Elon Musk Outlines Ambitious Plans for xAI, Aims to Run 50 Million H100 Chips in Five Years

Elon Musk plans for xAI to use 50 million H100 chips within the next five years, aiming to surpass Nvidia's capacity. xAI currently runs around 200k H100s in its Colossus data center, while Meta is set to reach 600k H100 capacity. Both companies are developing custom chips to reduce reliance on Nvidia and increase efficiency. Musk's strategy aims to outmaneuver opponents, but Meta's investment in hundreds of billions of dollars could give it an advantage. Capacity alone may not define the winner, as innovative use cases and partnerships will also be key.

Elon Musk has outlined ambitious plans for his artificial intelligence (AI) company, xAI, aiming to leverage 50 million H100 chips within the next five years. This aggressive expansion strategy is designed to surpass Nvidia's current AI compute capacity, which is a critical element in the AI race [2].

xAI currently operates around 200,000 H100 chips within its Colossus data center in Memphis. Meta, a significant competitor, is projected to reach 600,000 H100 chips, indicating a substantial lead in AI compute capacity [2]. However, Musk's plans suggest that xAI is not merely playing catch-up but aiming to outmaneuver its rivals.

Both xAI and Meta are developing custom chips to reduce their reliance on Nvidia and enhance efficiency. This strategy is part of a broader trend among major AI companies to create their own hardware solutions. Musk's strategy appears to be focused on leveraging the Grok chatbot and Colossus supercomputer to develop AI-driven software and video games, challenging Microsoft's dominance in the software market [1].

Musk's plans to open-source all of xAI's models, including Grok 2.5, are another significant move. This transparency aims to build trust and ensure that the company's AI processes are scrutinized by the broader community. However, the impact of this open-sourcing on xAI's competitive edge remains to be seen [2].

Nvidia's earnings report, scheduled for August 27, will provide further insight into the company's performance and strategy in the AI hardware market. Analysts are anticipating earnings of $1.01 per share, with revenue forecast to reach $46.0 billion, reflecting the growing demand for AI technologies [3].

While capacity is a crucial factor in the AI race, innovative use cases and strategic partnerships will also play a significant role in determining the winners. Musk's strategy appears to be focused on both increasing AI compute capacity and developing cutting-edge AI applications.

References:
[1] https://timesofindia.indiatimes.com/technology/tech-news/elon-musk-calls-upon-engineers-to-join-xai-and-take-on-microsoft-says-microsoft-does-not-/articleshow/123471356.cms
[2] https://www.socialmediatoday.com/news/elon-musk-outlines-plan-to-power-up-xai-h100-capacity/758458/
[3] https://www.kiplinger.com/investing/live/nvidia-earnings-live-updates-and-commentary-august-2025