Elon Musk has confirmed that Tesla and his AI startup, xAI, will continue to purchase GPUs from Nvidia and AMD to support their artificial intelligence initiatives. Musk emphasized that as long as these external suppliers offer superior technology, Tesla will persist in acquiring their products. This commitment comes as xAI plans to construct a facility near Memphis capable of housing 1 million GPUs, building upon the 200,000 already deployed in the region. Tesla, in turn, requires a significant number of GPUs for training its Optimus humanoid robots.

Earlier this year, Musk indicated that Tesla expects to spend between $3 billion and $4 billion on Nvidia hardware for AI training in 2024. He noted that for building AI training superclusters, Nvidia hardware accounts for around two-thirds of the cost.

Musk also confirmed plans to purchase chips from AMD, specifically the Radeon Instinct MI300X, which is designed to scale and supports up to 192 GB of HBM3 memory. This chip is comparable to Nvidia’s H100 in training tasks and is 1.6 times faster in AI inference scenarios.

In addition to these developments, xAI is investing over $400 million in building a supercomputer in Memphis, although local power grid limitations may pose challenges.

Musk’s statements underscore Tesla’s commitment to advancing its AI capabilit

ies through significant investments in GPU technology from leading suppliers.