Some thoughts on data centres and the environment

325 Front Street West III

People are worried about data centres.

People are worried about data centres and carbon emissions, especially if they read articles like these: Google’s carbon footprint balloons in its Gemini AI era, or Microsoft’s AI obsession is jeopardizing its climate ambitions or Google’s greenhouse gas emissions are soaring thanks to AI or Exxon Plans to Sell Electricity to Data Centers.

People are also worried about data centres and water usage after they read articles about how much water is used for each ChatGPT query.

And when people read pieces like this, Amid Arizona’s data center boom, many Native Americans live without power, they are no doubt worried about what data centres do to the communities where they reside.

My thoughts on this, as someone who has worked in data centres for 40 years, is that there are valid reasons to be concerned, but there are positive aspects to data centre growth and it’s important to keep those in mind.

Data centres are simply places with a concentration of information technology (IT). One time companies had data centres on a floor of their building, or in special buildings in locations like the one pictured about on 325 Front St in Toronto. These days, many companies are moving from hosting their own technology in their own buildings and moving that tech to cloud computing locations, which are just another form of data centre.

I believe that’s this migration to the cloud is a good thing. As this states:

Research published in 2020 found that the computing output of data centers increased 550% between 2010 and 2018. However, energy consumption from those data centers grew just 6%. As of 2018, data centers consumed about 1% of the world’s electricity output.

Moving workloads from on premise infrastructure to cloud infrastructure hosted in big cloud data centres saves on energy consumption.  One of the reason for the savings is this:

…normally IT infrastructure is used on average at 40%. When we move to cloud providers, the rate of efficiency using servers is 85%. So with the same energy, we are managing double or more than double the workloads.

You might think all this data centre growth is being driven by things like AI and crypto, but according to this IEA report:

Demand for digital services is growing rapidly. Since 2010, the number of internet users worldwide has more than doubled, while global internet traffic has expanded 25-fold. Rapid improvements in energy efficiency have, however, helped moderate growth in energy demand from data centres and data transmission networks, which each account for 1-1.5% of global electricity use.

That’s the key point: the demand for digital services is driving the growth of data centres. Every time you watch a video on your phone or pay your bills on your computer, you are using a data centre. Even things like the smart meter on your house or the computer in your car or the digital signs you see interact with a data centre. You use data centres pretty much all day, sometimes without knowing it.

The good news is that there are innovations to make them greener are happening with them, like this new method for liquid-cooling data centres that could make the waste heat useful. And it’s good that IT professionals are moving towards green cloud computing.  But it’s not good that with the rise of technologies like generative AI, IT companies are having a difficult time keeping up with the demand and sticking to their green targets.

Speaking of gen AI, I think energy costs associated with AI will peak and come down from the initial estimates. Indeed, when I read this article Data centres & networks – IEA, and in particular this…

Early studies focused on the energy and carbon emissions associated with training large ML models, but recent data from Meta and Google indicate that the training phase only accounts for around 20–40% of overall ML-related energy use, with 60–70% for inference (the application or use of AI models) and up to 10% for model development (experimentation). Google estimates that ML accounted for 10-15% of its total energy use in 2019-2021, growing at a rate comparable with overall energy growth (+20-25% per year over the same period).

… then I am optimistic that energy costs will not be as bad as initially estimated when it comes to AI. I am not optimistic that we will not decrease our demand for digital services any time soon. Because of that demand, we need not just more data centres, but better ones. It’s up to companies to build them, and it’s up to citizens to keep the companies accountable. The best way to keep them accountable is to better understand how data centres work. I hope this post went some small way to doing that.

P.S. All the opinions expressed here are my own and do not represent the views of my employer.

(Photo by Jack Landau on Flickr)

The fascinating story of Paris Green, a tint that literally was drop dead gorgeous

I love stories about colours and their origins, but this one on Paris Green is especially good: This Trendy But Toxic Shade of Green Left Thousands Dead in the Victorian Era.

Turns out 19th century patrons loved this tint that was produced using…arsenic. You can imagine how this turns out, but save your imagination and read the story.

Image and story from Town and Country, of all things. Not sure how I came across it, but I am glad I did.