Skip to content

AI Leaders Paint Bold Future: 10 Billion GPUs Needed

AI's future demands 10 billion GPUs. Companies are investing, but is it enough?

This is an article and here we can see planets, a machine and some text.
This is an article and here we can see planets, a machine and some text.

AI Leaders Paint Bold Future: 10 Billion GPUs Needed

Leading AI figures, including OpenAI's Sam Altman and Greg Brockman, along with Nvidia's Jensen Huang, gathered for an interview, painting a bold vision of the future. They discussed the current gap in AI computing power, the potential of AI as a 'superbrain' in daily life, and the need for vast computational resources to realize this future.

OpenAI President Greg Brockman believes the industry is 'three orders of magnitude off' in terms of AI computing power. He estimates that 10 billion GPUs or equivalent would be needed for the envisioned future. This stark contrast is highlighted by Microsoft's former CEO Bill Gates, who once envisioned a computer on every desk, a reality now, but far from the scale Brockman envisions.

Brockman suggests the world is heading towards 'compute scarcity' and an economy powered by compute. He implies that AI datacenter service could become a de facto currency in the future. To this end, no company explicitly plans to finance the production of 10 billion GPUs, but IREN is significantly expanding its AI cloud with tens of thousands of GPUs and $674 million funding. Mistral AI, with partners, plans to deploy about 18,000 Nvidia GPUs as part of an $8.5 billion project, supported by EU investment in AI chip fab facilities.

OpenAI CEO Sam Altman compares the scale of OpenAI's partnership with Nvidia to the Apollo program. Brockman suggests every person should have their own dedicated GPU for this purpose. He envisions AI working proactively for individuals while they sleep.

The interview underscores the vast computational resources needed for the future of AI, with estimates of billions of GPUs required. While companies are investing heavily in AI cloud expansion, the scale envisioned by Brockman remains a significant challenge. The future of AI, as a 'superbrain' embedded in daily life, hinges on overcoming this 'compute scarcity'.

Read also:

Latest