Cloud computing is often seen as a net addition to IT’s growing draw on the world’s energy supplies. But what if it could bring energy efficiency gains to offset its role as a booster of IT growth? A new study from Microsoft, Accenture and WSP Environment and Energy says that moving business applications to the cloud can cut the associated per-user carbon footprint 30 percent for larger, already-efficient clients and as much as 90 percent for the smallest and least efficient businesses.

If the real-world cloud deployments of providers like Microsoft, Amazon, Google and a host of others can deliver similar efficiency gains, that could spell new opportunities for making efficiency pay off for customers.

The Microsoft study’s results were based on models of cloud-based energy use, rather than real-world figures. The study found that more energy-efficient servers and state-of-art data centers versus smaller on-premise business environments played a small role in energy savings. But by far the largest share of energy savings came from more efficient operation linked specifically to the nature of cloud computing, such as dynamic provisioning, server multi-tenancy and improved server utilization.

The less a customer is able to accomplish those tasks on its own, the greater the efficiency gains to come by switching to a cloud infrastructure, the study found. For example, one 50,000-person consumer goods company found it could drop per-user carbon emissions by 32 percent by moving its email exchange infrastructure from on-premise to cloud-based infrastructure. But moving server, network and storage infrastructure for a 100-person organization could yield carbon reductions of up to 90 percent, given the anticipated inefficiencies of equipment and management techniques for such a small organization.

All of this information begs the question of just what role energy consumption plays in cloud providers’ business models. Indeed, trying to measure cloud power use gets complicated quickly. Customers will have already reduced their IT energy burden to zero, more or less, by shifting their IT assets and power bills to the cloud provider — unless it’s a private cloud, of course. But measuring energy use only in data centers might miss energy costs incurred by others in the chain, such as the communications providers that link clouds and their users.

Still, shifting all of that IT load to huge centralized data centers will increase the focus on just how those data centers are trying to green their operations. That could be by squeezing more efficiency out of their IT and facility assets, or securing more renewable power to supply them, whether it’s to meet regulatory benchmarks or fend off criticism from shareholder and watchdog groups. The Microsoft/Accenture/WSP study put its results in terms of carbon reduction, rather than energy reductions, after all. Microsoft already has some cloud customers, including the Scandic hotel chain, looking to carbon reductions as a key driver. Of course, going beyond public relations bragging rights for reducing one’s carbon footprint and actually monetizing those reductions is still a long way away.

Finally, it’s important to note that not everyone agrees with Microsoft’s findings. Researchers at University of Melbourne in Australia plan to release a report later this month showing that cloud computing’s energy draw can exceed in-office computing at times, more or less based on how much energy is used to transport data from one location to another. Data storage and software as a service (SaaS), for example, required lots of data moving, and thus were less efficient. Still, tasks such as consolidating workloads from multiple clients onto fewer servers — and using the latest, most efficient servers in the most up-to-date data center environments — are still net efficiency gains, the Australian report found.

Against these benefits, however, cloud providers looking to maximize efficiency may be forced to take limits on distance and data into account — just the things the cloud was designed to maximize for business value. The new math on balancing that efficiency-profitability equation may get tricky.

Question of the week