The future of computing is blasting off, and Europe needs to catch the rocket! The European Space Policy Institute (ESPI) is sounding the alarm: if Europe doesn't quickly develop a comprehensive strategy for space-based data centers, it risks being left behind in a critical area of digital infrastructure. But why is this so important, and what's the rush? Let's dive in.
The race to the stars for data processing is heating up, with the United States and China leading the charge. Tech giants like SpaceX, under the visionary Elon Musk, are exploring the vast potential of in-orbit cloud computing. This is driven by the soaring processing and energy demands of artificial intelligence.
According to an ESPI report released on November 18th, a significant €70 million ($81 million) of private capital has already poured into space-based data center ventures and related technologies over the past five years. This includes exciting projects like Florida-based Lonestar, which focuses on lunar data storage, and Starcloud, a company based in Washington that recently launched its first small satellite.
But here's where it gets controversial... Starcloud-1, weighing around 60 kilograms, carries an Nvidia processor designed to run AI models in orbit, including versions of Google’s Gemini. However, to truly scale up to commercially viable levels, these ventures will require massive infrastructure: kilometers of solar arrays and colossal radiators to manage the immense heat generated in space.
This sheer scale presents one of the biggest engineering hurdles. Despite considerable advancements, challenges remain in launch affordability, thermal management, and in-orbit assembly. While Europe has made initial strides with projects like ESA’s PhiSat AI processing missions and feasibility studies funded by the European Union, others are moving rapidly toward operational systems.
NASA, for instance, is managing over two dozen AI-driven edge computing projects, including experiments on the International Space Station. These projects focus on processing sensor readings in space rather than sending raw data back to Earth. China is also making significant moves, having launched the first 12 satellites of its planned Three-Body Computing Constellation – a massive 2,800-satellite AI computing network designed to perform heavy processing in space, reducing latency and data bottlenecks.
Powering the AI Revolution: McKinsey analysts predict that as much as $6.7 trillion in investment could be needed for data centers by 2030, with approximately $5.2 trillion driven by AI workloads.
"There is an increasing need in orbit for storage and processing in space for in-situ operations,” says Claude Rousseau, research director at Analysys Mason. He adds that space-based data centers are a “natural fit” to meet the growing needs for operational sovereignty, increased AI use, and enhanced data security.
Elon Musk, who also leads the advanced-model startup xAI, has stated that future generations of Starlink broadband satellites, launching on SpaceX’s Starship, could be scaled up to operate as orbiting data centers, linked via high-speed lasers. Amazon and Blue Origin founder Jeff Bezos has predicted gigawatt-scale data centers in space within the next 10 to 20 years. Amazon, which owns cloud computing giant Amazon Web Services, is also deploying a low Earth orbit broadband constellation called Amazon Leo to compete with Starlink.
ESPI warns that Europe risks becoming overly reliant on foreign orbital compute capacity if it doesn't take decisive action. To avoid falling behind, the think tank recommends the European Union take the following steps:
- Launch a European Space-Based Data Centre initiative as part of the 2028–2034 Horizon Europe Moonshot Projects.
- Utilize the European Space Agency’s General Support Technology Programme (GSTP) and its Advanced Research in Telecommunications Systems (ARTES) program as public-private testbeds.
- Establish a phased roadmap extending beyond R&D toward commercial orbital compute deployment.
What do you think? Is Europe's strategy the right one? Are there other approaches that could be more effective? Share your thoughts in the comments below!
