Elon Musk Goes All In To Bring A Whopping 1.2 Million NVIDIA’s Latest GPUs Online In His Mega Data Center To Take On OpenAI
Elon Musk's xAI plans to deploy 1.2 million NVIDIA GPUs in its Colossus data center project, aiming to compete with OpenAI and Anthropic. Colossus 2, the second phase, will primarily use NVIDIA's Blackwell GB300 GPUs, with 660,000 units expected by 2026. The data center currently operates 110,000 GB200 GPUs and 440,000 GB300 GPUs, with more to be added by year-end.
How this was made

The 30-second read
Why it matters
If realized, the massive compute capacity could accelerate xAI's model development, potentially affecting competitive dynamics with OpenAI and Anthropic.
Market read
While no immediate trading action is evident, the scale of the compute buildout may shape sentiment around AI‑related equities.
What to watch
Capital expenditure financing and power‑consumption costs are not addressed.
Background
Elon Musk's xAI (referred to as SpaceXAI) announced plans to bring hundreds of thousands of NVIDIA Blackwell GPUs online at its Colossus data center in Memphis.
Market effects
Potential boost to AI‑hardware demand and related semiconductor stocks.
Limited to U.S. AI and cloud infrastructure markets.
May influence global AI compute capacity outlook.
Counterpoint
The announced GPU rollout may be overstated; supply constraints could delay full deployment.
Key entities
- personElon Musk
CEO of xAI, providing the primary quote.
- companyNVIDIA
Supplier of the Blackwell GB300 GPUs referenced in the plan.




