California, United States – Meta is considering reorganizing the use of artificial intelligence tools within the company by setting a cap on the consumption of AI tokens per engineer over the next two years. This step comes amid a continuous rise in model operating costs, as part of the company’s efforts to strike a balance between supporting innovation and managing spending on technical infrastructure.
Dedicated Budget for AI Usage
Head of Instagram, Adam Mosseri, stated in remarks cited by “agencies” and followed by “Voice of the Emirates,” that the cost of using AI tools may soon reach a level comparable to the cost of hiring the engineer themselves. He explained that this makes allocating a specific budget for each employee to use these tools essential, similar to salaries or other technical resources within the company.
Mosseri added that AI resources should be managed in the same way as Graphics Processing Units (GPUs), Central Processing Units (CPUs), storage capacities, and memory. This will ensure that these resources are distributed among teams according to work priorities and the expected return from their use.
Usage Cap Will Not Be Uniform
The Head of Instagram explained that any future limits on the use of AI tokens will not be applied uniformly to all engineers. Instead, these limits will vary based on each employee’s ability to utilize these tools to achieve added value and a positive return for the company, ensuring resources are directed to the most productive uses.
He confirmed that Meta does not currently impose any restrictions on the internal use of AI tools. However, the continued expansion of reliance on them and the exorbitant rise in operating costs may push the company to adopt this regulatory approach in the coming period.
Rising Costs Prompt Resource Reorganization
These trends come at a time when major technology companies are facing mounting pressure due to the inflation of operating costs for AI models, as processing requests and responses has become one of the largest expenditure items on infrastructure for these companies.
Meta had previously discontinued an internal dashboard that displayed the ranking of employees who consumed the most AI tokens, after the experiment showed it encouraged excessive use without achieving real operational value. Mosseri suggested that the costs of using AI models are likely to decrease in the long term as competition among developing companies intensifies, but he stressed that prudent expenditure management will remain a core element in tech corporate strategies over the coming years.
