Key Takeaways
- Two major cloud outages in a week further proves how fragile the digital landscape really is.
- Edge and hybrid computing are being pitched as the antidote to single-provider dependence.
- We spoke with Bruce Kornfeld, Chief Product Officer at StorMagic, about how it can be a major business opp for hosts and MSPs.
When AWS’s US-East-1 region came to a halt earlier this month, thousands of services froze with it. Barely a week later, Microsoft Azure followed suit, causing a daylong outage. From banking apps and smart homes to gaming servers, most folks called the two days when “half the internet shut down.”
AWS’s outage stemmed from what was essentially a technical miscommunication between its DNS and system. It moved through DynamoDB and EC2, ultimately taking down hundreds of dependent services for hours, including IoT devices, Fortnite, and various banks and payment services.
When Azure suffered its own breakdown — caused by a faulty configuration change that caused a global outage that lasted most of the workday — it disrupted major services, including Microsoft 365, Xbox Live, and Starbucks, Capital One, and Alaska Airlines.
Bruce Kornfeld, who’s the CPO at StorMagic, a U.K.-based edge infrastructure company, told us these incidents are bigger than a single cloud outage: We’ve built a digital economy around a handful of single points of failure.
Analysts have warned for a long time now that events like these are inevitable. InfoWorld called the “one cloud, one region” model “untenable,” noting that even the best-designed redundancy strategies break down when too many things are dependent on a single provider.
A study titled “Identifying and Mitigating Cloud Concentration Risk” points out that relying on one cloud service provider (CSP) creates “systemic cloud concentration risks.” So it’s not that cloud infrastructure is fragile, but that it’s too centralized to fail without causing a ripple effect across the globe.
The point is that it shouldn’t — and that’s why, Kornfeld said, many companies are looking at edge computing instead. Ask any service provider and they’ll tell you the answer isn’t to abandon the cloud, but to know how to rebalance the way you work with it.
“The cloud has had a massive, long-lasting positive impact … However, the world has changed massively just in the last 10 years. Small, edge sites can no longer be 100% dependent on the cloud if uptime, cost, and performance are important to their business,” said Kornfeld.
The Shift to Edge and Hybrid
One way to mitigate the centralization issue is through edge computing. The market is booming — it’s grown into a $168 billion value and is set to hit $249 billion in the next decade.
“Edge wins out for many reasons — uptime, as evidenced by the recent AWS outage, along with simplicity, performance, and cost,” he said. “Due to technology advancements, the cost of an edge site is no longer hundreds of thousands of dollars, but more like tens of thousands of dollars.”
Think of it this way: Every time a hyperscaler like AWS or Azure goes down, smaller businesses feel the punishment. Edge computing limits that by keeping the most essential operations close to where the data is created.
Basically, instead of sending every request to a data center that could very well be on the other side of the world, local infrastructure will process, cache, and serve information on-site. That means even if a major cloud region goes down, those local networks keep running.
It’s almost funny when you think about where cloud computing even began. Pitched as a data center replacement, experts have noted that the ecosystem seems to have evolved faster than its infrastructure can keep up.
That doesn’t mean the cloud goes away. It still handles the heavy stuff, like AI/ML training and long-term storage. But Kornfeld said that for hosts and managed service providers (MSPs), they should start showing their clients a world beyond just the cloud.
“Instead of ‘cloud only’ or ‘cloud first,’ they can start educating on the benefits of a more balanced approach between an on-prem edge computing approach that is ‘cloud enabled’ but not too dependent on cloud,” he explained.
Still, there is no “one-size-fits-all” solution for building that balance. What separates the long-term winners, he said, is their ability to understand their client’s true needs — and not just slap on another gen-AI project, for example.
“Virtually all organizations are trying to figure out what the right approach is for them,” he said. “A lot of AI projects fail due to mis-set expectations and a poor understanding of desired outcomes.”
Kornfeld pointed to the State of AI in Business 2025 report, which found that nearly 95% of generative-AI pilots fail to deliver measurable results. And it’s not because they don’t work; it’s because they’re built on hype instead of a clear plan.
What Hosts and MSPs Can Do
Kornfeld stressed that the smartest service providers will leverage this point in time — a time when cloud dependency is huge, and edge computing is growing — and rethink their service offerings so that they can better guide their clients to a “more pragmatic approach to cloud computing.”
“The nice thing here for the service providers is that there is just as much (or more) revenue available to them, especially if they can help their customers lower their overall cost of compute,” Kornfeld said. “Let’s not forget — cloud computing can be very expensive, and filtering in some on-prem edge computing tech can help them lower costs significantly.”
There’s proof in the pudding: A Cornell study found hybrid edge-cloud architectures can reduce costs by 80%. Another analysis found that cloud computing can cost 40% more than buying hardware (on-premises/collocation/edge computing) in some cases.
“The tech isn’t as expensive or hard to deploy and manage as it used to be, and there is a huge opportunity for value-added services for the service providers,” Kornfeld added.
Actually, it’s a lot like what analysts at cyber risk management provider RB Advisory and others have urged:
- Diversify dependencies and design for failure
- Map which services rely on which clouds
- Split DNS, authentication, and data replication across multiple regions or vendors
- Build backup paths that don’t hinge on a single provider’s control plane
“The service providers that will win in the long term are those that can consult with their users to truly understand their requirements and build from there,” Kornfeld said. “If a cloud provider has an outage, an internet provider goes down temporarily or an unexpected local disaster prevents access to the cloud, there are architectures that can weather these storms.”
