Bits and Bytes, and Energy Demand

Bits and Bytes, and Energy Demand

Bits and bytes are not my speciality, however, they represent a trend that could affect demand for electricity.

Data is composed of bits and bytes, and data acquisition, transmission, manipulation and storage are growing at fantastic rates.

The growth rate for electricity demand since 2000 has been about 0.8% annually, which is approximately the population growth rate of the United States. In fact, growth in demand has been essentially zero over the past five years, which is most likely due to the rapid adoption of LEDs.

Statista estimates that LEDs will penetrate 61% of the global lighting market by 2020.

An earlier article explained why LEDs have probably caused this period of no-growth and that growth will likely resume when LEDs have achieved 80% penetration of type A-sockets used for more than an hour each day. See, Growth in Electricity Consumption – Part 1

After removing the effect of LEDs, baseline growth in electricity demand would be 1% annually. At this rate we can expect demand and the need for new power generation capacity to double in around 70 years.

But what happens if data growth continues to grow exponentially?

Mark P. Mills, senior fellow at the Manhattan Institute, published a five part series on the growth of data which provides a stating point for thinking about how the growth in data could affect demand for electricity.

It takes electrical energy to move, manipulate, transport and access stored data, which are bits and bytes.

While the amount of energy used to move a single piece of data is tiny, in the aggregate energy usage can be substantial. 

First, it requires energy to move and manipulate a piece of data on a three dimensional chip, i.e., with multiple processing layers. Then it takes energy to transport that data between processors for parallel processing. Both 3D chips and parallel processing are techniques that have improved efficiency. 

Future improvements in efficiency is another consideration when thinking about how much energy will be needed to handle increasingly large amounts of data.

One indication of how fast data is growing is the terminology used to describe it.

The world creates some 40 zettabytes of useable data each year, and CISCO predicts this rate will double in three years. (Cisco’s Global Cloud Index: Forecast and Methodology, 2016–2021 White Paper provides interesting information on data growth.)

And what is a zettabyte?

According to Mill’s, “A zetta stack of dollar bills would go from the earth to the sun — that’s 93 million miles away — and back, 700 thousand times.”

The world’s System of Units must adopt additional prefixes to represent what happens after yottabytes which is already being stretched to the limit.

In addition, there will be edge data centers because it takes too long to transfer large amounts of data any great distance. These edge data centers will keep the data close to where it will be used.

According to Data Economy, data centers worldwide in 2015 used 416 TWh of electricity, and, according to CISCO 48% of that, or 200 TWh, was used by data centers in North America.

Using CISCO’s estimate of a 13% growth rate, North America would be using 368 TWh in 2020. The annual increase would be over 40 TWh in 2020, which is roughly one percent of US usage, implying a 1% per year growth in America’s energy demand.

With a base growth rate of 1% and an additional growth rate of 1% from increased usage of data, total US energy demand would grow at a rate of 2%.

A 2% growth rate in demand would imply a doubling of power generation capacity by 2055.

While this logic train is imprecise, it demonstrates that growth in data usage could result in the need for twice as much power generation capacity by mid-century. 

And this could be a conservative estimate when all the technologies increasing data usage are considered, such as:

  • Artificial Intelligence
  • Internet of Things
  • Digital Twins
  • Autonomous Vehicles
  • Robots

Exponential growth in data could render renewables irrelevant, since only fossil fuels and nuclear power may be able to provide the energy needed by a data hungry world.

Growth in data could be a game changer.

. . .


Please follow and like us:

4 Replies to “Bits and Bytes, and Energy Demand”

  1. Thanks. Very interesting. In looking at your suggested link to ThorCon’s forecast of electricity demand I noticed the information is not dated. It showed US capacity at 528 GW which is about half of today’s capacity. I may have misunderstood the chart, but it’s important to know the installed capacity today, and what the growth will be for the segments in your analysis.
    The segment I was discussing is information technology and I was trying to compare your 300 GW increase by 2030 with my 2% annual growth rate. I can’t be sure whether the 300 GW in 2030 is related to installed capacity today or some date in the past.
    My understanding is that worldwide installed capacity today is around 6,500 GW, though it’s hard to get a real number since so much of the information is corrupted by the inclusion of renewables whose installed capacity overstates the true capacity.
    Can you provide information that will help me understand your data.

    • My GW numbers were not installed capacity, but average power generation/consumption.

      I’d hoped the plus sign on +300 GW would imply additional power consumption. The numbers I find vary all over the place, so take this with a grain of salt.

      My objective is to emphasize that world power consumption is going to increase dramatically, even though US demand is flat. The added demand by 2040 will be on the order of 600 GW. I’m trying to make the point to potential investors that there is a big demand for power plants. Greenpeace et al count 574 GW of new coal-fired plants.

      • I understand your objective and am sympathetic toward your being successful. I think you can help your objective by using current capacity information and then estimating the probable increases in demand.
        For my article, I focused on the US because I can get good capacity and demand information.
        If you have additional information on the growth in data I would like to see it.