GoKawiilxAI plans to add 660,000 GPUs by December, reaching 1.44 million total
Elon Musk said xAI will bring 220,000 Nvidia GB300 GPUs online next week, another 220,000 in November, and potentially 220,000 more by late December. Combined with existing hardware at Colossus 1 and Colossus 2, this would give xAI about 1.1 million GB300 GPUs and 1.44 million GPUs in total. Musk also noted that Colossus 1, an inefficient mix of Hopper and Blackwell chips for training Grok, has been rented out to Anthropic for inference.
GoKawiil's interpretation of the reporting above, not reported fact.
Reaching 1.44 million GPUs would put xAI's compute scale ahead of far older rivals like OpenAI and Anthropic, suggesting Musk is betting heavily on raw hardware capacity to compete despite the company's short history. The rapid buildout also highlights power supply as a growing bottleneck, since xAI has resorted to running its own gas turbines and power plant, a move that has already drawn a lawsuit over unpermitted equipment. How quickly xAI can resolve these power and permitting issues could determine whether this GPU expansion translates into usable training capacity on schedule.
- xAI expects to add 660,000 GB300 GPUs by the end of 2025, split across three rollout phases.
- Total GPU count across Colossus 1 and Colossus 2 could reach 1.44 million units, including 1.1 million GB300 chips.
- Power constraints have pushed xAI to run its own gas turbines, prompting a lawsuit over unpermitted equipment near its data center.
Source: tomshardware.com — Jowi Morales, 2026-09-25
Published there as: “Elon Musk's SpaceXAI to add another 660,000 AI GPUs this year, nearing a total of 1.44 million in operation”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.