A lack of resources forced Google to limit Meta’s access to Gemini

Google was forced to impose restrictions on Meta Platforms’ access to its Gemini AI models. According to Financial Times sources, this decision is dictated by the acute shortage of computing resources of Alphabet, which are needed to ensure the operation of large-scale AI systems.

Why are there restrictions on the use of Gemini?

The situation became an indicator of the extent to which modern AI developments exhaust the infrastructure and energy resources of data centers. Google is actively reviewing the distribution of limits for its partners, with Meta being among those who have felt the restrictions the most. Currently, Alphabet management strongly recommends optimizing the consumption of AI tokens.

Decreasing dependence on Google developments

Until recently, Meta actively relied on Gemini’s capabilities to automate internal processes, in particular for the moderation of malicious content and combating fraudulent activities, where Google’s models showed higher efficiency than Llama’s own efforts. In the current realities, Meta is forced to force the development of its own Muse Spark neural network in order to minimize its dependence on third-party cloud solutions.

Large-scale investments and infrastructure problems

In order to cover the rapid growth in demand for computing resources, Google entered into an agreement with SpaceX at the beginning of June this year. The amount of the deal was 30 billion dollars. The terms of the contract include monthly payments of $920 million for the provision of capacity through mid-2029.

Meta Transformation: Priorities and Losses

Artificial intelligence is a top priority for Mark Zuckerberg, so investment in this area continues to grow despite tough austerity measures. In May, the company embarked on a large-scale restructuring, cutting about 10% of the staff (about 8,000 jobs). About 7 thousand employees were transferred to positions focused exclusively on the development of AI.