AI’s Energy Crisis: Can Data Centres Keep Up With a World Demanding More Power?

The explosive growth of AI is driving an insatiable demand for data centre capacity, creating a collision course with the world’s finite energy supply. Are we prepared for the coming power crunch?

Share

The rise of artificial intelligence (AI) and generative AI has redefined the boundaries of data growth and ownership, creating a seismic shift in the tech industry. The only trend is clear: more.

Greater data centre capacity requires increasing power, yet energy availability is shrinking. As more homes are built, more businesses come online, and more people use AI, the power and data storage needs of both will grow.

This creates a fundamental conflict for both the tech industry, which is trying to meet customer expectations, and governments, who need to secure the energy supply for homes and businesses. How can both get what they need to satisfy these requirements?

The Scale of Data Centre Demand

Around 2% of global energy generation is consumed by data centres and transmission networks, according to the International Energy Agency (IEA). In terms of specific workloads, McKinsey & Co forecasts a 39% compound annual growth rate (CAGR) increase in Generative AI workload demands and a 16% CAGR on other workloads for global data centre capacity demand by 2030. To put this in perspective, a recently announced hyperscaler data centre is almost the size of Manhattan. This puts immense pressure on national grids.

Within the data centre, power is used for compute, storage, networking, and cooling. Historically, water cooling was widely used, but many operators have stopped this practice and are now using more traditional chillers and cooling strategies driven by electricity. This means less power is going to the compute load and it increases the overall power demand even further.

In terms of AI, a single, high-end GPU can consume as much daily energy as a standard four-person home (roughly 30 kWh). GPU manufacturers are shipping hundreds of thousands of these units every quarter. This demand for AI is taking electricity away from other sources. Put simply, the industry is behind the curve of demand.

Why Simply Generating More Power Isn’t the Answer

One might think the answer is to generate more power. However, it’s not a simple calculation: adding more electricity doesn’t necessarily mean more data centre capacity while keeping homes and businesses powered. For countries that are investing heavily in building data centres, the amount of electricity they require isn’t sustainable without fundamental change. To increase generation, key considerations include:

  • Grid Capacity: Can electricity providers supply more power from their existing grid connections?
  • Resilience: Can more backup coverage be installed to provide resilience, and what fuel or generator types would this require?
  • Cooling: Can more cooling be added to extract the additional heat that will be generated, and is there enough power for this as well?

A Call to Action for the IT Industry

For too long, the IT industry has had a pass on its energy consumption. That must change. Tech leaders need to embrace and champion different ways of doing things—and get in front of governments and policy makers to redesign, revise, and remodel for more energy-efficient technology. Some considerations include:

  1. Champion Sustainable Practices
  • Report power consumption and embedded carbon (CO2e) at a device and fleet level.
  • Focus on and use the most energy-efficient technologies available.
  • Build a circular economy to re-use and refurbish equipment.
  1. Invest in True Innovation
  • Try new things and embrace alternatives.
  • Look at the state-of-the-art components for compute, network, storage, and cooling.
  • Consider how software, like enhanced compression and deduplication algorithms, can impact overall usage.
  • Examine new technologies like ceramic data storage.
  1. Explore Bridge Technologies
  • Small modular reactors (SMRs) could help alleviate energy demands. They are, in the context of the industry, relatively simple to stand up and run and can be online in a few years, instead of a few decades for traditional nuclear plants.
  1. Practice Moderation
  • Ensure the outcome is worth the energy cost—don’t say yes to every AI project.

Innovation in Action: The Power of Efficiency Gains

Pure Storage launched its first product—a five-terabyte system—more than 10 years ago. Since then, the single system capability has increased by a factor of more than 1,200 to six petabytes. The product is now physically smaller and requires less power despite being orders of magnitude larger in capacity. If cars had improved at the same rate since 2013, today we would be able to circumnavigate the Earth in approximately 10 minutes on a single tank of fuel.

The Way Forward

There needs to be massive change in electricity generation, data centres, networking, and the grid—they’re all connected and all need to be updated. The data boom may have been the catalyst that highlights the need to upgrade, but it will not stop or slow down. Consumer and business demand has reached such a pitch that it’s obvious this industry is only growing. The only way to meet these needs is to invest in innovation, embrace change, and be sensible with the resources available.

The question is no longer if we will face an energy reckoning, but how we will rise to meet it.

Alex McMullan
Alex McMullan
CTO International, Pure Storage

Related

Unpack More