Musk's xAI Plans to Substantially Increase Nvidia AI Chip Count by Year-End

Elon Musk revealed that his artificial intelligence venture, xAI, is set to substantially increase the computing power of its Colossus 2 data center, with plans to roughly double its count of advanced Nvidia AI chips by the end of the year. The expansion underscores the intense and growing demand for high-performance hardware required to train and operate large-scale AI models.

The Expansion Plan

In a post on the social media platform X, Musk outlined the aggressive timeline for the ramp-up at the Memphis, Tennessee-based facility, which he has touted as the “world’s largest AI supercomputer.” According to the post, the center currently uses 110,000 of Nvidia’s GB200 chips and 440,000 of its GB300 chips.

The plan details a multi-stage increase in its Nvidia hardware:

• An additional 220,000 GB300 chips are scheduled to come online by next week.
• Another 220,000 are slated for installation in November.
• A further 220,000 GB300 chips could be added by late December, with Musk noting, “If we get lucky.”

Ad

Context and Implications

The Colossus supercomputer complex was initially built to train Grok, xAI's flagship AI model. However, the venture has since expanded its scope, renting out computing capacity to other major players in the AI field, including Anthropic and Alphabet’s Google, according to the source report.

This rapid scaling highlights the fierce competition among technology firms to secure the cutting-edge processors essential for AI development. The announcement did not specify whether the Colossus 2 facility has sufficient power and physical capacity readily available to support the massive influx of new hardware, a critical logistical challenge in large-scale data center expansions.