Instagram head Adam Mosseri suggested that Meta may need to implement caps on AI token spending per engineer within the next one to two years, comparing it to managing other operational expenditures like payroll.
The potential for AI token spending caps highlights the significant and growing costs associated with AI development and deployment, forcing major technology companies to re-evaluate resource allocation and potentially impacting the pace of AI innovation.
Adam Mosseri, head of Instagram at Meta, has indicated that the company may need to implement caps on AI token spending for individual engineers within the next one to two years. He drew parallels between managing AI token costs and other operational expenditures like payroll, hardware resources, and labeling budgets, suggesting that such limits would become necessary as AI token consumption potentially rivals an engineer's salary.
Mosseri's comments come amid increased scrutiny of AI costs across the tech industry. Meta recently shut down an internal AI token spend leaderboard after the company projected billions of dollars in AI-related expenses by 2026. Other tech giants, including Uber and Microsoft, have also recently adjusted their approaches to AI experimentation due to soaring token costs. Uber reportedly exceeded its 2026 AI coding budget by April, while Microsoft canceled Claude Code licenses, consolidating engineers onto its own Copilot CLI tool.
Meta currently does not have token caps for its employees, but Mosseri believes their implementation could be beneficial. He anticipates that AI token costs will eventually decrease as AI model makers engage in pricing competition to attract users. For the immediate future, Meta has focused on curbing unnecessary AI usage, such as the now-defunct token spend leaderboard, which Mosseri described as creating little value.