Meanwhile, there is a breakdown happening at a larger scale but so slowly that we hardly notice it: The environment around us. The data centre industry has an enormous effect on this environmental breakdown.
While estimates vary, the data centre industry currently consumes three to five percent of the world's electricity and is responsible for about two percent of global carbon dioxide emissions. The global infrastructures that work invisibly behind our touch screens thus contribute to the oncoming climate catastrophe in the same degree as the aviation industry globally. A data centre is like a gigantic airliner that flies through the clouds with all of our data onboard and never needs to land to refuel.
Since 2006, power-usage effectiveness (PUE) has been used across the data centre industry to measure, manage and compete over the energy efficiency of data centres. It measures the amount of electricity used by the actual servers and other IT machinery of a data centre in relation to how much power the facility consumes as a whole. The number, which ideally is 1.0 – meaning that all power is used for computing – only says something about what the energy is used for inside the centre and nothing about where it comes from and how it is generated or even how much energy is actually used.
In 2018, according to their own figures, Google used 10.1 terawatt hours of electricity. If the company was ranked as a country, it would come in at 91st place based on total energy consumption just above Uruguay.
Just as the data centre industry as a whole contributes to global warming, each data centre produces substantial heat. A major factor in the energy cost of data infrastructure is cooling them down and diverting excess heat elsewhere. This has lead big companies to build their new facilities in colder climates, like Facebook's data centre in Luleå, Sweden and the one in Odense, Denmark. Similarly, Iceland, with its combination of chilly weather and cheap renewable energy has poised itself as a haven for data centres.
However, a large part of the recent data centre explosion in Iceland is due to bitcoin and blockchain based companies. Bitcoin and the blockchain – while itself a decentralised data infrastructure – has a massive power consumption and corresponding environmental impact. And unlike the Icelandic data centres catering to cryptocurrency companies, most Bitcoin farms elsewhere on the planet are less likely to invest in green energy solutions than PR-sensitive corporations like Google and Facebook.
Since 2010, Greenpeace has campaigned internationally to pressure technology companies to switch their operations to renewable energy and with a great deal of impact. Giants like Apple, Facebook and Google are investing in renewable energy to power their data centres, in some cases even planning wind and solar farms of their own.
Besides the obvious PR-arguments from the corporate side there are also business arguments for the switch. As infrastructure journalist Ingrid Burrington wrote, "Tech companies aren’t underwriting wind turbines for their data centers or embracing renewable energy solely out of some altruistic concern for the environment or fear of bad press. If that were the case, they'd be using their billions in profit to build wind turbines to power the grids of entire cities, not just cancelling out their own environmental impact."
Burrington notes that, while it might be cheaper in the short term, power from fossil fuels such as coal are relying on volatile commodities whose market prices can vary wildly leading to uncertain energy costs. Renewable energy, on the other hand, is less affected by global market speculation and even if it is slightly more expensive, it is easier to plan for in the long-term.
But giants investing in renewable energy solve only one part of the environmental impact of the data industry. First of all, they represent only a small part of the whole big data ecosystem - albeit the most public facing one. Greenpeace IT analyst Gary Cook wrote in 2017, that only 20% of the energy used to power data centres in the world comes from renewable energy with fossil fuels accounting for the remaining 80%.
Secondly there are much wider environmental costs that are not solved by giants going green. Issues like land use factors in. The data centres take up physical space and require other public infrastructure than just electricity and water for cooling. And then there are the supply chains for the materials like rare earth minerals used to manufacture the IT equipment. These get mined leaving massive lasting imprints on physical landscapes and human communities, often in the Global South similar to where a lot of e-waste ends up at the end of the technology life cycle.
Finally, there is the raw cost issue at the core of the problem from the business side. As the amount of data being collected, stored and processed about us continues to rise, the energy requirements and thus the carbon footprint of data centres also grows. Switching to renewable energy sources does not reduce the cost of collecting, storing and processing more and more data. And while the industry is doing what it can to make the data centres more energy efficient, they still have to keep up with what some predict will be exponential growth in the amount of data in the world. As the amount of data grows, the cost of running data centres therefore go irreversibly upwards.
polypoly's aims to change the trajectory towards environmental breakdown, by revolutionising the way businesses understand and handle personal data. Instead of scrambling to collect more of it like digging up more fossil fuels to power the engines of the digital market, the guiding idea of the company is that you can do more with less. In fact, the less data a business has to collect, store and process, the lower the costs. Companies that reduce their data dependency, not only reduce their power usage, and thereby environmental impact, but also reduce operational expenses.
“We have researched the cost structure of the big tech companies. In our analysis of their main weaknesses, all of them have one thing in common and it’s something they can’t change any more. And that is the data centres, the centralised storage of data. This will get more and more expensive, mainly because of energy costs. Of course there’s optimisation. They can buy green energy and build their data centres where they can buy the energy cheaper than in San Francisco. But at the end of the day, energy costs grow more and more expensive,” says Thorsten Dittmar, founder of polypoly.
The energy cost of running and cooling the equipment is the main part of the rising information technology operational expenditures – or ‘IT OpEx’ in the management vocabulary - for companies that collect, store and process data in a centralised manner. But the costs of establishing the infrastructure (so-called capital expenditures or 'CapEx') are in themselves staggering.
“Even the investment costs are unbelievable. Google alone recently invested 13 billion dollars in data centres. And the more data they collect and the more they want to use artificial intelligence, the more expensive it will get over time,” says Thorsten Dittmar.
Instead of mining mountains of collected user data in a centralised data centre, polypoly turns the digital services model on its head. Rather than the company collecting user data and storing it centrally, the users keep their own data locally on their devices in the polyPod, polypoly's core product. It is a data safe under control by the user, but also a platform for running algorithmic computations that the company refers to as 'features'.
Instead of processing massive amounts of user data in the same place at the same time, polypoly is developing a special purpose query language, where a company offering digital services can send an algorithmic question (e.g. “would you like to see an ad for a car?”) that the user can choose to answer or not based on the data they have on their own device (e.g. “no thanks, I bike to work every day but I do need a new front tire”). It is also possible for researchers to query populations at large, without having to gather, store and protect the data themselves.
The economic benefit of this model, and the business case for both polypoly and the company's partners, is that this model saves cost for everyone involved. And it does so in a way that overseas data giants cannot compete with, having to maintain infrastructures that keep growing not only in how much data they handle but also how much energy they require to handle it.
“The IT OpEx costs for us are tending towards zero because the system is 100% decentralised. It doesn’t matter how much they are optimising their IT OpEx cost. It can’t be lower than zero,” says Thorsten Dittmar.