Instagram chief Adam Mosseri says the way companies use AI may soon require tighter financial discipline, including possible token caps for individual engineers at Meta.
Speaking on Lenny's Podcast, Mosseri suggested that within the next one or two years, the cost of AI usage could become significant enough to manage like payroll, storage, or computing capacity. In his view, token spending should be treated as a measurable business resource rather than an open-ended experiment.
AI token spend refers to the cost of processing prompts and responses through AI models. Meta has already begun adjusting its internal approach, including ending an employee token-spending leaderboard after costs climbed sharply. Mosseri said the company does not currently impose token limits on staff, but he believes that could change as usage grows.
The broader tech industry is also rethinking AI budgets. Companies are increasingly comparing model usage with other operational expenses, especially as teams rely more heavily on coding assistants and generative tools. Mosseri expects that competition among AI model makers may eventually push prices lower, making access more efficient over time.
For now, Meta is focusing on removing low-value uses and directing resources toward work that delivers clearer returns. The shift points to a future where AI access may be managed with the same precision as other core business assets, shaping how teams build, test, and scale digital products.
