AI, roughly two-thirds of the audience remained standing. When he asked whose results were important enough to interrupt the chief executive's summer vacation, only about eight people stayed up. His conclusion was that deployment is advancing, but slowly.
That pace matters because technology companies are funding an enormous expansion of data centers and chips. Sustaining it requires customers to keep buying AI services, ideally at higher prices. Companies can increase usage while moving toward cheaper open-weight models, however, which would spread the technology without necessarily generating enough revenue for frontier labs and infrastructure providers.
The audience exercise is an anecdote, not a controlled survey, and “measurable” can cover very different outcomes. It therefore cannot establish economy-wide return on investment. It does capture a useful distinction for buyers: evidence of any benefit is not the same as a result large enough to reshape budgets or strategy.