X
Business

Data centers want to be a lot greener. One big problem is holding them back

Technology companies have set ambitious goals for themselves, but more needs to be done if they are to make good on their promises.
Written by Daphne Leprince-Ringuet, Contributor

Since 2017, Google has been purchasing enough renewable energy to match how much the company consumes in electricity every year, effectively zeroing out the carbon footprint of its entire power use. Amazon has set the same objective for 2030; Facebook has achieved 100% renewable energy in the social media giant's global operations as of this year; Microsoft is looking to hit the milestone in the firm's data centers by 2025. 

More than a trend, investing in green data centers is becoming an imperative for big tech players – an environmental duty that most technology companies have now taken upon themselves to champion.  

With good reason: the huge industrial warehouses that host the countless digital transactions that happen globally every day are estimated to consume a hefty 3% of all electricity around the world, most of which is used for servers and cooling systems. Data is expected to only increase, reaching 175 zettabytes by 2025. One zettabyte is a trillion gigabytes: and there is no sign that data center activity will subside anytime soon. 

SEE: IT Data Center Green Energy Policy (TechRepublic Premium)

In that light, the aggressive climate pledges coming from big tech seem justified – but between pledging and succeeding, there remain many gaps and holes that need to be overcome. "If the technology companies sit and wait for some magical technologies to come along, they might get it eventually. But will they get it on the timescale they have put forward to achieve 100% renewability?" Stuart Adler, researcher in energy storage at the Clean Energy Institute (CEI) at the University of Washington, tells ZDNet. "I'm not sure." 

The commitments that have been made for a carbon-free future sound impressive, but it's always necessary to check the details, too. For example, companies typically commit to "100% renewable energy", meaning that every year, they buy as much renewable energy as they have consumed. This is different, however, to powering all of a data center's operations with solar or wind-based electricity, at all times of the day – a stricter interpretation of "100% renewable", which is much harder to achieve. 

Data centers typically connect to the local electricity grid, meaning that when renewable sources of electricity aren't available, power needs to be produced by alternative sources – often, carbon-based ones. To compensate for those fluctuations, companies can contract renewable energy from sources located on different grids, to make sure that their overall purchase of renewable power still matches their total consumption of electricity. 

Much of the tech giants' efforts have focused, therefore, on buying renewable energy – any renewable energy. In 2017 alone, Google purchased as much power from solar and wind farms than is used yearly by the state of Rhode Island; Facebook, for its part, contracted enough renewable energy in 2019 to power more than 1.6 million US households.  

But powering a data center at every hour of the day with carbon-free sources of energy, which has been openly presented as the end-goal by companies like Microsoft and Google, requires a different strategy: to load the local grid with renewable power that could be stored for later, when the sun isn't shining or the wind isn't blowing. To do so, companies are building their own sources of renewable energy directly next to their data centers; this is the case of Google, but also Amazon, which is currently operating 86 renewable energy projects worldwide. 

As Adler explains, however, this only solves half of the equation. "Right now, the renewable energy production side is much more mature than the renewable energy storage side," he says. "If we don't find a way to store that energy, and redistribute it when we need it and to who needs it, then all of these are only half measures." 

There are many different ways that electricity can be stored. By far the most established solution is the use of lithium-ion batteries, which currently accounts for 90% of the global grid battery storage market. The batteries can store excess energy produced by solar or wind sources, and feed it back into the grid when demand surpasses supply.  

Lithium-ion batteries, however, cannot store energy well over long periods of time, and are rather designed to keep electricity only for a few hours. Looking forward, it is inevitable that a mix of technologies will be needed to enable 24x7 renewable energy supply. 

Alternative technologies that can complement batteries exist already. Scientists are working on sophisticated systems such as fuel cells, which convert electricity from wind and sun into a chemical fuel that can then be stored indefinitely, before reversing the process when necessary to generate energy again. With fuel cells, huge amounts of electricity can be stored and for a very long time, for example in the form of hydrogen.

SEE: Technology's next big challenge: To be fairer to everyone

The technology that underpins fuel cells is advanced enough that the devices can be deployed in real-world environments. Microsoft, for instance, recently unveiled a fuel cell pilot in one of its Azure data centers, to replace the building's diesel-powered generators. The company presented an even bigger vision for the technology, in which data centers fitted with fuel cells could be integrated with the local power grid to generate electricity during periods of high demand. 

Technology alone is not the stumbling block, therefore. "Do we have fuel cells today that can operate on hydrogen and last for the five-year period that data centers need to last? The answer is yes," says Adler. "But can we make them in large numbers at a feasible cost? The answer is no, that's not a solved problem yet." 

"That is one of the big problems. If you want to outfit a giant data center with fuel cells, there is no one today that you can go to and order a million of those systems from. The manufacturing infrastructure for that product simply isn't big enough yet," he continues. 

The market for energy storage technology remains much too small, explains Adler. Only a handful of companies are currently manufacturing fuel cells, and most of them for small consumer bases. In Japan, for example, electronics company Kyocera builds solid oxide fuel cells (SOFC), but only for use in residential settings. The infrastructure in place for those customers, therefore, is radically different to the one that would be needed for the needs of data center operators across the world. 

For the energy storage systems to mature, demand needs to increase, and fast – but demand, as is often the case with emerging technology, is hesitant. Baosen Zhang, researcher in energy systems at the Clean Energy Institute, explains that this needs to change if technology companies are to make good on their promises. "We are not waiting for a technology that is impossible to develop," Zhang tells ZDNet. "If these companies want to get there by 2040 or 2050, they can. It's just a question of how much investment they will deploy for it." 

Being the first company to invest heavily in a new technology like fuel cells is risky; it requires jumping into an ecosystem with few suppliers and little certainty; and costs are very hard to estimate, because that would require speculating about a manufacturing process that essentially doesn't exist yet. 

SEE: This powerful new supercomputer will let scientists ask 'the right questions'

Ss the energy storage technologies that are needed for big tech to achieve its climate goals, are not available at a commercial scale yet to match the relentless expansion of digital services – and it is impossible to know how long it will be before the market is mature enough. 

"It's the chicken and the egg problem," says Adler. "You need customers to buy the product, but then you need the factory to build the product, and it's about knowing which one comes first." 

"If you ask me whether tech companies will meet their goals at some point in the future, then yes, I am very confident," he adds. "Will they reach them by 2030? I don't know. That's an open question." 

To meet the aggressive objectives they have set for themselves, tech giants such as Google and Microsoft will have to take more risks, says Adler – and switch some of the focus from renewable energy production to renewable energy storage. Contracting small suppliers, despite the potential higher costs and uncertainty, is the only way to accelerate the growth of a market that urgently needs to expand, if it is to cope with the explosion of hyperscale data centers. 

There is certainly reason to remain optimistic. Data centers have the potential to make a significant difference to global energy trends, and they are operated by companies that combine deep pockets with an overt commitment to fighting climate change. 

"If the right companies put enough effort into it, it can be done," says Zhang. "Plus, we're going to keep building data centers – so they are probably one of the best places to showcase new storage technology, to show that we can build large-scale green buildings."  

From power-hungry, industrial warehouses that are mostly misunderstood by the public, therefore, it seems that data centers are edging towards trend-setting buildings that could lead a green revolution in the digital sector. It is up to big tech to seize the opportunity. 

Editorial standards