Nvidia expands into Intel and AMD territory with Meta partnership

Meta is strengthening its collaboration with Nvidia, in what the AI chip leader calls a "multigenerational" partnership.

The arrangement involves Meta establishing data centres powered by millions of Nvidia's existing and future chips designed for AI training and operations.

This development highlights how Meta is increasing its reliance on Nvidia while also developing its own proprietary chips and working with rivals such as AMD.

There have also been reports suggesting Meta has looked into utilising TPUs — chips crafted by Google, a competing company.

The Nvidia partnership might reduce speculation regarding Meta's rumored TPU discussions, according to Patrick Moorhead, chief analyst at Moor Insights & Strategy — though, technology giants often evaluate several suppliers simultaneously.

The agreement comes at a time of heightened competition in the AI infrastructure landscape. 

Although Nvidia is leading the sector, competitors like Google, AMD, and Broadcom are challenging its supremacy.

Importantly, the collaboration will involve Meta deploying not just Nvidia's GPUs but also CPUs.

CPUs, which have been long dominated by Intel and AMD, function as the central processors that collaborate with GPUs within data centres. 

They're utilised for general computing and form the heart of nearly all contemporary computing setups, while GPUs handle specialised, power-intensive tasks like AI training and gaming graphics. 

By supplying both, Nvidia could capture even more of Meta's investment and solidify its position in Meta's AI setup.

Though this heightens competitive dynamics, Moorhead noted that the demand for infrastructure has grown so extensively that Nvidia's competitors are unlikely to experience immediate decreases.