Matt Garman, chief executive officer (CEO) of Amazon Web Services (AWS), said that a space data center touted by Tesla CEO Elon Musk as the future of AI is "realistically still a long way off." For now, the economic and technological hurdles outweigh the benefits, he argued.
Garman said at the Cisco artificial intelligence (AI) Summit in San Francisco on the 3rd, "There still aren't enough rockets to launch a million satellites," adding, "When you think about the expense of sending payloads into space today, it's not economical."
Garman noted that while space data centers are conceptually intriguing, they are an unworkable option in the near future due to the logistical challenges of maintaining high-performance hardware in the harsh environment of space and the enormous expense of launching and maintaining such infrastructure.
Recently, the concept of building data centers in space has drawn attention. The logic is that orbital facilities can bypass constraints such as land use and cooling on the ground. Furthermore, the argument goes that space-based servers could offer a way to generate the massive AI computing power needed for artificial general intelligence (AGI) in a cost-efficient manner. Musk is leading this outlook.
Just a day earlier, Musk cited "building space data centers" as the backdrop for his decision to merge his space company SpaceX with his artificial intelligence (AI) company xAI. Musk said that unlike ground data centers, which require massive power and cooling facilities, space can use nearly limitless solar energy as a power source and eliminates concerns about cooling.
Garman drew attention in particular for expressing skepticism about space data centers even though Amazon founder Jeff Bezos is investing heavily in the space race through Blue Origin.
Garman's remarks suggest that AWS will continue to focus on expanding ground data centers. But there are challenges to this strategy as well. AWS is currently facing the risk of delays in data center expansion due to power grid delays in Europe.