Energy constraints are forcing technology companies to seek new paths. Interest in artificial intelligence has intensified, and discussions about the Moon and space are becoming increasingly relevant.
Global magnates like Jeff Bezos are expressing serious ideas about building space data centers that could efficiently process AI models, providing energy supply at a new level.
Bezos recently noted, "The Moon is a gift from the Universe," emphasizing its potential as a base for space projects. While companies like Blue Origin and SpaceX are working to reduce the cost of space travel, it remains unclear which will emerge faster: lunar bases or super-advanced AI. These ideas seem to be intertwining, raising concerns about a potential bubble around AI.
At present, creating space data centers is economically unfeasible, but according to forecasts by Professor Phil Metzger from the University of Central Florida, this situation may change in the future, possibly within ten years.
“Space has always attracted enthusiasts looking for economic justifications for the exploration of other planets,” he shared on X, commenting on the growing interest in space data centers. “AI servers in space represent the first real basis for such steps, opening new horizons.”
Meanwhile, on Earth, President Trump has already declared a state of emergency in energy, stating the need to increase capacity to meet the growing demand for electricity in the coming years.
Companies developing AI are actively enhancing their energy resources while awaiting improvements in the power supply system. For example, Elon Musk's xAI uses gas turbines as temporary energy sources, while OpenAI is calling for collaboration with the government to increase capacity by 100 gigawatts annually.
Let's consider 100 gigawatts in context. Forty years ago, the film "Back to the Future" became iconic due to the need to generate such colossal energy for time travel to seem plausible. The main character sought 1.21 gigawatts of energy, equivalent to a lightning strike, to return home in a DeLorean time machine.
Now, 1 gigawatt, which once equated to half the power of the Hoover Dam, seems insignificant. Names like Musk, Bezos, and Sundar Pichai are uttering words that sound like science fiction to the younger generation.
The essence of their argument is that the needs for AI will soon reach such scales that humanity will be forced to seek solutions in space, where solar energy can be utilized more efficiently.
In space, sunlight can be received directly and continuously, allowing solar panels to collect energy without interference from clouds or rain. Additionally, the vacuum of space reduces the need for cooling equipment.
Moreover, many limitations faced by companies on Earth when building new power plants for data centers do not exist in space.
“In the next two decades, we will be able to surpass the cost of ground data centers,” Bezos claimed at the latest technology conference. “Space will become a place where technologies will continue to develop Earth.”
Although it may be too early to talk about it, Google has announced the Suncatcher project, aimed at scaling machine learning in space. Plans are in place to launch two satellite prototypes by 2027 to test equipment in orbit.
“Like any project, this will require solving many complex engineering challenges,” Pichai noted on social media. Nvidia has also begun collaborating with the startup Starcloud to develop space data centers.
Musk, eager not to fall behind, presented his updated concept. He has long targeted Mars as the future of SpaceX, but recently he has increasingly spoken about the possibility of using his spacecraft to launch new versions of Starlink satellites equipped with solar panels and high-speed lasers to create space data centers.
At a recent event, Musk reiterated that such AI satellites could produce 100 gigawatts of solar energy per year, which is about a quarter of the average annual consumption of the United States. “We have a plan for how to implement this,” he told investor Ron Baron. “It’s truly incredible.”
Musk previously noted that he would need four to five years to achieve this goal, and he also shared ideas that 100 terawatts per year could be generated on a lunar base producing AI satellites powered by solar panels and launching them into orbit using a catapult.
According to his assumptions, the lunar base will produce satellites that will be launched into orbit using a catapult, and the solar panels of these satellites could generate 100,000 gigawatts of energy per year.
“I am confident that we will see further development of AI, where a large portion of solar energy will be used for computations,” he stated at a technical conference in September.
Despite the seemingly fantastical nature of these ideas, a few years ago, Musk also faced ridicule when urging energy companies to expand their capacities, warning of a potential shortage due to the growing demand for AI, electric vehicles, and other technologies. Now, the answer to this challenge is being sought in space.
As the famous phrase from "Back to the Future" goes: if you put in the effort, you can achieve anything. Perhaps.
