Between February 2024 and December 2025, a series of subsea cable cuts across the Baltic and the Red Sea served as a blunt reminder to network teams everywhere: the physical paths their data travels are exposed to events they cannot predict or control. The 2025 blackout across the Iberian Peninsula reinforced the point, taking national grids offline and, with them, the connectivity that businesses in the region assumed would always be there. And when the AWS US-East-1 outage hit in October 2025, a failure in a single cloud region cascaded through thousands of companies worldwide. Cloud and connectivity, for all their built-in resiliency, can still fail.
None of these incidents targeted any specific company. That is precisely what makes them instructive. A bank in Lisbon and a logistics operator in Rotterdam discovered the same thing at the same time: their resilience depended on infrastructure whose physical routes and failure modes they couldn’t control.
The evaluation criteria have expanded
For most of the past two decades, enterprises evaluated connectivity on a familiar set of measures. Coverage came first, then performance. Cost usually closed the discussion. Those still matter. What changed is that a second set of questions has moved out of the compliance department and into the network architecture review.
Who governs the infrastructure my traffic runs on, and where does the data path actually go? If something upstream fails, can operations continue independently?
Part of the shift comes from where data now lives. Two decades of moving workloads into the cloud and outsourcing infrastructure created enormous efficiency. It also placed critical data in the hands of third parties whose terms, jurisdictions, and priorities are constantly re-evaluated. Outages are only the visible part of the risk. The underlying question is ownership: who ultimately controls the environment a business depends on.
The rest comes from experience. Heavy Reading’s 2025 survey found that 81% of enterprises see connectivity loss hitting operations and customer experience immediately. Computer Weekly reported in late 2025 that IT outages cost a UK investment bank around £600,000 per hour, with costs rising across the most dependency-heavy sectors, banking and energy among them. And research from Kiteworks found that one in three organizations ran into an incident tied to who could access or move their data. Some were breaches or transfer violations; others involved unexpected government access.
How satellite fits
Satellite connectivity has become central to this conversation for a structural reason: it offers a data path that does not depend on the terrestrial infrastructure whose fragility started the whole discussion. When cables are cut or a national grid goes down, a link that runs through space is on a genuinely separate failure domain. As space becomes a larger component of global networks, the sovereign questions naturally extend to it.
The interesting development is architectural. Carefully designed networks allow enterprises to obtain sovereign characteristics on commercial infrastructure, whether that means dedicated capacity and controlled gateways or simply auditable routing. Operators like SES are building along these lines, layering sovereign design principles onto existing constellations rather than treating fully dedicated, customer-owned infrastructure as the only serious answer.
For enterprises, this changes the economics. Sovereignty stops being a binary choice between full ownership and none at all, and becomes a set of network design decisions.
Bringing sovereign principles to commercial services
When most people hear “sovereign” they think of governments and defense departments, but that association is now too narrow. Sovereign thinking has moved into more and more network design discussions, and the organizations driving this conversation are commercial.
A regulated bank that needs compliant connectivity for branches in under-served regions. An energy operator running smart grids and offshore platforms where terrestrial networks simply do not reach. A multinational whose core operations cannot tolerate a prolonged outage in any one country. Forrester’s analysis points the same way, defining sovereignty by who controls network technology and how it is operated and regulated. It concerns who makes routing and governance decisions as traffic crosses borders, and how much visibility the enterprise has into those decisions.
Control is not one-size-fits-all
Full control over every segment, covering the space layer and everything on the ground, gets expensive quickly. Most commercial operations do not need it everywhere, and pretending otherwise inflates project costs without improving resilience.
The real exercise is mapping. An end-to-end network path has distinct legs, and each one can be evaluated on its own terms. An enterprise might need strict in-country routing and customer-managed encryption for its financial transactions, while accepting shared commercial infrastructure for everyday corporate traffic. An energy company might prioritize path auditability across jurisdictions for its pipeline telemetry and treat office connectivity as a standard procurement decision.
Regulation is pushing in the same direction. DORA in finance and NIS2 for critical infrastructure tie governance requirements to specific categories of data and operations. The EU Data Act adds cross-sectoral obligations on top, and none of them treats the network as a single undifferentiated whole. Matching the level of control to the actual requirement, segment by segment, is what separates a sovereign architecture that works commercially and one that exists only in a procurement document.
The work ahead
For any organization that depends on connectivity it cannot afford to lose, the near-term work is unglamorous. It starts with an inventory of the data paths and which segments carry what. From there, the exercise is mapping regulatory and operational requirements, along with what each application actually needs, against the options available at each leg. And those options are judged on more than control; performance and stability weigh just as heavily. Space, gateway, ground network, terrestrial handoff, each evaluated on its own terms.
The organizations that start this mapping now will be deciding their exposure on their own schedule, rather than during the next disruption.
