Cloud Computing in Critical Sectors: Innovation Engine or Systemic Risk? Lessons from the AWS Outage

Yesterday morning, October 20, 2025, went down in tech history as the "Black Monday" of the digital age. The global Amazon Web Services (AWS) outage, centered on the critical US-EAST-1 region, was not just another minor hiccup. It was a systemic heart attack. For more than fifteen hours, we watched a domino effect that paralyzed not only streaming services and social media, but also banking systems, airlines and even, most painfully for us, the services of Poczta Polska, the Polish national postal operator.

Caused by a mundane DNS error in the DynamoDB database service, the incident exposed a frightening truth about our civilization: we have built our global economic and social system on infrastructure controlled by just three (well, maybe four) corporations.

Read more: To learn how this outage was followed by the Cloudflare incident, see our recap, "The Autumn of the Middle Ages".

As a cybersecurity editor, I have followed the cloud debate for years. For a long time, the "Cloud First" narrative dominated. The cloud stood for modernity, agility and innovation. After yesterday's chaos, that narrative looks not just naive but outright dangerous.

Today I don't want to ask whether companies should move to the cloud. I want to ask whether the state's strategic sectors (the military, healthcare and critical infrastructure) can afford this model at all. And if they can, at what price?

The Siren Song of Innovation: Why We Fell in Love with the Cloud

Before we get to the risks, let's be fair. The cloud is not a marketing invention. It is a fundamental revolution in access to technology, and it offers three powerful advantages.

First, economics. The cloud pulled off a small miracle by turning massive capital expenditures (CAPEX) into flexible operating expenses (OPEX). Instead of spending years building their own server rooms, buying expensive hardware and worrying about cooling and power, companies could "rent" computing power by the minute. This democratized innovation and let startups compete with giants.

Second, scalability. The cloud solved the "Black Friday" problem. Our infrastructure can now stretch like rubber, shrinking overnight and expanding to unimaginable scale at peak load.

Third, and perhaps most important today, innovation. The real value of the cloud is no longer virtual machines. It is ready-to-use advanced services: platforms for Big Data analytics, machine learning (ML) and, above all, generative artificial intelligence (AI). No serious company today "builds" its own AI model from scratch. It rents one from Google, Microsoft or AWS. That is why cloud migration is no longer optional. It has become a competitive imperative.

The Giants' Hangover: Concentration Risk and the Loss of Sovereignty

And then, at the height of our dependence, came October 20. The AWS outage showed the other side of the coin, and it is no longer a theoretical risk. It is a tangible reality.

The problem isn't that the cloud can fail. The problem is that the whole cloud can fail at once.

This is systemic concentration risk. We have built the global digital economy on a handful of pillars. When one of them, in this case the US-EAST-1 region, starts to wobble, the whole world feels the quake. Yesterday's incident was a textbook example of a cascading failure. DNS for DynamoDB went down and took 64 other, seemingly independent AWS services with it. This wasn't a server failure. It was a Jenga tower collapsing: pull out one block at the base and the whole structure comes down.

The second problem, just as serious, is sovereignty. Why did a data center outage in Northern Virginia affect Poczta Polska? That rhetorical question shows how the abstract concept of "digital sovereignty" translates into hard operational reality. Our data and processes, including those essential to the functioning of the state, depend physically and logically on the infrastructure of a commercial company subject to the jurisdiction of another country.

For an online store or a dating app, an outage lasting more than a dozen hours means a huge financial and reputational loss. But for a hospital, a power plant or an army command system, it is a catastrophic scenario.

Sector by Sector: When "Cloud First" Meets Reality

Let's look at what this new reality means for strategic sectors. For them, cost and innovation are not enough. What matters most is resilience and mission assurance.

1. Military and Defense: The Digital Battlefield

Armed forces around the world, including NATO, are investing in the "Combat Cloud" concept. The idea is for data from drones, satellites, tanks and soldiers (C4ISR) to flow into one place, be analyzed by AI and give commanders a real-time picture of the situation.

The problem: Public cloud architecture runs on a "best-effort" model. The military has no room for that. Command systems must be deterministic: they must always work, predictably and with guaranteed response times. You cannot afford an "authentication error" or a "race condition" while a missile is being launched or intelligence about an attack is being analyzed.

Takeaways: Relying on a commercial, foreign public cloud for core command systems is strategic suicide. An outage like yesterday's could blind an army at the critical moment of a crisis. That is why the military model has to be hybrid. Historical data analysis, training and logistics can run in the public cloud. But real-time C4ISR systems must stay in a private (on-premise) cloud, often physically isolated (air-gapped), with key computing capabilities pushed out to the "edge" (edge computing): to vehicles, ships and forward bases.

2. Healthcare: A Patient with a Near-Zero RTO

Healthcare is going through a cloud-driven revolution. Electronic Health Records (EHR), telemedicine platforms, AI analysis of diagnostic images (X-ray, MRI) and rapid genomic testing all happen in the cloud.

