Meta is rolling out new spending controls and usage tracking to manage its artificial intelligence costs, warning 6,000 employees about the rising expenses. The company plans to introduce token budgets by 2027 and an AI Gateway dashboard to monitor usage across teams.
A warning to 6,000 employees
The company told 6,000 staff that AI costs are surging. The warning came as part of a broader push to get a handle on spending. Meta didn't disclose the exact dollar figures or the current rate of increase, but the scale of the internal alert — reaching thousands of workers — signals the issue is a priority.
Employees across multiple teams were notified. The company is asking them to be more mindful of how they use AI resources. The move follows a pattern of cost-cutting measures Meta has taken over the past year, including layoffs and budget tightening in other areas.
Token budgets and a new dashboard
By 2027, Meta plans to introduce token budgets. These budgets would cap how much AI computing power each team or project can consume. Tokens are a common way to meter usage of large language models and other AI services. The system would force teams to prioritize which AI tasks get funded.
Alongside the budgets, Meta is building an AI Gateway dashboard. The dashboard will track AI usage across teams in real time. It's meant to give managers visibility into who is using what, and at what cost. The company hasn't said when the dashboard will go live, but the token budgets are set for 2027.
The dashboard and budgets are part of a broader effort to avoid runaway spending. AI models are expensive to train and run, and Meta has been investing heavily in them. The company's AI research division, FAIR, and its product teams have been competing for computing resources. The new controls aim to bring some order to that competition.
Meta didn't say whether the token budgets would apply to external developers using its AI platforms, or only to internal teams. That detail remains unclear. The company also didn't specify what happens if a team exceeds its token budget — whether projects get paused or simply flagged for review.
The warning to 6,000 employees and the planned controls come as Meta faces pressure from investors to show that its AI spending is generating returns. The company has been pouring money into AI infrastructure, including data centers and chips. The new cost management tools are a sign that Meta is trying to balance ambition with discipline.
