One of the biggest questions in the tech sector has been if AI is the future, where does all the infrastructure go?<\/p>\n
Some space industry leaders \u2014 including Elon Musk and Jeff Bezos \u2014 believe the answer could be in orbit, via space-based data centers.<\/p>\n
That idea has gained attention in recent weeks. Musk is reportedly considering a merger<\/a> between SpaceX and his generative artificial AI company xAI ahead of his space firm\u2019s planned IPO later this year. In addition, in a\u00a0filing<\/a>\u00a0with the FCC late Jan. 30, SpaceX proposed an orbital data center constellation<\/a> of up to 1 million satellites in low Earth orbit.\u00a0<\/p>\n But industry analysts say that while the technological challenges can be surmounted, it\u2019s not yet clear if the business case for data centers in space holds up.\u00a0<\/p>\n Deep learning models built by frontier labs like OpenAI, Anthropic and xAI all depend on massive clusters of advanced processors operating in data centers. As more deep learning models are built and deployed, more processing power is required \u2014 last year companies invested $61 billion in building more data centers. That\u2019s a new record, according to S&P Global, with real estate trackers expecting capacity to double by 2030.<\/p>\n However, there are obstacles facing new data centers on Earth, including environmental permitting, access to electricity for power and water for cooling and community opposition. Microsoft, for example, expects its data center water usage to triple by 2030, according to the New York Times<\/a>. Data center projects built by Amazon and Chile have been canceled over resource concerns like those.\u00a0<\/p>\n \u201cThe biggest problems are cooling, security, power transmission \u2014 all those things can be solved if you just move it into space,\u201d Chris Quilty, a space industry analyst, told SpaceNews. \u201cThe big question that has to be solved, can you do it economically?\u201d<\/p>\n Jeff Bezos, long an advocate of moving polluting industries off-world as CEO of Blue Origin, kick-started this discussion with remarks at last year\u2019s International Astronautical Congress.\u00a0<\/p>\n \u201cWe\u2019re going to start building these giant gigawatt data centers in space,\u201d Bezos said. \u201cWe will be able to beat the cost of terrestrial data centers in space in the next couple of decades.\u201d<\/p>\n Elon Musk and SpaceX are also onboard. Musk told audiences at the World Economic Forum that it will be cheaper to build data centers in space within three years \u2014 although that forecast depends on achieving full reusability for Starship, which Musk expects in 2026.\u00a0<\/p>\n Meanwhile, start-ups like Starcloud are attempting to build companies around this business, while Google is collaborating<\/a> with Earth observation firm Planet to launch its own data center pilot in 2027. Chinese companies like ADA Space and Beijing Astro-Future are plotting ambitious space compute architectures.<\/p>\n \u201cThe rapidity with which the trend has taken hold has shocked me, even for an old hand,\u201d Quilty said. \u201dIt\u2019s very similar to what happened with direct-to-device \u2014 everyone and their brother agreed it was a dumb idea right up until the moment that AST raised a round on the concept.\u201d<\/p>\n Still, much uncertainty remains. \u201cWhether putting data centers in space makes sense or not, and what technologies are needed, depends on the business model,\u201d Akhil Rao, a former NASA economist and managing partner at the frontier technology consultancy Rational Futures, told SpaceNews via email. \u201dI haven\u2019t seen a business model articulated clearly enough to understand how the system architecture serves particular use cases well.\u201d<\/p>\n