Meta May Set AI Token Consumption Limits for Engineers
Instagram head Adam Mosseri stated that in the next one to two years, the cost of AI Token consumption for top engineers may approach their salaries or total hiring costs, at which point Meta may set limits based on input-output ratios.
The company currently has no limits and has closed the internal Token consumption leaderboard to avoid encouraging ineffective spending.
Under market mechanisms, AI cost control drives companies to optimize resource allocation, benefiting efficient engineers and high-return projects while limiting low-value usage.
Source: Public Information
ABAB AI Insight
As a heavy user of AI, Meta initially encouraged experimentation through open Token consumption but is now shifting to strict ROI assessments reflecting increased cost pressures.
On the capital front, the company will transition from indiscriminate incentives for computing resources to precise investments, motivated by the need to control explosive spending and enhance overall productivity returns.
Similar to other tech giants tightening AI budgets, Meta is in a phase of transitioning from scale expansion to efficiency prioritization.
This essentially represents capital concentration under technological substitution, with the realization of Token costs prompting funds to concentrate on high-value applications and model optimization.
ABAB News · Cognitive Laws
The era of Token burning ends with ROI calculations.
Leaderboards encourage waste, while limits extract value.
Engineer output is assessed by returns, not consumption.