Is Jeff Bezos Right? Is Space the Next Frontier for AI?

Considering all he’s achieved, why wouldn’t Jeff Bezos be looking at space as the next big opportunity?
The Amazon founder and Executive Chair has laid out a bold 10 to 20-year timeline for building gigawatt-scale data centres outside the Earth’s boundaries, claiming they could eventually exceed the capacity and performance of terrestrial equivalents.
The plan is aimed at meeting the ever increasing demand for AI infrastructure that’s accelerating global energy use.
Speaking at Italian Tech Week in Turin, Bezos discussed his space-based vision in a conversation with John Elkann, Chair of both Ferrari and Stellantis, outlining a scenario in which orbital data centres become the next major step in AI infrastructure.
"One of the things that’s going to happen next – is we’re going to start building these giant gigawatt data centres in space," he said.
The scale of these kinds of developments is vast – a gigawatt equals one billion watts, comparable to the output of a large nuclear power plant.
As organisations invest in larger AI models and expand cloud computing networks, confronting the rising energy demands these systems create is unavoidable.
Data centres, which house servers that store and process digital data, already consume large amounts of electricity and water for cooling. These resource demands are prompting global concerns about long-term sustainability and grid pressure.
Solar power beyond weather or night
Bezos identified round-the-clock solar energy as the main reason why space offers a viable alternative.
He explained that space-based solar panels can bypass the limitations that terrestrial systems face, such as interruptions from cloud cover or night-time hours.
"These giant training clusters, those will be better built in space, because we have solar power there, 24/7. There are no clouds and no rain, no weather," he said.
- Space offers uninterrupted solar power with no weather or night interruptions
- Orbital solar energy can outperform Earth-based renewable and grid power
- Space data centres could reduce costs versus terrestrial facilities over time
- AI training clusters need continuous, high-power energy only space can provide
- Existing satellite systems prove space infrastructure can reliably support Earth-based operations
- Moving data centres off-planet avoids Earth’s growing electricity and water constraints
Training clusters refer to large groups of interconnected computing systems that train AI models. These require intensive computing resources and large energy supplies. Continuous solar power availability, without the variability of weather or daylight, could improve efficiency and reduce operating costs.
According to Bezos, the long-term result is that space-based operations could eventually undercut terrestrial data centres in cost.
"We will be able to beat the cost of terrestrial data centres in space in the next couple of decades," he said.
A pattern in space-based infrastructure
The shift towards orbital data centres, Bezos argued, fits into a longer trend in which space-based systems begin to support life and business on Earth.
He referred to how existing satellite infrastructure already provides foundational services. Weather satellites deliver the data required for accurate forecasting. Communication satellites enable internet access and telecommunications across the globe.
"It already has happened with weather satellites. It has already happened with communication satellites. The next step is going to be data centres and then other kinds of manufacturing," Bezos explained.
He said this as a natural progression, with data centres and potentially manufacturing following similar deployment paths.
Engineering challenges and launch costs
A transition of this scale isn’t without its hurdles, though. Bezos acknowledged that launching and maintaining space-based infrastructure presents logistical and financial obstacles.
Maintenance requires an entirely different model compared to land-based systems, where physical access is relatively straightforward.
If hardware upgrades or repairs are needed, new components must be launched into orbit, introducing higher costs and risks.
Despite lower launch costs made possible by reusable rocket technologies from SpaceX and Bezos’ own Blue Origin, failed launches can result in the total loss of high-value equipment.
Nonetheless, he believes that the timeline remains within reach.
"It’s hard to know exactly when, it’s 10 plus years – and I bet it’s not more than 20 years," Bezos said.


