The first days were the shock. Then came the constant pressure, and now a much more uncomfortable phase is beginning, but the cloud is still working.
Applications keep responding, systems have not gone down, and much of the digital infrastructure in the Middle East remains operational. From the outside it might seem like the worst is over, but inside technical teams the feeling is very different, because the problem is no longer collapse, it is attrition.
The scenario has changed. We are no longer managing incidents, we are living under continuous pressure
For weeks, several cloud regions in the Middle East have been operating under conditions they were not designed to sustain indefinitely. Some availability zones remain degraded, others are running at reduced capacity, and much of the critical workload has had to shift to European regions to maintain operational stability. AWS continues to recommend migrating services out of certain affected regions while repairs progress and uncertainty over new attacks persists.
As we already saw when the cloud went from absorbing blows to becoming a direct target, the conversation around cloud resilience used to revolve around isolated incidents. A power outage, a fire, a network failure, or even the complete loss of an availability zone. Everything was designed to absorb specific events and return quickly to normal. What is happening now is different, the infrastructure is not managing an incident. It is operating under continuous geopolitical pressure.
And that is where limits begin to emerge that are rarely part of the public conversation. Because a cloud region is not just distributed software. It depends on something far more physical and far more fragile: stable energy, constant cooling, international connectivity, logistics, maintenance, fuel, technical staff, and network capacity all working in a coordinated way. When one of those pieces fails, the architecture absorbs the impact. When several remain degraded for weeks, the system starts to operate differently.
Energy, latency, and submarine cables — the physical limits nobody mentions
In Bahrain and the UAE, part of the cloud infrastructure continues to run in contingency configurations following the attacks of recent weeks. Some operations have shifted to other regions to reduce risk and maintain availability, particularly to Europe. Frankfurt, Ireland, and other European regions have absorbed part of the traffic and workloads that were previously processed locally. But moving workloads out of the region carries a cost that normally stays invisible, distance.
More latency, greater pressure on interconnections, and an even heavier dependence on submarine cables and international traffic routes. Because the cloud does not eliminate geography. It only makes us not think about it — until something happens. This is where one of the most delicate factors in the entire situation comes into play: energy.
Data centers are prepared for brief supply outages. They have UPS systems, batteries, diesel generators, and redundant architectures designed to absorb temporary failures — but all of that infrastructure is built for time-limited incidents, not for operating for weeks in an environment where energy stability has become a structural problem. And that is starting to have real consequences.
Backup systems are running more hours than planned. Generators are consuming more fuel. Cooling is losing its operational margin. Preventive maintenance is becoming more complex. And energy costs are rising just as the conflict is also affecting the region's oil infrastructure and logistics. The real danger is not running out of energy — it is no longer being able to rely on it. Because a data center does not need to go completely dark to become a problem. It is enough for the predictability it runs on to disappear, and when that happens, the entire logic changes.
Redundancy wears out too, when the cloud enters survival mode
Decisions that are normally made in pursuit of efficiency start being made in pursuit of operational survival. Moving workloads ahead of schedule, keeping redundancies permanently active, over-provisioning capacity, accepting higher latencies, prioritizing stability even at greater cost. In other words, the cloud enters war mode, and when that happens, even the most solid principles of cloud start showing their limits.
Redundancy works very well when problems are temporary. But redundancy also wears out. Multi-region architectures are designed to absorb incidents, not to live indefinitely on contingency mechanisms. Each additional migration consumes resources, each region absorbing external traffic loses its own margin, each increase in latency adds operational complexity. Meanwhile, another layer of risk keeps growing beneath the sea. The submarine cables in the Gulf and the Red Sea remain one of the most sensitive points in the entire global infrastructure.
Although no confirmed large-scale outages have occurred in recent weeks, several projects and maintenance operations continue to be affected by regional instability, and the problem is not only physical damage. It is recovery time, because repairing a submarine cable does not take hours. It can take months. And in a scenario where traffic is already being redistributed across continents, every available route becomes even more critical. That is precisely what makes this phase so unsettling. The cloud is still working, but it is increasingly dependent on temporary configurations, alternative routes, forced redundancies, and operational decisions made under constant uncertainty. The system has not collapsed, but it has not truly returned to normal either. And perhaps that is the most important lesson in all of this.
For years, the cloud was presented as something abstract, distributed, and practically intangible. But recent weeks have demonstrated something different: the cloud wears out too. Not only from technical failures, but from sustained pressure on the physical infrastructure that supports it. Because behind every digital service there are still buildings, energy, water, networks, fuel, people, and geopolitical decisions, and when all of that remains under strain for too long, even the most resilient infrastructure starts to change the way it operates.
The cloud has not gone down, but it is no longer working the way it used to.
Fuentes
• Tom’s Hardware — degradación prolongada de regiones AWS en Oriente Medio y migración de cargas
• DataCenterDynamics — impacto operativo sostenido sobre infraestructura cloud regional
• Financial Times — presión sobre infraestructuras energéticas y logística en Oriente Medio
• Rest of World — riesgos sobre conectividad y dependencia de cables submarinos
• AWS regional status y comunicaciones técnicas sobre recomendaciones de migración de workloads