Meta was reportedly minding its own business last March, just trying to push the Gemini token, and suddenly Google said it was cut off. This is according to an anonymously sourced story in the Financial Times.
In March, it was revealed that Meta was one of the largest companies participating in the tokenmaxing trend – literally evaluating employees based on how many AI tokens they were using at work. This moment coincides with another fad: for a TeaOxygen-hungry agentic AI platforms like OpenClave were being used by anxious software engineers to achieve apparently unprecedented new levels of workplace efficiency.
Citing “three people familiar with the matter”, the Financial Times now says that Google informed Meta that it was not able to continue its AI use, and placed limits on the company’s use of the Gemini model.
The move has been taken, says the FT:
“…disrupted and delayed some of Meta’s internal AI projects. Because of the restrictions, which are still in place, as well as a broader effort to streamline AI costs, Meta has encouraged employees to be more efficient with AI tokens — the units that measure AI usage, multiple people said.”
In other words, the meta replaced tokenmaxing with judicious token-counting. Sad!
The burden on Google’s resources, meanwhile, could have helped drive Google’s decision to rent compute from xAI’s parent company SpaceX in early June for $920 million a month.
The FT says other big companies also pressured Google’s AI capabilities and had limits placed on them, but it seems those problems weren’t as severe. According to FT’s sources, Meta was also a standout among other AI high-rollers.
Meta and Google declined FT’s requests for comment.
<a href