AI is revolutionizing business models, but its success depends on a factor that is often underestimated: digital infrastructure. No matter how powerful the AI, businesses can only scale as fast as network infrastructure will allow. This came through loud and clear at SatShow 2026, where conversations pointed to an industry very much in transition, from the rise of direct-to-device connectivity and multi-orbit architectures to a renewed focus on sovereignty and resilience. Satellites will play a defining role, but proximity between data, users, and data centers on the ground will ultimately determine how fast businesses can move – an article by Ivo Ivanov, CEO of DE-CIX.
According to Gartner, 40% of enterprise applications will feature task-specific AI agents by the end of the year, up from less than 5% in 2025. By 2027, those task-specific AI agents will be replaced by collaborative, end-to-end agents, and by the end of the decade, full AI agent ecosystems spanning multiple joined-up applications will become the norm. It’s little wonder businesses are pursuing agentic AI so determinedly, when already two-thirds of organizations are reporting “significant” product gains, according to a recent IBM survey. The takeaway? Gaining a competitive advantage is no longer solely down to great products, but by building intelligent processes to support, sustain, and enhance those products from the ground up.
Everything came back to this at last week’s SatShow 2026 in Washington DC, where satellite operators, telco providers, equipment vendors, and policymakers attended panels and roundtable discussions to discuss the road ahead. Among the key takeaways: a sharp rise in demand for sovereign space infrastructure, growing momentum behind direct-to-device services, and a shift toward multi-orbit architectures designed to balance coverage, capacity, and latency. At the same time, competition is intensifying as mega-constellations scale and new entrants push into the market, while regulatory frameworks – perhaps predictably – struggle to keep pace with the speed of innovation.
The lower the latency, the greater the success: Shorter distances, better quality
Vehicles navigating traffic autonomously or security systems reacting instantly to threats: In intelligent applications, every millisecond counts. AI only works reliably when all components – from cloud to edge to networks – operate in perfect sync. The challenge: If data is processed too far from where algorithms are needed to steer cars, control robots, or regulate machines, problems arise. In time-critical applications, slow AI responses can lead to inefficiencies and heightened risk.
Geographic distance directly impacts quality of use: data packets travel as light signals between cloud platforms, edge units, and applications. The time this round-trip takes is called latency – you could say this delay is essentially distance measured in milliseconds. The takeaway: The closer AI, data, and applications are to each other, the lower the latency. And the shorter this time, the faster and more reliably intelligent applications, smart services, and AI agents can be used.
This emphasis on proximity was echoed repeatedly at SatShow, particularly in discussions around multi-orbit architectures, where operators are actively combining LEO, MEO, and GEO systems to balance latency and coverage in real-world deployments.
Infrastructure as a location factor: globally connected, locally challenged
What’s merely annoying in video streaming can become a risk in real-time processes. More bandwidth or faster devices rarely solve the issue. The root cause lies deeper: in the structure of many networks, often shaped by the physical locations of digital infrastructure. Despite global connectivity, geographic distance remains the limiting factor. To shorten the gap between servers, storage, data, and AI, infrastructure must move closer to businesses. And data paths across the Internet need better control and optimization. But not every company is located near major digital hubs like Frankfurt or New York, where data centers and cloud providers interconnect via Internet Exchanges (IXs) to serve intelligent business models.
An Internet Exchange (IX) physically connects networks so that Internet service providers (ISPs), cloud providers, content platforms, enterprises, and other operators can exchange data directly. This means data flows don’t have to detour through the public Internet but move straight from the source network to the destination. The principle: to process and transport local data as locally as possible, saving time and reducing latency. This controlled data journey between AI agents, devices and networks enables the speed required for AI to become part of every business success story.
This challenge is becoming more pronounced as governments and enterprises place greater emphasis on digital and space sovereignty, a theme that dominated the exhibition in Washington DC as countries look to exert more control over where data is processed and how it moves across borders.
The Internet between Earth and space: physically connected, seamlessly integrated
As AI growth is testing the limit of our digital infrastructure on earth, the Internet from space is gaining attention and the discussion about digital ecosystems of the future has centered around it. The benefits are clear: Satellites in orbit can serve locations previously out of reach – on land, at sea, or in the air – and space could potentially provide the limitless energy required to train and run AI models. Space based systems are set to become an essential part of our digital backbone alongside terrestrial data centers, edge locations, and high-capacity network hubs.
To bridge the data gap between Earth and orbit, infrastructure must transport information as directly and efficiently as possible. That is why Internet Exchange operators on Earth are stepping up. On the ground, more than 60 DE-CIX locations worldwide are already providing pathways to exchange data in the most efficient way possible. Now, Space-IX brings the concept into orbit: DE-CIX already integrates low Earth orbit (LEO) satellite networks into existing terrestrial interconnection ecosystems – which has opened new opportunities for broadband access, mobile backhaul, and edge connectivity.
In the ESA project OFELIAS, DE-CIX and the German Aerospace Center (DLR) are working on developing new protocols, processes, and algorithms to make data exchange between Earth and space via laser links more stable and efficient. The goal is to integrate optical satellite communication seamlessly into terrestrial networks. Another building block for the next big leap. In the coming years, the first Internet Exchange in space may well interconnect the growing number of space-based networks directly in orbit, enabling data to flow even more efficiently.
AI Success Between Earth and Space: Minimize Latency, Maximize Efficiency
Where interest exists, demand follows: According to McKinsey, the space economy is booming. By 2035, revenues of up to $1.8 trillion are realistic. Huge players like Google are already investing heavily into research towards building AI datacenters in space – Project Suncatcher will see two satellites launch in 2027 to explore the potential of large-scale data center clusters that orbit the Earth. If we were to take one thing away from SatShow 2026, it’s that the real future isn’t “Earth versus Orbit”, it’s how intelligently we can move data between them – and we need to lay the groundwork now.