We are entering an era in which cloud computing is no longer a separate IT category — it is IT. The use of the cloud in all its forms is growing exponentially, and investment in cloud infrastructure by big hyperscale cloud providers — data centers, servers, storage and data transit — to meet demand is running at over $70 billion per year. As a result, there is a lot of interest in, on the one hand, just how much energy hyperscalers consume, and, on the other hand, the carbon reduction potential of the cloud. So, what is going on, and what is the data based on?
Energy conservation (saving) relies on better efficiencies, so the most effective strategies will focus on squeezing out as much of these efficiencies as possible, getting the energy requirement as low as possible and then looking at how to decarbonize that. Simply moving to using green energy is not going to deliver the desired — or the best — carbon reduction. The best place to go from a carbon reduction perspective is the cloud. This is the case despite a significant amount of overprovisioning in some regions — and underprovisioning in others — because the hyperscalers do not have sufficient visibility around demand to be able to build less than 30% extra capacity into their data center servers and storage to avoid capacity constraint. Not all workloads can go to the cloud, so for those organizations the next best thing may be a multi-tenant facility whose power utilization effectiveness, or PUE, score will be lower than could be achieved on premises. Multi-tenant providers will have better access to green energy. Only after that does an enterprise data center make sense — perhaps for proximity or multicity requirements.
Isn't the carbon reduction potential of the cloud an oxymoron?
Using the results from its 2020-21 surveys, 451 Research's data center team built a data model that looks at how much energy is used by typical IT equipment and how much energy is used by typical data centers. With this, it compared enterprise setups with cloud setups to see if they are similar or different in their energy use. The model determined that cloud servers are much more efficient in general than enterprise servers, partly because they are newer and there is more new equipment in the server farm of a typical cloud supplier. But that is only part of the story: Cloud servers are also much more highly utilized, which makes them more efficient. The team found the savings amounted to as much as a 60% efficiency improvement. The implication is that by just switching servers — moving to the cloud — enterprises can save up to 60% of their energy usage.
Next, our team looked at the data center building itself and found that typical cloud facilities are also much more efficient. Although they are seen as consuming a lot of electricity, they are typically much more efficient than enterprise data centers that are not fully utilized, especially those with older equipment. Ignoring the change in hardware mentioned above, simply moving to the cloud provider's data center delivers an additional 10% to 15% boost in energy efficiency.
Overall, 451 Research data finds that if they move their IT to the cloud, enterprises can save up to 85% of energy usage, resulting in a smaller carbon footprint. In some cases, even more of a reduction can be seen if the cloud facilities deploy completely renewable carbon-free energy. Here, organizations can improve just by moving to the cloud.
By contrast, when an organization houses its own IT stack, it must account for all of its energy use and, therefore, waste. By moving to the cloud, it is further abstracting its applications from the devices themselves — from an accounting perspective, anyway — and therefore the organization is only responsible for the emissions associated with the energy it consumes. It is like pay per use but in terms of emissions. This really only works in the cloud, as in a leased data center it is a bit harder for an organization to absolve itself of the responsibility of added waste. This can be debated, but it is way less cut and dry with the cloud.
Thus, there are essentially three levers that determine the level of saving: Equipment is more efficient, mostly because of enhanced efficiency of processors, over and above the now diminishing impact of Moore's Law; equipment is better utilized, which is one of the biggest determinants of overall operating cost, according to 451 Research's Cloud Price Index; and access to a green energy grid brings additional benefits. This presents a strong underlying message in terms of cloud migration.
The IT savings found are fairly consistent geographically — 60% up to 68%, depending on the country. The U.S., for example, is at the higher end. The typical U.S. enterprise virtualizes servers much more and tends to have greater utilization of servers compared with Asia-Pacific and EU enterprises.
On the data center side, the U.S. has the most efficient data centers among the enterprises surveyed, followed by Europe and the Asia-Pacific; that is partly due to climatic differences, as a hotter climate means greater energy use for cooling. The key differences, however, come back to the access to essentially carbon-free power, and here Europe has far greater access to green power than the U.S. or the Asia-Pacific.
What about PUE?
Energy transformations taking place in most organizations are focused on PUE. The industry and enterprises have focused so much on that data center efficiency metric — partly because it is relatively easy to at least envision, and there are actions that can be taken that make immediate and obvious enhancements. The servers, however, are what is important, though it is much harder for enterprises to virtualize their servers and ensure each server is being deployed at a high level of efficiency. This is where the cloud suppliers are the answer.
It is also an organizational issue. Facility managers whose responsibility is the building itself will fixate on PUE. To them, it is the network administrators who are dealing with the servers. And so, from their perspective, the whole utilization story is not their job. The opposite is also the case, where IT sees access to green energy as the facility's problem. Looking on the data center floor and then outside the data center walls, that has always been somebody else's responsibility. The same has been the case for multi-tenant data center providers, but they are realizing that incentivizing customers to actually have utilization rates helps drive down the PUE of the facility and increases overall efficiency.
Additional factors now include consumer demand for sustainable and carbon-free IT services as part of their supply chains. There are now software packages, such as Ledger8760, that can show customers the greenness of the energy they consume on an hour-by-hour basis — something that was not available 12 months ago. In addition, cloud providers are starting to expose their end users to the relative greenness of each workload.
Enterprises will increasingly face investor and regulatory requirements for sustainability. The largest cloud suppliers and even some of the biggest enterprises are also coming under pressure from Greenpeace and other organizations.
This article was published by S&P Global Market Intelligence and not by S&P Global Ratings, which is a separately managed division of S&P Global.
451 Research is part of S&P Global Market Intelligence.