Data Center Power Requirements

As has always been the case, data center power requirements are driven by the capabilities and constraints of the data center IT equipment (ITE), and the demands of the workloads running on that gear.

The increasingly widespread adoption of power-intensive workloads is driven by exponentially rising demand for more data, real-time data consumption, and real-time analytics—particularly as organizations embrace AI for an increasingly wide range of applications. For example, researchers found that a ChatGPT-4 query consumes 3-33 times more energy (3.6-36 kJ) than a Google search query (1.08 kJ).

To support those types of workloads, chipmakers are producing increasingly powerful chips. Modern graphics processing units (GPUs) are around four times more power-dense than CPUs. But in a shift, those increasingly powerful chips are also increasingly power-hungry. As infrastructure leaders look to pack more power into the same space, average data center densities are rising dramatically. And to support it all, critical loads are higher than ever.

Chip performance is increasing—and so is power consumption.
For over a decade, chip performance increased while power consumption remained relatively flat. But since about 2018, rising performance has meant rising power consumption. (This is the Power Wars Era.) Now, more computation requires more watts. As just one example, the maximum power consumption of NVIDIA’s latest GPU is 160% higher than that of the company’s previous generation chips.

AI workloads run on high-performance graphics processing units (GPUs) that are incredibly power intensive. By 2027 all the AI servers in the world, running at full capacity, would consume at least 85.4 terawatt-hours of electricity annually, according to new research. Indeed, “The GPUs that are used to train generative AI models have large power draws,” said Sean Graham, IDC’s research director for cloud to edge data center trends. That’s raising data center power requirements and density and forcing changes in how the data center is cooled.

As infrastructure leaders look to pack more power into the same space, average data center densities are rising dramatically—the AI data center is a high density data center.

Supporting more power-hungry workloads, utility power requirements are increasingly massive. By the end of 2021 there were more 20+ megawatt data center requirements in the market than 1 MW requirements, according to JLL. Most data centers being built today are well over 30 MWs. Data centers are still increasingly efficient, but exponential growth in the workloads they support means significant growth in the size of data centers and the amount of electricity they consume in aggregate. According to the International Energy Agency, combined electricity use by Amazon, Microsoft, Google, and Meta more than doubled between 2017 and 2021, rising to around 72 TWh in 2021.

Explore more in our whitepaper Data Center Development in an AI-Driven Market.