The trade-off between resilience and efficiency in cloud-based data storage models begs consideration.

Cloud-based models make an individual’s data available seamlessly, regardless of the device used. Data—and even the software used to process it (such as Microsoft Office 365)—are no longer tied to a specific location. The IT landscape has been revolutionized as almost all IT-related activities can be outsourced to cloud platforms. Firms and their employees no longer need to maintain local data storage facilities, back up data, or updated software versions. Capital expenditures—once seen as lumpy investments—can now be spread out over time as regular payments for infrastructure as a service (IaaS), platform as a service (PaaS), or software as a service (SaaS).

These services have been marketed for their flexibility, scalability, accessibility, and availability of specialized expertise and innovation. Reportedly, 94 percent of enterprises use cloud services to boost efficiency. And maybe they should be, insofar as firms should make decisions based upon their own direct costs and benefits and so long as the services purchased are available seamlessly at all times and locations. However, when the supply chain breaks—as it does for a variety of reasons—productivity ceases entirely until the rupture is repaired. Furthermore, when productivity is disrupted, so are the productivity and outcomes for all those relying on the firm to provide products and services in real time.

Firms accrued local productivity gains from moving to Saas, PaaS, IaaS, and cloud storage (“cloud providers”). And the firms mostly have contracts with cloud providers that make the cloud provider liable for specified performance levels and restoring service as fast as possible when failures occur. But these contracts rarely compensate for lost productivity of the client firm, let alone that incurred by the client’s clients, and so on.

This is the classic problem of social costs and benefits being accrued beyond the boundaries of the firms concerned. In such circumstances, firms in the data supply chain will invest no more than what is necessary for their resilience; no one is responsible for maintaining society’s overall productivity when outages inevitably occur. The savings from moving to more efficient operations are pocketed, but the new risks are passed down to the users of services provided and society generally. These include academics who could not research and teach and students who could not learn when a botched university upgrade severed access to OneDrive, their files, even their applications, such as Word and Excel. Or, for example, air travelers worldwide when the CrowdStrike outage in 2024 severed airport check-in desks from all passenger flight data required to manage flights and airport activities, and in some cases, from the programs managing such operations.

Fortunately, cloud providers and their clients can make their systems more resilient without losing the functional benefits of the cloud model. Cloud storage has removed some risks by ensuring central data repository. There are also economies of scale in archiving large quantities of data together.

However, cloud storage has removed all of the local flexibility of having data and software available locally when either the internet or the cloud intermediary goes down. Transporting data and software to where it is needed trades the cost of transport for the cost of local storage and the risks of not being able to transport it when needed. And while the costs of both have fallen substantially in recent years, the costs of storage have fallen much faster than the costs of transport.

Moreover, the synchronization technologies has also improved substantially. At small additional cost of local storage, it is feasible for data to be stored locally (say, on a laptop) and synchronized with the cloud storage (say, OneDrive) in a manner that is seamless to the laptop user. Similarly, local versions of software applications can be maintained. When the cloud or the internet go down, local operations that do not require interaction with other parties(e.g., scheduling medical appointments or completing check-in and baggage tagging processes)—can continue. Such arrangements are rapidly becoming as important in the resilience debate as providing battery backup at cell phone towers to maintain connectivity during electricity outages and climate -induced catastrophes.

There is a small monetary cost for this additional resilience, and there is a debate to be had as to how it should be funded, but the societal benefit is undisputed. The cloud model is not quite as superior economically as first thought, but responsible and resilient cloud supply chains will help reduce the social costs of outages.