In the next year or so, a pair of satellites linked to Google’s Project Suncatcher will take to orbit to assess the potential of large-scale data center clusters in space. Similarly, AWS is working on a new ground station designed for satellite data to be integrated directly into cloud environments. And at the European Space Agency, Project OFELIAS is well underway, a partnership with the German Aerospace Center exploring new protocols, processes, and algorithms to make data exchange between Earth and space via laser links more stable and efficient. This is 2026, and while the above may sound like something from a sci-fi novel, it’s happening now, as you read this article.
Space is full of untapped potential. It offers limitless solar energy, few physical constraints, and a vantage point that could reshape how data is processed and distributed around the world. As industries continue to scale AI across everything from logistics and finance to connected cars and remote healthcare, connectivity is once again a spotlight issue. But this time the question is less binary – it’s not just a case of coverage or bandwidth but where compute is located and how quickly it can be accessed. Latency can make or break real-time AI-powered applications, which is why data center construction is now booming and “edge hubs” are becoming increasingly common. No matter how powerful AI models become, the time it takes for data to get from A to B and back again will the ultimate bottleneck.
This subject was front and center at this year’s SatShow 2026 in Washington DC. AI may be advancing at remarkable speed, but the infrastructure supporting it is under increasing strain. Gartner predicts that 40% of enterprise applications will feature task-specific AI agents by the end of 2026, rising from less than 5% the previous year, with more complex, collaborative systems already on the horizon. At the same time, research from IBM shows that two-thirds of organizations are already seeing meaningful product gains – and that momentum is unlikely to slow. The direction is clear, but so is the pressure it creates. AI doesn’t operate in isolation – it depends on how efficiently data moves between systems, across networks, and increasingly, between Earth and orbit.
When geography becomes a performance constraint
Once AI is pushed into real-world environments, distance dominates. Every interaction relies on data moving between devices, applications, and compute environments, often across hundreds or thousands of kilometers. That journey happens in the blink of an eye, but even then, the round trip introduces delay, and in time-sensitive scenarios, those milliseconds add up quickly. An autonomous vehicle responding to changing road conditions, a robotic system adjusting in a live production environment, or a security platform identifying and reacting to a threat all depend on near-instant feedback loops. When data has to travel too far before it can be processed and returned, responsiveness drops and reliability begins to suffer. The closer data, compute, and applications are to one another, the more effectively AI systems can operate, which is also what is driving interest in architectures that blend terrestrial networks with non-terrestrial ones – but even then, latency is factor.
The foundation for Earth-space interconnectivity
Moving data via optical feeder links, which use laser communication between satellites and ground stations, will undoubtedly be part of the solution, but that introduces a new set of variables that don’t exist in traditional terrestrial networks. Cloud cover alone can interrupt these links, creating fluctuations in availability, delay, and jitter that are difficult to predict. For AI-driven applications that rely on steady, real-time data exchange, that inconsistency simply isn’t tolerable. Projects like the ESA’s OFELIAS are focused on developing algorithms and protocols that allow data to be rerouted dynamically between multiple optical ground stations whenever a link is disrupted. In practical terms, that means building a system where satellite constellations, ground stations, and network control layers work together to maintain continuity, even when environmental conditions are less than perfect.
OFELIAS is part of a broader effort to make space-based connectivity more dependable. For instance, optical inter-satellite links are already being deployed to allow data to travel between satellites before being sent to Earth, reducing reliance on any single ground station. SpaceX has implemented this approach within its Starlink network using laser links between satellites, enabling data to be routed in orbit rather than immediately downlinked. Meanwhile, the European Space Agency’s HydRON initiative is exploring how to create a high-capacity optical network in space, effectively extending connectivity with fiber-like performance into orbit. These developments don’t eliminate the challenges of operating between Earth and space, but they show how the industry is working to manage them, building a more stable and flexible foundation for the next generation of connected systems.
As these terrestrial and orbital networks become more tightly interwoven, the role of interconnection points will only grow in importance, quietly shaping how data is exchanged, optimized, and delivered at scale. Whether on the ground or reaching into orbit, it’s this underlying foundation of connectivity that will determine how far and how fast AI can truly go.
