Ai, Elon Musk's xAI startup raises 6 billion and works on a supercomputer for Grok

The US News Agency The Information published a report stating that xAIthe startup founded last year by Elon Musk which is in direct competition with OpenAI And Alphabet (the holding company that controls Google), is planning a new supercomputer to train with Grok, the proprietary paid chatbot on X equipped with generative artificial intelligence. For this operation, Elon Musk would have collected 6 billion dollars extra thanks to a fundraiser from Silicon Valley investors.

The new supercomputer, nicknamed “Gigafactory of Compute” by Musk, apparently could be created in collaboration with Oracle and be operational in just over a year, inautumn of 2025. At the moment neither xAI nor Oracle have commented on the report in question, which was drawn up by The Information on the basis of some statements that Musk himself made during a presentation held with investors this month.

What the new xAI supercomputer will be like

According to rumors, xAI's new supercomputer will be extremely powerful. In the aforementioned presentation, Musk revealed that this will use something like 100,000 Nvidia H100 GPUs architecture-based Hopperwhich would make it at least 4 times larger than competitor GPU clusters, like those used by Meta (the company that, among other things, develops Facebook and WhatsApp) to train its models. A decidedly larger number than the ones 20,000 GPUs used to train Grok 2.0which gives us an idea of ​​how ambitious Musk's project is.

The purchase of so many GPUs will naturally require a huge economic outlay on the part of xAI which, as announced in a post on its official blog, has recently raised $6 billion in investmentsthanks to “actors” of the caliber of Andreessen Horowitz, Sequoia Capital and Kingdom Holding.

Doubts about the timing

According to what was reported by The InformationElon Musk considers himself personally responsible of the creation of the supercomputer by deadline expected, to indicate how important he considers the success of the project to be. It is good to keep in mind, however, that over the years Musk has accustomed us to sensational announcements which, at times, have failed to meet expectations regarding the timing of implementation. So we will still have to wait to understand if the project will actually be completed on time and if it will lead to the promised results.