Tesla Dojo supercomputing debut, Musk: The computing power of AI training by the end of the year will be approximately equal to 8,000 NVIDIA H100 GPUs

According to news from this site on July 24, Tesla CEO Elon Musk said in an earnings call today that the company is about to complete the largest artificial intelligence training cluster to date, which will Equipped with 20,000 NVIDIA H100 GPUs.

Musk also told investors on the company's earnings call that Tesla will work hard to develop its Dojo supercomputer because GPUs from Nvidia are expensive.

This site’s translation of part of Musk’s speech is as follows:

The road to competing with NVIDIA through Dojo is difficult, but I think we have no choice, we are now too dependent on NVIDIA.

From Nvidia’s perspective, they will inevitably raise the price of GPU to a level that the market can bear, but the price is too high, so we pin our hopes on the Dojo supercomputer.

Musk said in a tweet today: "By the end of this year, Dojo 1 will launch a training program with a computing power equivalent to 8,000 Nvidia H100s. Although this scale is not large, it is not small."

The above is the detailed content of Tesla Dojo supercomputing debut, Musk: The computing power of AI training by the end of the year will be approximately equal to 8,000 NVIDIA H100 GPUs. For more information, please follow other related articles on the PHP Chinese website!