In a recent interview, Adam Mosseri, CEO of Instagram and Meta executive, suggested that in just one or two years, there might be an upper limit on the amount of AI token spend per engineer. This move comes as soaring costs have already forced other tech giants to reassess their AI budgets.

Metas decision to shut down its internal AI token spend leaderboard is indicative of a shift towards more controlled expenditure. Mosseri believes that AI tokens, like payroll or operating expenses, should be managed as resources, with proportional caps for each engineer based on their expected ROI-positive usage.

The cost implications extend beyond Meta; Uber and Microsoft have also faced budget overruns due to skyrocketing token costs. As models improve, the hope is that prices will eventually come down, but in the meantime, companies are reining in expenses by ceasing non-essential spending.

“It’s not that hard to build a token incinerator,” said Mosseri, referring to shutting down wasteful expenditures. This shift could signal a broader trend towards more responsible AI development and deployment, but it may also slow down some cutting-edge research and experimentation.

Metas approach is currently experimental, with no current caps in place. However, as the tech landscape evolves, token budgets for engineers might become a standard practice, reflecting a new era of resource management in the AI-driven world.