How do cloud service outages happen?

As users unsuccessfully tried to access Washington Examiner content, listen to music, or scroll on social media, they were met with a warning of an error in the cloud and a sense of deja vu. The Cloudflare outage was one of two major cloud service outages in the span of a month, following the Amazon Web Services outage in late October. 

Behind the scenes of users clicking out of their interrupted website, is a world of cloud service providers, computing researchers, and policy experts seeking ways to make the internet more resilient and combat outages like those of the past several weeks. 

TRUMP BIDS FAREWELL TO MTG AFTER SHOCK ANNOUNCEMENT OF HER RESIGNATION

The Washington Examiner spoke with a few experts to better understand what causes these outages, what risks they pose to consumers and providers, and what is being done in the private and public sectors to best prevent, or handle, future outages.

Cloud service providers, explained

AWS became the first cloud service provider in the 21st century, fundamentally changing the way companies and individuals access their data experience. 

Before cloud providers, users would need to physically invest in data processing infrastructure on their own. Dr. Timothy DeStefano, a research professor at the Georgetown University McDonough School of Business, said this used to require companies to buy, install, and maintain their own servers, storage devices, and software, which cost a pretty penny.

“What cloud computing allows businesses to do is it negates all that upfront, sunk investments, and it allows them to access those technologies through it through a subscription,” DeStefano said. 

DeStefano explained that cloud service providers give access on demand, allowing companies to pay for and access services only when they need it, not at a constant rate. This benefit, he said, combined with the lack of a need for physical data processing infrastructure, allows newer companies to grow quickly and older companies to efficiently restructure.

The top three cloud service providers are AWS, Microsoft Azure, and Google Cloud Platform. Together, they make up 63% of the industry market share, as of Q3 2025, according to Synergy Research Group. The rest of the market includes providers like Cloudflare, IBM Cloud, Alibaba, Oracle, Digital Ocean, and Dell Technologies. 

The market is rather centralized. Dr. Timothy Wood, a computer science department professor at George Washington University, said this isn’t necessarily a bad thing.

“I wouldn’t necessarily call it a monopoly, but there’s a relatively small number of providers that we rely on,” Wood said. “I think part of what makes this complex is that the reason that computing and Internet services have gotten so cheap is partly because of the economy of scale that these providers get from being so huge. So the bigger you are, the cheaper you can offer these services.”

Wood also said that by being big, these providers can also fundamentally provide better services.

On the other hand, a centralized market can cause significant global outages when one of these big providers has a glitch.

What causes cloud service outages?

There are essentially limitless ways that cloud services can fail, though it can often come from imperfections like bugs and human error. Wood told the Washington Examiner that “it can be anything.”

“It ranges from electricity striking something and bringing down power to some part of a data center, to a shark chewed through a cable off of the coast of Africa and part of Africa lost internet for a long time,” Wood said. “Or it can be malicious. It can be attackers actively trying to bring down these services.”

According to Cloudflare’s CEO Matthew Prince, Tuesday’s outage was caused by a bug that essentially triggered a file to double in size and overwhelm the system. Prince said the outage was the largest in Cloudflare’s recent history since 2019 and clarified that it “was not caused, directly or indirectly, by a cyber attack or malicious activity of any kind.”

The Oct. 20 AWS outage was caused by error rates and latencies in the provider’s US-East-1 region in its data center cluster in Northern Virginia.

HOW LOUDOUN COUNTY’S COMMERCIAL TAX BASE INITIATIVE TURNED VIRGINIA INTO A DATA CENTER HUB

What risks do outages pose to providers?

DeStefano simply summed up the risk that outages pose to cloud service providers in three words: “loss of customers,” he said.

Once an outage happens, it makes room for other providers, in the small, hyper-competitive landscape, to pounce.

“There will be an effort to try and increase the numbers of customers that they get, and try and detract customers from companies that have an issue,” DeStefano said.

“You can imagine the competitive pressure that this is putting on cloud providers to make sure that these types of things don’t happen again,” DeStefano said.

Wood also said outages can potentially cause providers to not be as profitable, forcing them to drive down their prices if they are not reaching the level of service they have guaranteed to their customers.

“They do have an economic incentive to try to keep their systems online as much as possible,” Wood said.

Other risks from outages simply depend on the type of malfunction that causes the incident.

“In the event of a fire where there’s damage to the physical infrastructure, you can imagine that that would have long term effects on the providers themselves,” DeStefano said, for example.

What risks do outages pose to consumers?

Thankfully, on the consumer side, the only real risk of an outage is a short term inconvenience to cloud users, DeStefano told the Washington Examiner.

Wood emphasized that the main risk of outages to everyday users is just that the availability of websites is downgraded. He did note, however, that if websites which rely on a cloud provider experiencing an outage temporarily disable that cloud service, it could expose the website to attackers in the event that the provider also included security services, like Cloudflare can.

“They could potentially disable their use of Cloudflare during that outage,” Wood said, referring to companies using cloud systems. “They could just be downgrading the performance of the system, but potentially it could be exposing themselves to more attacks. Because Cloudflare is really providing, both, performance and it’s usually providing some kind of security hack detection and mitigation systems.”

What can be done to mitigate outages?

DeStefano, who researches how cloud computing impacts financial firms, said firms and companies are experimenting with layers of computing services to have backup in case an outage occurs.

“One of the things that we know firms are doing is they use multiple cloud providers,” DeStefano said. “They typically have a hybrid system where they rely on Cloud for some of their compute and data processing needs, but they also have their own internal servers within the organization. So having a hybrid approach for the user is something that helps build resilience.”

Wood said that cloud computing providers also increase their reliability by having different service locations in different geographical areas, which significantly helps in the instance of a weather-related outage.

Some lawmakers and policy experts have tossed around regulatory changes targeted at the cloud service provider industry to mitigate centralization.

In May, Sens. Elizabeth Warren (D-MA) and Eric Schmitt (R-MO) introduced the Protecting AI and Cloud Competition in Defense Act, aimed at preventing large artificial intelligence and cloud computing companies from boxing out smaller competitors.

“It’s a mistake to let Silicon Valley monopolize our AI and cloud computing tools because it doesn’t just stifle innovation, it increases costs and threatens our national security,” Warren said in a May statement. “Our bill will make sure the military can access cutting-edge tools and will keep our markets strong and our information secure.”

JUDGE RULES META DID NOT CREATE MONOPOLY WITH PURCHASES OF INSTAGRAM AND WHATSAPP

When asked about public policy and regulatory approaches to outages, Wood said it could help with the centralization and thus could help prevent “single point of failure” outage concerns, but cautioned that regulatory approaches could have significant effects on consumers.

“So you could regulate that these things needed to be a smaller scale, perhaps, but then it would affect the cost and it might affect some of the capabilities that they could provide. Unless you deal with things like different providers having to work together and so on, which may be possible, but it’s complicated,” Wood said.