Musk Charges into AI Computing Power! Colossus 2 Expansion Progress Revealed: Chip Count to Surpass 1.2 Million by Year-End

Elon Musk (SPCX-US), CEO of SpaceX, said in the early hours of Friday (25th) that his company xAI plans to dramatically increase the number of advanced NVIDIA (NVDA-US) chips used at Colossus 2, an artificial intelligence (AI) computing center under construction in Memphis, USA, before the end of this year.

On Friday, Musk released the most detailed timeline yet for the expansion of Colossus 2. According to his posts on the X platform, Colossus 2 is currently equipped with 110,000 NVIDIA GB200 chips and 440,000 GB300 chips, totaling 550,000 GPUs.

He stated that an additional 220,000 GB300 chips are scheduled to come online next week, with another 220,000 to be added in November. If progress continues smoothly, a further 220,000 GB300 chips could be deployed by the end of December, bringing the total to 1.21 million.

Overall, if the November chip additions go live as planned, the number of chips in Colossus 2 will increase by approximately 80% compared to current levels. If the final batch of 220,000 chips is successfully operational before year-end, the total will reach about 2.2 times the current capacity.

Musk described Colossus as the "world's largest AI supercomputer." Colossus 1 was initially built in 2024 to train xAI’s flagship AI model Grok, but the facility has since been continuously expanded and has begun providing computing power for other AI projects. Earlier this year, xAI also began leasing computing capacity from this facility to Anthropic and Google (GOOGL-US), a subsidiary of Alphabet.

Previously, xAI stated its goal to equip the Memphis facility with one million GPUs by the end of 2026. Musk did not clarify whether Colossus 2 currently has sufficient ready computing capacity and power infrastructure to support the planned chip additions.