Rethinking sustainability

Lithium is a difficult mineral to extract. Extracting it is primarily done through either evaporation from salty underwater pools, or by traditionally blast-mining spodumene, a lithium-rich mineral. According to some sources, extracting a tonne of lithium can consume up to half a million litres of water and generates fifteen tonnes of CO2.

However, lithium is at the heart of many sustainability drives; an EV battery uses around eight kilograms of it. But once it’s there, it greatly supports our drive to become greener – for example, EVs put out about 60% of the emissions that traditional petrol vehicles do. 

The point of the exercise is to show that we need to look at the bigger picture and avoid simplistic thinking. Progress always has a cost, and we shouldn’t flinch from looking at the broadest possible context to make sure that our actions don’t have unintended consequences. 

Seizing the Opportunity to Rethink our Industry

The cloud industry has developed some bad habits, and it’s clear that in some cases, we’ve built on bad foundations. There’s a reason why a certain brand of electric car is popular: the designers went back to the drawing board and thought about what a truly modern car should do and how it should do it, retaining the good without being afraid to throw out the bad. 

In our industry, for example, there is an unhealthy fascination with using air in datacentres. Compared to water, air is difficult to direct, and the volumetric heat capacity of water is thousands of times better than that of air. Although air is useful in removing heat from some areas, water is far more efficient when cooling very specific and well-defined areas, such as CPUs and GPUs. According to studies, an average air-cooled datacentre has a PuE of around 1.55, compared to 1.2 for Direct Liquid to Chip (DLC) cooling. 

And as we know, liquid cooling can be taken further by immersing the entire server in liquid or even combining this with DLC cooling to ensure that the hottest parts of the server get the attention they need. 

Clearly, this does make maintenance harder, but liquid cooling at scale has been deployed by many organisations – including ours – for a number of years. 

Thinking Bigger

However, reinventing how we think about technology has to go far beyond manufacturing and operations – it’s also crucial to think about what happens at end-of-life, which brings us back to minerals. 

When servers reach end-of-life, it’s rarely the case that all the components need replacing at the same time. The unit might be nearing failure on certain benchmarks, but that doesn’t mean parts of it can’t be re-used. Some parts – hard drives, for example – may need to be securely disposed of in specific cases to comply with data protection regulations, but other parts can be removed and re-used in lower-spec ranges. 

And when whole components can’t be re-used, it’s possible to re-use certain metals. For example, using electrolysis, you can separate copper and other metals from even the most complex and tricky part of the server, the motherboard. This copper can be re-sold and re-used, turning waste into revenue. 

However, there’s also a geo-political component to this. Much of our copper comes from Poland and Germany, but around 70% of the EU’s cobalt, used in batteries, and almost a third (29%) of the EU’s nickel (nickel-plated copper is common in heatsinks) used to come from Russia (European Central Bank, 2023). Magnesium, to give another example, is frequently used in laptops and parts with complex casings because it’s both lightweight and strong – but in 2023, 97% of the EU’s magnesium was processed in China. 

Furthermore, there’s also a need to take stock and re-examine the materials we use. For example, could we use aluminium rather than copper for cooling components? With some work and in some cases, yes. Aluminium is about four times cheaper than copper, because it’s over a thousand times more common. It does take more energy to produce, but in the long run, it emits less carbon and tends to be less harmful to marine environments when disposed of poorly, for example. 

On the other hand, it’s worth noting that we do have to be a bit discerning. Many of the new towns in the UK were created with aluminium rather than copper telephone wires because of post-war shortages. Aluminium only conducts sixty percent as well as copper, and as a result, areas like Milton Keynes experienced atrocious broadband speeds for some years, until they were re-wired with copper and fibre optic cables later on. 

Getting Used to Complexity 

Thinking broadly can be taxing, but it’s worthwhile. After all, as we’ve shown, using more efficient technologies can directly cut power bills, and recycling rare metals can be a source of revenue.

There’s no doubt that our industry is complex, and sometimes it’s easy to only see the challenges. However, it is becoming increasingly clear how important it is to understand every link in the chain so that we can tackle the problems honestly, seize the opportunities when they arise, and make sure that we take even the smallest steps towards being a better, cleaner industry. Because as the saying goes, a journey of a thousand miles still begins with the first step.