About two weeks ago, Elon Musk was one of the bigwigs who signed an open letter (opens in new tab) calling for a pause in AI development. It is rumored that this Elon Musk has purchased 10,000 GPUs. What are these high-performance graphics cards for? [It's a generative AI project known as X.AI Corp, of course (via Ars Technica (opens in new tab)).

Yes, Elon Musk is now holding the very wand he tried to stop and may be shipping his newly acquired GPUs to Twitter's Atlanta data center as we speak. According to a report from our sister site Tom's Hardware (opens in new tab), Musk has recruited former DeepMind employees into the venture to help give birth to a large-scale language model (LLM) comparable to ChatGPT, which is currently in the conceptual stage.

Estimated to be worth tens of millions of dollars, the exact configuration of the hardware Mask is reportedly purchasing is unknown. Most likely, it will include a significant number of Nvidia's Hopper H100 GPUs and/or GPUs similar to the A100 and ChatGPT GPUs.

Having co-founded Open AI (creators of ChatGPT) in 2015 and then left the company on unfavorable terms in 2018, this certainly seems like what Musk would see as a logical business step.

We have investigated the rumors and found that X.AI itself is not necessarily directly associated with Twitter. However, it is likely that a large amount of Tesla data as well as Twitter data will be used to train the anticipated X.AI LLM.

This is because training chatbots on social media (open in new tab) has never gone terribly wrong before.