The problem: Two metrics are critical in healthcare: RTO (Recovery Time Objective, or how quickly a system must come back after a failure) and RPO (Recovery Point Objective, or how much data we can afford to lose). For a hospital EHR system that an emergency department depends on, RTO must be close to zero. A doctor must have immediate access to the patient's history, allergies and medications.

Takeaways: In healthcare, a multi-hour AWS outage is not a business problem. It is a direct threat to life and health. Picture an ER doctor who cannot access the records of a critically ill patient. Remember that ransomware attacks on hospital networks are no longer theoretical but a reality, and a cloud outage can have equally catastrophic consequences. That is why relying on a single public cloud provider in healthcare, whatever its marketing promises, is unacceptable. It calls for a hybrid architecture or, more likely, a multi-cloud one, where data is replicated and systems are ready for immediate failover to another, independent provider.

3. Critical Infrastructure: When a Digital Failure Becomes a Physical One

This is the backbone of the state: energy, transportation, water supply and the financial sector. These sectors are turning to the cloud more and more boldly to manage smart grids, optimize energy distribution, control traffic and analyze risk.

The problem: Critical infrastructure is where IT (information technology) and OT (operational technology) converge. Put simply, digital systems control physical equipment, and an IT failure can trigger a disaster in the physical world. If a power grid balancing system built on cloud-based analytics stops working, the result could be a widespread blackout. If rail traffic control systems fail, the result could be chaos or a disaster. It is also worth remembering that the Polish energy sector is the target of thousands of cyberattacks, so the added risk of cloud dependence could be the straw that breaks the camel's back.

Takeaways: This sector needs the most conservative strategy. It should adopt a "private cloud first" policy for all OT systems. SCADA systems that control the flow of energy or water should never be directly exposed to the risk of a commercial public cloud outage. The public cloud is an excellent tool for business analytics (IT), demand forecasting and customer service, but the operational core must remain under full, sovereign control.

The Polish Context: Fast Adoption and Strategic Debt

Against this backdrop, the Polish cloud market is fascinating. We are in a phase of rapid growth, fueled by Google, Microsoft and AWS opening cloud regions in Poland. Companies and the public sector have rushed to seize the new opportunities, mostly in pursuit of AI innovation.

The problem is that our adoption is often tactical rather than strategic. We lack maturity. As reports show, many companies and institutions, excited about their new toys, are building their strategies around a single provider.

In doing so, we are taking on massive "strategic debt." We migrate quickly to cut costs and gain access to AI, but we don't invest in a resilient architecture. We become hostages to a single ecosystem (vendor lock-in). Yesterday's outage showed how painful paying off that debt can be. Our digital ecosystem, built in a hurry and without diversification, may be disproportionately vulnerable to future hyperscaler outages. It is also worth remembering that the NIS2 Directive requires essential and important entities to ensure business continuity, and relying on a single cloud provider could be seen as a breach of those requirements.

From "Cloud First" to "Cloud Smart": Architecting for Resilience

So what should we do? Turn our backs on the cloud and go back to building server rooms in the basement? Absolutely not. That would set us back a decade and mean giving up on innovation.

The events of October 20, 2025, do not mark the end of the cloud era. They mark the end of the era of the naive cloud. "Cloud First" is dead. We need to replace it with a "Cloud Smart" strategy.

"Cloud Smart" means understanding that the cloud is not a monolith, and that resilience is not a feature you can "switch on" but a foundation you have to design. The answer to concentration risk is not to run away. It is diversification.

Multi-cloud is becoming an architectural imperative.

Relying on a single provider is a strategic mistake. True systemic resilience requires spreading critical workloads across at least two independent providers (e.g., AWS and Azure, or GCP and a private cloud). There are three basic models for this kind of architecture:

1. Backup and Restore: The simplest model. Backups are kept with a different provider, and after an outage systems are restored manually. Recovery time (RTO) is long (hours or days), but the data is protected. A good fit for archival systems.
2. Active-Passive: The primary system runs in Cloud A, while a "warm" standby environment in Cloud B is ready to take over traffic. Data is replicated continuously, and in the event of an outage, traffic is switched over. RTO is measured in minutes or hours. This is the bare minimum for most critical services.
3. Active-Active: The gold standard. The application runs in both clouds at the same time, with traffic distributed between them. If one cloud fails, end users don't notice: traffic automatically flows to the other. RTO and RPO are close to zero. This approach is expensive and complex, but for real-time banking, emergency medical systems or infrastructure control, it is the only acceptable option.

Of course, multi-cloud is complex. It requires standardization, automation (e.g., with tools like Terraform) and containerization (Kubernetes) so that applications are "portable." It also demands deep expertise from IT teams. But after October 20, we know that the cost of this complexity is still lower than the cost of systemic paralysis.

Yesterday's outage was a brutal but necessary wake-up call. The cloud will remain an engine of innovation. But for the strategic sectors that our physical and economic security depends on, we can no longer treat it as some magical, infallible entity. We have to treat it as what it is: a powerful but complex tool that demands mastery, risk diversification and strategic prudence.

Aleksander