top of page

The Cloud Goes Down. Does the Hospital?

Austin Delaney
45 minutes ago
10 min read

What a week of major technology outages reveals about healthcare's growing dependence on digital infrastructure.


By Austin B. Delaney, EVP, Technology Design & Advisory


It seems like every time we turn around lately, another major technology platform is experiencing an outage. And while I’d originally planned this article for a later release, the recent string of major disruptions made it a conversation worth having sooner rather than later.

 

Just this month, we've seen disruptions affecting Microsoft, Apple services, the National Security Agency, airline operations, and major financial institutions. The disruptions have affected communications, transactions, travel, and other everyday services on which millions of people depend.

 

And it's not just a matter of perception. According to Cisco ThousandEyes' November reporting, the United States experienced 282 monitored network outage events during the week of November 9–15, a 21% increase over the preceding week.

 

For most of us, these incidents amount to an inconvenience. Maybe we can't check our email, access an app, or complete a transaction. Frustrating, certainly, but usually temporary.


But what happens when the systems that go down are supporting patient care?

 

It's a question worth asking as hospitals become increasingly digital, interconnected, and dependent on technologies that extend far beyond their physical walls.  And from my perspective, working at the intersection of healthcare technology, facility design, and operational planning, it's a question that deserves considerably more attention.



A Week of Outages, and a Much Bigger Question

 

The recent disruptions weren't necessarily connected, nor were they all the result of cyberattacks. In fact, that's part of what makes them interesting.

 

A software update, an authentication failure, a network configuration issue, or an infrastructure malfunction can produce widespread consequences without a malicious actor ever entering the picture. Look at what happened with JPMorgan Chase last week. A software bug involving an on-premises hardware system disrupted transactions and created ripple effects across other financial institutions and the Zelle payment network. Some banks experienced transaction delays not because their own infrastructure had failed, but because they depended on transactions involving Chase.

 

That's the part worth paying attention to.

 

The takeaway isn't that technology is becoming inherently less reliable. It's that the systems supporting our daily activities have become extraordinarily interconnected, with layers of AI and automation being added as new dependencies and potential failure modes continue to emerge.  A single service failure can affect numerous applications, organizations, and end users, sometimes in ways that aren't immediately obvious.

 

Healthcare is no exception. In fact, the stakes are considerably higher.



The Modern Hospital Is More Than a Building

 

Think about what it takes to operate a contemporary hospital.

 

Electronic health records support clinical documentation, medication management, and care coordination. Nurse call and clinical communication platforms connect caregivers with patients and their loved ones. Real-time location systems help locate equipment and support operational workflows. Building automation systems manage environmental conditions, while access control and security systems protect sensitive spaces. Add virtual nursing, ambient documentation, robotics, artificial intelligence, remote monitoring, and other emerging technologies to the mix. Many of these capabilities increasingly exchange data with one another, often across systems supplied by different vendors.

 

The list goes on, and the number of potential interdependencies grows along with it. The federal government's SAFER Guides recognize the importance of examining the safety implications of interconnected health IT, including how electronic systems support clinical processes and how organizations prepare for their unavailability. Many of these capabilities depend on shared infrastructure, centralized applications, cloud services, external vendors, and network connectivity. This interconnectivity is what makes the modern smart hospital possible, but it also introduces dependencies that deserve careful consideration.

 

As the lines between physical and digital infrastructure continue to blur, hospitals are becoming less a collection of independent systems and more a single, interconnected environment. That creates tremendous opportunities for innovation, but also new vulnerabilities when those systems don't work as intended.

 

Consider a hypothetical example.

 

A hospital implements an integrated patient room where nurse call, room controls, digital signage, patient engagement, and clinical communication share information through a common platform. Under normal circumstances, the result can be a more coordinated patient and caregiver experience.

 

But what happens when one of those integrations becomes unavailable?

 

Does the nurse call system continue operating independently? Can caregivers still receive critical notifications? Does the patient retain control of their environment? Are essential workflows preserved, or do multiple capabilities become unavailable simultaneously? And perhaps more importantly, have these dependencies been identified, documented, and tested before the first patient occupies the room?

 

The measure of a well-designed smart hospital isn't just what its systems can accomplish when everything works. It's also what they can continue doing when something doesn't.



Redundancy Isn't the Same as Resilience

 

Healthcare facilities have long understood the importance of redundancy.

 

Emergency power, redundant network pathways, backup equipment, and other safeguards are established elements of critical infrastructure planning. Some represent industry best practices, while others are governed by applicable codes and standards.

 

For example, FGI's guidance on healthcare facilities responding to emergency conditions discusses the importance of maintaining critical functions during emergencies, including requirements for essential electrical systems. Similarly, NFPA 99 (Health Care Facilities Code), establishes risk-based requirements for healthcare infrastructure, while NFPA 70 and NFPA 110 address electrical installations and emergency power systems.

