Bezos and Musk are competing to establish data centers in space, aiming to handle AI computing without the constraints faced on Earth. While the concept is intriguing, experts point out significant engineering challenges and cost concerns that might hinder its feasibility.

The article critiques the push for data centers in space, arguing that the immense costs and logistical challenges outweigh the benefits. It highlights the growing risks of satellite congestion and the competitive edge of ground-based energy sources, suggesting that such ventures are driven more by hype than feasibility.

Starcloud, backed by Nvidia, has successfully trained an AI model in space using its Starcloud-1 satellite equipped with a powerful H100 GPU. This milestone demonstrates the potential for orbital data centers to operate complex AI models while reducing energy costs and environmental impact compared to Earth-based facilities.

Google is developing Project Suncatcher, a plan to deploy AI data centers in space using solar-powered satellites. This approach could provide a solution to high energy costs on Earth by utilizing more efficient solar panels in orbit and creating a distributed network of computing resources. Challenges include maintaining high-speed communication between satellites.