ALBAWABA - Billionaire tech tycoon Elon Musk has informed investors that he intends to construct a supercomputer described as a "gigafactory of compute," according to a report by the Economic Times citing industry outlets, to aid in the ongoing expansion of his artificial intelligence business, xAI, developer of AI chatbot Grok.
The supercomputer will reportedly run by the power of over 100,00 Nvidia chips, with Musk looking forward to a start of operations by fall of 2025, saying he will hold himself personally responsible for delivering it on time,” mentioning that xAI and Oracle could collaborate to build such a tremendous computer.
When finished, the interconnected chip clusters of Nvidia's flagship H100 graphics processing units (GPUs), are going to be at least fourfold bigger than the largest GPU arrays currently in use, Reuters notes citing the Information quoting Musk from an investors meeting earlier this month.
xAI is currently developing Artificial Intelligence chatbot Grok that trains on social media posts on Musk’s own X platform (formerly Twitter), was launched as a competitor of popular ChatGPT by OpenAI, which Musk co-founded but quit in 2018 due to the company’s profit-focused business practices by his description.
Musk stated earlier this year that it required approximately 20,000 Nvidia H100 GPUs to train the Grok 2 model. These GPUs are among the most powerful in the data center processor space for artificial intelligence, but are in short supply do to huge demand, according to Reuters, he added that 100,000 Nvidia H100 chips would be needed for the Grok 3 model and beyond.
