In the rush to adopt AI, many teams celebrate rising token counts and GPU usage as success. But this post argues that such metrics often reflect 'shovel sellers'—infrastructure providers—rather than actual miners finding gold. The real question is whether your AI application solves a user problem and creates tangible value. For product leaders, this means shifting focus from consumption metrics to outcome metrics. It's a useful framework for evaluating AI investments and avoiding the trap of vanity metrics. The author encourages teams to ask: are we building something people truly need, or just feeding the hype? This perspective is especially relevant for startups and enterprises navigating the AI landscape.
A critical look at AI adoption metrics, distinguishing infrastructure growth from real user value.