Elon Musk Makes Massive Nvidia Move at SpaceX

Elon Musk is preparing another massive expansion of AI computing capacity, saying xAI's Colossus 2 cluster could more than double its current number of Nvidia chips by year-end. The timetable underscores just how aggressively Musk is scaling AI infrastructure across his businesses and gives Nvidia investors another sign that demand for its newest accelerators remains enormous.

Colossus 2 currently runs 110,000 Nvidia GB200 chips and 440,000 GB300s, according to Musk.

Another 220,000 GB300s are expected to be fully operational next week, followed by another 220,000 in November.

Musk added that a further 220,000 GB300s could come online by late December if we get lucky.

That would bring Colossus 2 to roughly 1.21 million Nvidia GB200 and GB300 chips if every planned tranche arrives on schedule, more than double the 550,000 chips Musk said are currently installed.

The buildout is especially notable because xAI is already operating Colossus 1 with 150,000 Nvidia H100s, 50,000 H200s and 30,000 GB200s.

Musk's companies are racing to secure enough compute for increasingly demanding AI models, putting xAI among the industry's most aggressive infrastructure builders.

The expansion also matters for Nvidia because the planned additions are concentrated in its latest-generation GB300 accelerators, highlighting continued demand for its most advanced AI hardware.

The SpaceX unit previously said it plans to equip its Memphis facility with 1 million graphics processing units by 2026.

For Nvidia investors, the most important signal is the sheer pace of deployment.

If Musk's timetable holds, hundreds of thousands of additional GB300 chips could be installed within only a few months, reinforcing Nvidia's position at the center of frontier AI infrastructure spending.

For SpaceX and Tesla investors, the broader question is how Musk allocates capital, computing resources and AI development across his companies.

Investors should watch whether the December expansion arrives on schedule, how quickly the additional capacity becomes productive and whether xAI continues scaling beyond the million-chip threshold.

The opportunity is clear: AI demand remains extraordinarily compute-intensive.

The risk is equally straightforward. Infrastructure growth this fast requires enormous capital, power and execution, making utilization and returns increasingly important as the buildout gets larger.