Sam Altman envisions OpenAI running 100 million GPUs

OpenAI CEO Sam Altman has sparked industry-wide discussion after hinting at a long-term vision to operate 100 million GPUs.

While the remark might have been delivered lightheartedly, it underscores the AI company's growing ambition in the artificial intelligence (AI) space.

The company has already confirmed it expects to run over one million GPUs by the end of 2025.

That figure, unmatched by competitors, places the company at the forefront of AI compute infrastructure.

For comparison, the tech billionaire Elon Musk’s AI venture xAI currently uses about 200,000 GPUs to power its Grok 4 model, just a fifth of what OpenAI plans to deploy by year’s end.

However, scaling from one million to 100 million GPUs presents enormous challenges.

Industry analysts have estimated that such an expansion could cost upwards of $3 trillion and would require breakthroughs in hardware manufacturing, energy efficiency, and data centre logistics.

"Very proud of the team, but now they better get to work figuring out how to 100x that lol," Altman stated in a social media post.

OpenAI primarily relies on Microsoft Azure for its cloud computing needs, but has also partnered with Oracle and is reportedly exploring Google’s TPU chips as alternatives.

This diversification reflects a broader industry shift, with major tech firms like Meta, Amazon, and Google increasingly turning to custom in-house chips to support AI workloads.