 

Beyond physical infrastructure, the HHS Healthcare Cybersecurity Performance Goals outline recommended practices addressing system protection, recovery, and organizational preparedness. These codes and frameworks address different aspects of resilience, but none eliminates the need to understand how critical technologies and workflows interact.

 

A hospital can have redundant infrastructure and still experience a significant operational disruption. Two internet connections don't necessarily help if both rely on the same upstream service. Multiple applications don't provide meaningful independence if they share a common authentication platform. Backup servers accomplish little if the underlying workflows cannot function without an unavailable external dependency.

 

And even when the technical safeguards perform exactly as intended, the people operating the facility still need to know what happens next. This is where the distinction between redundancy and resilience becomes important.

 

Redundancy provides alternatives. Resilience preserves essential operations.

 

A resilient healthcare environment requires more than redundant hardware or connectivity. It requires an understanding of which services are truly critical, how they depend on one another, what happens when those dependencies fail, and how clinical operations will continue. The U.S. Department of Health and Human Services reinforced this point in its October cybersecurity guidance, encouraging organizations to prepare for situations in which critical systems, or even internet connectivity itself, may become unavailable.

 

The guidance also emphasizes testing incident response plans and maintaining alternatives such as paper-based workflows and radio communications. Having a recovery plan is important. But the ability to operate safely while systems are unavailable is equally important.



Healthcare Has Already Seen the Consequences

 

Unfortunately, we don't have to look far back, or very hard, to find examples of widespread technology disruptions affecting hospitals.

 

In March, Vanderbilt University Medical Center experienced an approximately four-hour electronic health record outage, temporarily placing Vanderbilt University Hospital on ambulance diversion. The incident was not attributed to a cyberattack or ransomware.

 

In September, three Johns Hopkins Health System hospitals experienced a technology network disruption that temporarily restricted ambulance arrivals. Again, the health system indicated the incident was not related to a cyberattack.

 

And just last month, an October network outage affected Mass General Brigham, forcing hospital staff to activate downtime procedures and rely on paper documentation and manual processes. Fans of The Pitt might recognize a similar scenario from Season 2, Episode 9, when the fictional emergency department has to abandon its computers and return to paper charts, handwritten orders, and old-fashioned communication. For healthcare professionals who've experienced an actual downtime event, that scenario probably hits a little too close to home.

 

Three separate incidents, each demonstrating how technology disruptions can have immediate consequences for clinical operations, even without evidence of malicious activity. But the July 2024 CrowdStrike incident demonstrated how a single technology failure could have consequences on a much larger scale.

 

A faulty software update distributed by one widely deployed cybersecurity vendor disrupted computer systems around the world, including those supporting healthcare delivery. It illustrated how an issue originating with a single technology provider could simultaneously affect organizations operating across entirely different industries and geographic regions. The impact extended well beyond any single healthcare institution.

 

A 2025 study in JAMA Network Open examined the impact of the CrowdStrike incident on U.S. hospitals. Researchers detected network disruptions at 759 of the 2,232 hospitals studied, identifying nearly 1,100 affected digital services. Among the classified services were patient-facing applications, clinical information systems, staff scheduling tools, and other operational capabilities. Although the study did not establish clinical disruption at every affected hospital, its findings demonstrated just how broadly a single technology failure could affect healthcare infrastructure.

 

Technology outages rarely respect the organizational boundaries we've created around individual systems.

 

An EHR outage isn't necessarily just an EHR problem. A communications outage isn't necessarily just a telecommunications problem. And a software issue originating outside the hospital can affect clinical and operational functions inside it. What begins as a technical failure can quickly become a patient care challenge.



We Need to Design for the Downtime, Too

 

This is where I believe healthcare technology planning needs to evolve.

 

On capital projects, we spend considerable time determining what technologies a facility should have, how they should integrate, and what the future patient and caregiver experience should look like. And those are all important conversations. But resilience needs to be approached as a shared responsibility across digital infrastructure, the physical environment, and clinical operations.

 

Hospitals are no strangers to downtime planning. Most have extensive policies, emergency procedures, and, yes, those enormous binders outlining what to do when systems become unavailable. But having a documented process and successfully executing it under pressure are two very different things.

 

When multiple systems fail, staff are juggling competing priorities, and patient care is increasingly affected; how well do those procedures actually hold up? That's not something we should be discovering for the first time during a real incident.

 

The 2025 SAFER Guide for EHR Resilience addresses this directly, emphasizing contingency planning, staff preparedness, and the ability to maintain safe clinical operations during system disruptions. Similarly, AHRQ-supported research into EHR downtime has examined how disruptions affect hospital workflows and how contingency plans can be evaluated before they're needed.

 

These principles should extend beyond the EHR to the broader technology ecosystem supporting a modern hospital. That starts with asking more deliberate questions throughout the project lifecycle.

