Saturday, January 18, 2025

The Data Bill: Examining the Energy Demands of Data Centres

The Data Bill is making its way through the House of Lords, and it’s crucial to focus on where all that data is stored and the costs associated with it. Datacentres, once a specialized topic only tech enthusiasts discussed, are now popping up in communities everywhere. We need to think about how these facilities use power and where that energy comes from.

To address this, I proposed an amendment to the Data Bill. It states, “Consultation: datacentre power usage. On the day this Act is passed, the secretary of state must start a consultation on the impact of this Act on the power usage and energy efficiency of datacentres.” During our debate, I pointed out, “It’s odd to have a Data Bill without addressing datacentres’ energy consumption and environmental effects.”

This concern is more urgent than ever. A recent MIT Technology Review article warns that “AI emissions are set to skyrocket.” It reports a threefold increase in datacentre emissions since 2018. Their insights are based on research from Harvard and UCLA, which examined over 2,100 datacentres in the U.S., accounting for 78% of the facilities there.

We often forget that the energy consumed isn’t just from training massive AI models—especially in the U.S., where fossil fuels, primarily coal, are the main power sources for these activities. Every time we interact with any AI model, we draw on that energy.

Is our data growth sustainable? AI is expanding beyond just text into video, music, and more, amping up our power demands. And it’s not only AI. Datacentres are essential for everything we do online, from storing photos to hosting websites, all needing significant energy.

In the U.S., much of this energy is still fossil-fuel-based due to the physical locations of datacentres and the reliable output of coal. But is that approach sustainable in the long run?

The UK has a significant opportunity to lead in datacentre technologies. If the government seizes this chance, it can make strides in energy efficiency, sustainability, and environmental responsibility.

This isn’t just about technology or environmental issues; it’s existential. We have to ask ourselves how we power our lives. Even if we generate renewable energy, there are practical concerns about storage, deployment, and connecting to the grid given the current constraints.

Power usage effectiveness (PUE) is the primary metric for assessing datacentre energy efficiency, and I’m keen to hear thoughts on its effectiveness. During the debate, I asked the government about their view on the current PUE standard. Does it instill confidence in consumers?

The government promised to share a detailed response. I hope to gain clearer insight into their plans on this critical environmental, economic, and technological issue.

Data isn’t simply the “new oil”; it relies on numerous sources of old oil to function. While we might eventually turn to nuclear and sustainable energy for datacentres, we need to consider the opportunity cost of those resources.

Data and the technologies it enables hold immense potential for our economy and society. Ultimately, it’s in our hands—the conversations we foster, the choices we make, and the communities we shape will guide our future. We have the power to influence this direction.