Frankfurt hosts one of the densest concentrations of cloud and interconnection infrastructure in Europe, which is precisely why a single facility problem there can ripple across so many unrelated businesses at once. On August 15, 2026, AWS Direct Connect customers with connections at Equinix’s FR5 site began seeing increased packet loss and degraded connectivity. AWS attributed it to a facility infrastructure issue affecting the network equipment in that room, and unverified reports circulating among affected customers pointed to a cooling failure caused by a water leak.
A contained incident, if you had already planned for it
The practical impact split cleanly along a line most organisations do not think about until it matters: customers with multi-site or redundant Direct Connect configurations spanning other locations were unaffected. Traffic simply routed around the problem. Customers whose connectivity depended solely on that one Equinix room had two options while AWS engineers worked to restore the equipment: fail over to VPN, where that path existed, or absorb the packet loss until the facility issue was resolved.
That is not a design flaw in AWS Direct Connect. It is exactly how a single point of physical presence is supposed to behave when the room it sits in has a problem. The service did what it was architected to do. The gap sat entirely on the customer side, in whichever connections were provisioned without a second physical path.
The pattern Europe keeps re-learning
This is the third notable AWS reliability incident in as many months, following network device failures in US-East-1 and US-West-2 earlier this year. Each one has reignited the same conversation about Europe’s dependence on a small number of hyperscale providers, and each one has produced the same conclusion: the fix is rarely “leave AWS,” it is “stop assuming one region, one facility, or one connection is a redundancy strategy.” The European Commission’s own target, cutting reliance on non-EU technology providers from over 80 percent today to 40 percent by 2030, is aimed at exactly this kind of concentration risk, but sovereignty and redundancy are separate problems. An organisation can run entirely on AWS and still be exposed to a single leaking pipe in Frankfurt if nobody provisioned a second Direct Connect location or a VPN fallback.
For most mid-sized European businesses, a full multi-cloud strategy is overkill and expensive to maintain. A second Direct Connect location, or even a properly tested VPN failover path, is not. The businesses that rode out August 15 without noticing are not the ones with the biggest infrastructure budgets. They are the ones who had already asked what happens if this specific room goes dark, and built an answer before they needed one.
If you are not certain whether your organisation’s cloud connectivity has a single point of failure sitting in one building, or you want a proper resilience review of your AWS, Azure or other cloud architecture before the next facility incident finds the gap for you, contact Excello Digital. We help European businesses design infrastructure that survives the failures that are ordinary, not just the ones that make headlines.