 

During programming and visioning: Which clinical and operational functions absolutely must continue during a technology disruption? What systems support them, and which dependencies introduce unacceptable risks or unresolved unknowns?

 

During design: Are technology vendors' network, cybersecurity, integration, and infrastructure requirements being reviewed and coordinated with the same rigor as other critical project submittals? Are these requirements compatible with the Owner's enterprise standards, and have their implications for facility infrastructure and system operation been addressed?

 

During procurement: What do vendors actually guarantee regarding availability, failover, data access, interoperability, and recovery? How will their systems interact with the existing enterprise environment, and does the Owner have the internal expertise, resources, and support arrangements necessary to sustain them?

 

During commissioning and activation: Are we validating not only that each system functions correctly, but that critical workflows can continue when a system is unavailable? Are realistic technology failures incorporated into Day in the Life exercises, tabletop scenarios, and staff readiness activities?

 

These questions cannot be answered by the technology team alone. They require deep collaboration among clinicians, operational leaders, IT, facilities, design professionals, technology consultants, vendors, and the teams responsible for bringing a new facility online. And they require a willingness to challenge assumptions before those assumptions become embedded in the building.

 

In my work, I see the value of connecting these disciplines early, particularly when technology decisions made during design will shape how a hospital operates for decades. One practical approach is to take a critical clinical workflow, identify the technology dependencies supporting it, and examine what happens when one or more of those dependencies becomes unavailable.  What still works? What fails? What can continue locally? Who is notified? And, perhaps most importantly, what does the caregiver actually do next? That exercise connects digital design decisions with their real-world operational consequences.

 

A system that passes its technical acceptance tests hasn't necessarily demonstrated that the hospital can continue operating when it fails.



The Next Generation of Hospitals Raises the Stakes

 

There's an interesting tension emerging in healthcare technology.

 

We're moving toward hospitals that are more automated, more intelligent, and more responsive to the needs of patients and caregivers. Artificial intelligence may increasingly coordinate activities across multiple systems. Robotics will assume a greater and more autonomous role in logistics. Virtual care will extend clinical operations beyond the hospital campus. Digital experiences will connect patients to services throughout their care journeys, from home to hospital and back again.

 

All these developments offer considerable opportunities for improved efficiency, care delivery, and the overall patient experience. But as more functions become connected, a failure in one system can potentially affect several others.

 

Consider an AI-enabled hospital workflow that detects an operational issue, identifies an appropriate response, and initiates actions across multiple platforms. That's a powerful capability when the underlying systems are available, the information is reliable, and the AI behaves as intended.

 

But what happens when the platform loses access to one of those systems? Does it recognize incomplete or outdated information? Can it distinguish between a system that is unavailable and one reporting that nothing is wrong? Does it stop taking action when the conditions supporting its decisions are no longer reliable?  And if the automation itself produces an unexpected response, can a human intervene before that response cascades into other connected systems?

 

These questions become increasingly important as healthcare explores more autonomous technologies.

The concern isn't simply that AI might fail. It's that a system designed to coordinate activities across multiple platforms can introduce new dependencies, amplify errors, or affect workflows well beyond its original purpose. NIST's AI Risk Management Framework emphasizes the importance of validity, reliability, safety, and ongoing risk management throughout the AI lifecycle.

 

In healthcare, those considerations must extend into the built environment and the operational systems with which AI may interact. The ability to recognize degraded performance, maintain appropriate human oversight, and preserve essential functions should be part of how we evaluate these technologies from the outset.

 

The smarter our hospitals become, the more deliberately we need to design how they behave when they're not so smart.



Building Hospitals That Can Bend Without Breaking

 

None of this is an argument against cloud computing, artificial intelligence, or connected healthcare technologies. Quite the opposite.

 

The benefits of integration are substantial, and healthcare should continue pursuing technologies that improve care delivery, operational efficiency, and the patient experience. But progress needs to be accompanied by an equally serious commitment to resilience. Digital infrastructure and operational preparedness aren't separate considerations. They're complementary parts of the same responsibility.

 

That responsibility extends from capital planning and facility design through procurement, commissioning, activation, and ongoing governance. It requires helping nontechnical stakeholders understand the operational implications of the systems they're selecting, the infrastructure supporting them, and the dependencies being introduced.

 

Above all, it requires recognizing that individual systems can be highly reliable while the workflows connecting them remain vulnerable. No amount of investment will eliminate every possible failure.

 

When it comes to technology outages, the question isn't if, but when. But an outage doesn't have to become a crisis. With thoughtful planning, resilient design, and operational preparedness, the hospitals we're designing today can be ready to keep caring for patients when that day comes.

Comments


HBSLogo_White_RGB.png

800 Township Line Road
Suite 150
Yardley, PA 19067

888-269-4418

  • LinkedIn

© MMXXIV Healthcare Building Solutions LLC

bottom of page