Thursday, July 11, 2024

xAI supercomputer w/o Oracle cloud

Elon Musk on X: "@xDaily xAI contracted for 24k H100s from Oracle and Grok 2 trained on those...

"xAI contracted for 24k H100s from Oracle and Grok 2 trained on those. Grok 2 is going through finetuning and bug fixes. Probably ready to release next month. 

xAI is building the 100k H100 system itself for fastest time to completion. Aiming to begin training later this month. It will be the most powerful training cluster in the world by a large margin. 

The reason we decided to do the 100k H100 and next major system internally was that our fundamental competitiveness depends on being faster than any other AI company. This is the only way to catch up. 

Oracle is a great company and there is another company that shows promise also involved in that OpenAI GB200 cluster, but, when our fate depends on being the fastest by far, we must have our own hands on the steering wheel, rather than be a backseat driver."


Musk xAI Ditches Oracle Cloud to Build Massive GPU Cluster for Grok 3

xAI already rents around 16,000 Nvidia GPUs from Oracle, making it one of the largest customers of the cloud service.

xAI plans to build "the world’s most powerful supercomputer" in Memphis, Tennessee. Musk said that he expects the supercomputer to open by the fall of 2025.



No comments: